libvpx: Necessary modifications

- convert ARM assembly (NEON) files,
- add rtcd for run-time CPU features detection,
- modify "system_state.h",
- "arm_cpudetect.c" fixes.

thirdparty/libvpx/rtcd/vp8_rtcd_arm.h

@@ -0,0 +1,240 @@
+#ifndef VP8_RTCD_H_
+#define VP8_RTCD_H_
+
+#ifdef RTCD_C
+#define RTCD_EXTERN
+#else
+#define RTCD_EXTERN extern
+#endif
+
+/*
+ * VP8
+ */
+
+struct blockd;
+struct loop_filter_info;
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+void vp8_bilinear_predict16x16_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+void vp8_bilinear_predict16x16_neon(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+RTCD_EXTERN void (*vp8_bilinear_predict16x16)(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+
+void vp8_bilinear_predict4x4_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+#define vp8_bilinear_predict4x4 vp8_bilinear_predict4x4_c
+
+void vp8_bilinear_predict8x4_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+void vp8_bilinear_predict8x4_neon(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+RTCD_EXTERN void (*vp8_bilinear_predict8x4)(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+
+void vp8_bilinear_predict8x8_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+void vp8_bilinear_predict8x8_neon(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+RTCD_EXTERN void (*vp8_bilinear_predict8x8)(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+
+void vp8_clear_system_state_c();
+#define vp8_clear_system_state vp8_clear_system_state_c
+
+void vp8_copy_mem16x16_c(unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch);
+void vp8_copy_mem16x16_neon(unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch);
+RTCD_EXTERN void (*vp8_copy_mem16x16)(unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch);
+
+void vp8_copy_mem8x4_c(unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch);
+void vp8_copy_mem8x4_neon(unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch);
+RTCD_EXTERN void (*vp8_copy_mem8x4)(unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch);
+
+void vp8_copy_mem8x8_c(unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch);
+void vp8_copy_mem8x8_neon(unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch);
+RTCD_EXTERN void (*vp8_copy_mem8x8)(unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch);
+
+void vp8_dc_only_idct_add_c(short input, unsigned char *pred, int pred_stride, unsigned char *dst, int dst_stride);
+void vp8_dc_only_idct_add_neon(short input, unsigned char *pred, int pred_stride, unsigned char *dst, int dst_stride);
+RTCD_EXTERN void (*vp8_dc_only_idct_add)(short input, unsigned char *pred, int pred_stride, unsigned char *dst, int dst_stride);
+
+void vp8_dequant_idct_add_c(short *input, short *dq, unsigned char *output, int stride);
+void vp8_dequant_idct_add_neon(short *input, short *dq, unsigned char *output, int stride);
+RTCD_EXTERN void (*vp8_dequant_idct_add)(short *input, short *dq, unsigned char *output, int stride);
+
+void vp8_dequant_idct_add_uv_block_c(short *q, short *dq, unsigned char *dst_u, unsigned char *dst_v, int stride, char *eobs);
+void vp8_dequant_idct_add_uv_block_neon(short *q, short *dq, unsigned char *dst_u, unsigned char *dst_v, int stride, char *eobs);
+RTCD_EXTERN void (*vp8_dequant_idct_add_uv_block)(short *q, short *dq, unsigned char *dst_u, unsigned char *dst_v, int stride, char *eobs);
+
+void vp8_dequant_idct_add_y_block_c(short *q, short *dq, unsigned char *dst, int stride, char *eobs);
+void vp8_dequant_idct_add_y_block_neon(short *q, short *dq, unsigned char *dst, int stride, char *eobs);
+RTCD_EXTERN void (*vp8_dequant_idct_add_y_block)(short *q, short *dq, unsigned char *dst, int stride, char *eobs);
+
+void vp8_dequantize_b_c(struct blockd*, short *dqc);
+void vp8_dequantize_b_neon(struct blockd*, short *dqc);
+RTCD_EXTERN void (*vp8_dequantize_b)(struct blockd*, short *dqc);
+
+void vp8_loop_filter_bh_c(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+void vp8_loop_filter_bh_neon(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+RTCD_EXTERN void (*vp8_loop_filter_bh)(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+
+void vp8_loop_filter_bv_c(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+void vp8_loop_filter_bv_neon(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+RTCD_EXTERN void (*vp8_loop_filter_bv)(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+
+void vp8_loop_filter_mbh_c(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+void vp8_loop_filter_mbh_neon(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+RTCD_EXTERN void (*vp8_loop_filter_mbh)(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+
+void vp8_loop_filter_mbv_c(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+void vp8_loop_filter_mbv_neon(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+RTCD_EXTERN void (*vp8_loop_filter_mbv)(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+
+void vp8_loop_filter_bhs_c(unsigned char *y, int ystride, const unsigned char *blimit);
+void vp8_loop_filter_bhs_neon(unsigned char *y, int ystride, const unsigned char *blimit);
+RTCD_EXTERN void (*vp8_loop_filter_simple_bh)(unsigned char *y, int ystride, const unsigned char *blimit);
+
+void vp8_loop_filter_bvs_c(unsigned char *y, int ystride, const unsigned char *blimit);
+void vp8_loop_filter_bvs_neon(unsigned char *y, int ystride, const unsigned char *blimit);
+RTCD_EXTERN void (*vp8_loop_filter_simple_bv)(unsigned char *y, int ystride, const unsigned char *blimit);
+
+void vp8_loop_filter_simple_horizontal_edge_c(unsigned char *y, int ystride, const unsigned char *blimit);
+void vp8_loop_filter_mbhs_neon(unsigned char *y, int ystride, const unsigned char *blimit);
+RTCD_EXTERN void (*vp8_loop_filter_simple_mbh)(unsigned char *y, int ystride, const unsigned char *blimit);
+
+void vp8_loop_filter_simple_vertical_edge_c(unsigned char *y, int ystride, const unsigned char *blimit);
+void vp8_loop_filter_mbvs_neon(unsigned char *y, int ystride, const unsigned char *blimit);
+RTCD_EXTERN void (*vp8_loop_filter_simple_mbv)(unsigned char *y, int ystride, const unsigned char *blimit);
+
+void vp8_short_idct4x4llm_c(short *input, unsigned char *pred, int pitch, unsigned char *dst, int dst_stride);
+void vp8_short_idct4x4llm_neon(short *input, unsigned char *pred, int pitch, unsigned char *dst, int dst_stride);
+RTCD_EXTERN void (*vp8_short_idct4x4llm)(short *input, unsigned char *pred, int pitch, unsigned char *dst, int dst_stride);
+
+void vp8_short_inv_walsh4x4_c(short *input, short *output);
+void vp8_short_inv_walsh4x4_neon(short *input, short *output);
+RTCD_EXTERN void (*vp8_short_inv_walsh4x4)(short *input, short *output);
+
+void vp8_short_inv_walsh4x4_1_c(short *input, short *output);
+#define vp8_short_inv_walsh4x4_1 vp8_short_inv_walsh4x4_1_c
+
+void vp8_sixtap_predict16x16_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+void vp8_sixtap_predict16x16_neon(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+RTCD_EXTERN void (*vp8_sixtap_predict16x16)(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+
+void vp8_sixtap_predict4x4_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+#define vp8_sixtap_predict4x4 vp8_sixtap_predict4x4_c
+
+void vp8_sixtap_predict8x4_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+void vp8_sixtap_predict8x4_neon(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+RTCD_EXTERN void (*vp8_sixtap_predict8x4)(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+
+void vp8_sixtap_predict8x8_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+void vp8_sixtap_predict8x8_neon(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+RTCD_EXTERN void (*vp8_sixtap_predict8x8)(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+
+void vp8_rtcd(void);
+
+#ifdef RTCD_C
+#include "vpx_ports/arm.h"
+static void setup_rtcd_internal(void)
+{
+    int flags = arm_cpu_caps();
+
+    vp8_bilinear_predict16x16 = vp8_bilinear_predict16x16_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vp8_bilinear_predict16x16 = vp8_bilinear_predict16x16_neon;
+#endif
+    vp8_bilinear_predict8x4 = vp8_bilinear_predict8x4_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vp8_bilinear_predict8x4 = vp8_bilinear_predict8x4_neon;
+#endif
+    vp8_bilinear_predict8x8 = vp8_bilinear_predict8x8_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vp8_bilinear_predict8x8 = vp8_bilinear_predict8x8_neon;
+#endif
+    vp8_copy_mem16x16 = vp8_copy_mem16x16_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vp8_copy_mem16x16 = vp8_copy_mem16x16_neon;
+#endif
+    vp8_copy_mem8x4 = vp8_copy_mem8x4_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vp8_copy_mem8x4 = vp8_copy_mem8x4_neon;
+#endif
+    vp8_copy_mem8x8 = vp8_copy_mem8x8_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vp8_copy_mem8x8 = vp8_copy_mem8x8_neon;
+#endif
+    vp8_dc_only_idct_add = vp8_dc_only_idct_add_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vp8_dc_only_idct_add = vp8_dc_only_idct_add_neon;
+#endif
+    vp8_dequant_idct_add = vp8_dequant_idct_add_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vp8_dequant_idct_add = vp8_dequant_idct_add_neon;
+#endif
+    vp8_dequant_idct_add_uv_block = vp8_dequant_idct_add_uv_block_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vp8_dequant_idct_add_uv_block = vp8_dequant_idct_add_uv_block_neon;
+#endif
+    vp8_dequant_idct_add_y_block = vp8_dequant_idct_add_y_block_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vp8_dequant_idct_add_y_block = vp8_dequant_idct_add_y_block_neon;
+#endif
+    vp8_dequantize_b = vp8_dequantize_b_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vp8_dequantize_b = vp8_dequantize_b_neon;
+#endif
+    vp8_loop_filter_bh = vp8_loop_filter_bh_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vp8_loop_filter_bh = vp8_loop_filter_bh_neon;
+#endif
+    vp8_loop_filter_bv = vp8_loop_filter_bv_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vp8_loop_filter_bv = vp8_loop_filter_bv_neon;
+#endif
+    vp8_loop_filter_mbh = vp8_loop_filter_mbh_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vp8_loop_filter_mbh = vp8_loop_filter_mbh_neon;
+#endif
+    vp8_loop_filter_mbv = vp8_loop_filter_mbv_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vp8_loop_filter_mbv = vp8_loop_filter_mbv_neon;
+#endif
+    vp8_loop_filter_simple_bh = vp8_loop_filter_bhs_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vp8_loop_filter_simple_bh = vp8_loop_filter_bhs_neon;
+#endif
+    vp8_loop_filter_simple_bv = vp8_loop_filter_bvs_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vp8_loop_filter_simple_bv = vp8_loop_filter_bvs_neon;
+#endif
+    vp8_loop_filter_simple_mbh = vp8_loop_filter_simple_horizontal_edge_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vp8_loop_filter_simple_mbh = vp8_loop_filter_mbhs_neon;
+#endif
+    vp8_loop_filter_simple_mbv = vp8_loop_filter_simple_vertical_edge_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vp8_loop_filter_simple_mbv = vp8_loop_filter_mbvs_neon;
+#endif
+    vp8_short_idct4x4llm = vp8_short_idct4x4llm_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vp8_short_idct4x4llm = vp8_short_idct4x4llm_neon;
+#endif
+    vp8_short_inv_walsh4x4 = vp8_short_inv_walsh4x4_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vp8_short_inv_walsh4x4 = vp8_short_inv_walsh4x4_neon;
+#endif
+    vp8_sixtap_predict16x16 = vp8_sixtap_predict16x16_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vp8_sixtap_predict16x16 = vp8_sixtap_predict16x16_neon;
+#endif
+    vp8_sixtap_predict8x4 = vp8_sixtap_predict8x4_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vp8_sixtap_predict8x4 = vp8_sixtap_predict8x4_neon;
+#endif
+    vp8_sixtap_predict8x8 = vp8_sixtap_predict8x8_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vp8_sixtap_predict8x8 = vp8_sixtap_predict8x8_neon;
+#endif
+}
+#endif
+
+#ifdef __cplusplus
+}  // extern "C"
+#endif
+
+#endif

thirdparty/libvpx/rtcd/vp8_rtcd_c.h

@@ -0,0 +1,117 @@
+#ifndef VP8_RTCD_H_
+#define VP8_RTCD_H_
+
+#ifdef RTCD_C
+#define RTCD_EXTERN
+#else
+#define RTCD_EXTERN extern
+#endif
+
+/*
+ * VP8
+ */
+
+struct blockd;
+struct loop_filter_info;
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+void vp8_bilinear_predict16x16_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+#define vp8_bilinear_predict16x16 vp8_bilinear_predict16x16_c
+
+void vp8_bilinear_predict4x4_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+#define vp8_bilinear_predict4x4 vp8_bilinear_predict4x4_c
+
+void vp8_bilinear_predict8x4_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+#define vp8_bilinear_predict8x4 vp8_bilinear_predict8x4_c
+
+void vp8_bilinear_predict8x8_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+#define vp8_bilinear_predict8x8 vp8_bilinear_predict8x8_c
+
+void vp8_clear_system_state_c();
+#define vp8_clear_system_state vp8_clear_system_state_c
+
+void vp8_copy_mem16x16_c(unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch);
+#define vp8_copy_mem16x16 vp8_copy_mem16x16_c
+
+void vp8_copy_mem8x4_c(unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch);
+#define vp8_copy_mem8x4 vp8_copy_mem8x4_c
+
+void vp8_copy_mem8x8_c(unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch);
+#define vp8_copy_mem8x8 vp8_copy_mem8x8_c
+
+void vp8_dc_only_idct_add_c(short input, unsigned char *pred, int pred_stride, unsigned char *dst, int dst_stride);
+#define vp8_dc_only_idct_add vp8_dc_only_idct_add_c
+
+void vp8_dequant_idct_add_c(short *input, short *dq, unsigned char *output, int stride);
+#define vp8_dequant_idct_add vp8_dequant_idct_add_c
+
+void vp8_dequant_idct_add_uv_block_c(short *q, short *dq, unsigned char *dst_u, unsigned char *dst_v, int stride, char *eobs);
+#define vp8_dequant_idct_add_uv_block vp8_dequant_idct_add_uv_block_c
+
+void vp8_dequant_idct_add_y_block_c(short *q, short *dq, unsigned char *dst, int stride, char *eobs);
+#define vp8_dequant_idct_add_y_block vp8_dequant_idct_add_y_block_c
+
+void vp8_dequantize_b_c(struct blockd*, short *dqc);
+#define vp8_dequantize_b vp8_dequantize_b_c
+
+void vp8_loop_filter_bh_c(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+#define vp8_loop_filter_bh vp8_loop_filter_bh_c
+
+void vp8_loop_filter_bv_c(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+#define vp8_loop_filter_bv vp8_loop_filter_bv_c
+
+void vp8_loop_filter_mbh_c(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+#define vp8_loop_filter_mbh vp8_loop_filter_mbh_c
+
+void vp8_loop_filter_mbv_c(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+#define vp8_loop_filter_mbv vp8_loop_filter_mbv_c
+
+void vp8_loop_filter_bhs_c(unsigned char *y, int ystride, const unsigned char *blimit);
+#define vp8_loop_filter_simple_bh vp8_loop_filter_bhs_c
+
+void vp8_loop_filter_bvs_c(unsigned char *y, int ystride, const unsigned char *blimit);
+#define vp8_loop_filter_simple_bv vp8_loop_filter_bvs_c
+
+void vp8_loop_filter_simple_horizontal_edge_c(unsigned char *y, int ystride, const unsigned char *blimit);
+#define vp8_loop_filter_simple_mbh vp8_loop_filter_simple_horizontal_edge_c
+
+void vp8_loop_filter_simple_vertical_edge_c(unsigned char *y, int ystride, const unsigned char *blimit);
+#define vp8_loop_filter_simple_mbv vp8_loop_filter_simple_vertical_edge_c
+
+void vp8_short_idct4x4llm_c(short *input, unsigned char *pred, int pitch, unsigned char *dst, int dst_stride);
+#define vp8_short_idct4x4llm vp8_short_idct4x4llm_c
+
+void vp8_short_inv_walsh4x4_c(short *input, short *output);
+#define vp8_short_inv_walsh4x4 vp8_short_inv_walsh4x4_c
+
+void vp8_short_inv_walsh4x4_1_c(short *input, short *output);
+#define vp8_short_inv_walsh4x4_1 vp8_short_inv_walsh4x4_1_c
+
+void vp8_sixtap_predict16x16_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+#define vp8_sixtap_predict16x16 vp8_sixtap_predict16x16_c
+
+void vp8_sixtap_predict4x4_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+#define vp8_sixtap_predict4x4 vp8_sixtap_predict4x4_c
+
+void vp8_sixtap_predict8x4_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+#define vp8_sixtap_predict8x4 vp8_sixtap_predict8x4_c
+
+void vp8_sixtap_predict8x8_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+#define vp8_sixtap_predict8x8 vp8_sixtap_predict8x8_c
+
+void vp8_rtcd(void);
+
+#ifdef RTCD_C
+static void setup_rtcd_internal(void)
+{
+}
+#endif
+
+#ifdef __cplusplus
+}  // extern "C"
+#endif
+
+#endif

thirdparty/libvpx/rtcd/vp8_rtcd_x86.h

@@ -0,0 +1,247 @@
+#ifndef VP8_RTCD_H_
+#define VP8_RTCD_H_
+
+#ifdef RTCD_C
+#define RTCD_EXTERN
+#else
+#define RTCD_EXTERN extern
+#endif
+
+/*
+ * VP8
+ */
+
+struct blockd;
+struct loop_filter_info;
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+void vp8_bilinear_predict16x16_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+void vp8_bilinear_predict16x16_mmx(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+void vp8_bilinear_predict16x16_sse2(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+void vp8_bilinear_predict16x16_ssse3(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+RTCD_EXTERN void (*vp8_bilinear_predict16x16)(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+
+void vp8_bilinear_predict4x4_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+void vp8_bilinear_predict4x4_mmx(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+RTCD_EXTERN void (*vp8_bilinear_predict4x4)(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+
+void vp8_bilinear_predict8x4_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+void vp8_bilinear_predict8x4_mmx(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+RTCD_EXTERN void (*vp8_bilinear_predict8x4)(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+
+void vp8_bilinear_predict8x8_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+void vp8_bilinear_predict8x8_mmx(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+void vp8_bilinear_predict8x8_sse2(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+void vp8_bilinear_predict8x8_ssse3(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+RTCD_EXTERN void (*vp8_bilinear_predict8x8)(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+
+void vp8_clear_system_state_c();
+void vpx_reset_mmx_state();
+RTCD_EXTERN void (*vp8_clear_system_state)();
+
+void vp8_copy_mem16x16_c(unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch);
+void vp8_copy_mem16x16_mmx(unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch);
+void vp8_copy_mem16x16_sse2(unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch);
+RTCD_EXTERN void (*vp8_copy_mem16x16)(unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch);
+
+void vp8_copy_mem8x4_c(unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch);
+void vp8_copy_mem8x4_mmx(unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch);
+RTCD_EXTERN void (*vp8_copy_mem8x4)(unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch);
+
+void vp8_copy_mem8x8_c(unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch);
+void vp8_copy_mem8x8_mmx(unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch);
+RTCD_EXTERN void (*vp8_copy_mem8x8)(unsigned char *src, int src_pitch, unsigned char *dst, int dst_pitch);
+
+void vp8_dc_only_idct_add_c(short input, unsigned char *pred, int pred_stride, unsigned char *dst, int dst_stride);
+void vp8_dc_only_idct_add_mmx(short input, unsigned char *pred, int pred_stride, unsigned char *dst, int dst_stride);
+RTCD_EXTERN void (*vp8_dc_only_idct_add)(short input, unsigned char *pred, int pred_stride, unsigned char *dst, int dst_stride);
+
+void vp8_dequant_idct_add_c(short *input, short *dq, unsigned char *output, int stride);
+void vp8_dequant_idct_add_mmx(short *input, short *dq, unsigned char *output, int stride);
+RTCD_EXTERN void (*vp8_dequant_idct_add)(short *input, short *dq, unsigned char *output, int stride);
+
+void vp8_dequant_idct_add_uv_block_c(short *q, short *dq, unsigned char *dst_u, unsigned char *dst_v, int stride, char *eobs);
+void vp8_dequant_idct_add_uv_block_mmx(short *q, short *dq, unsigned char *dst_u, unsigned char *dst_v, int stride, char *eobs);
+void vp8_dequant_idct_add_uv_block_sse2(short *q, short *dq, unsigned char *dst_u, unsigned char *dst_v, int stride, char *eobs);
+RTCD_EXTERN void (*vp8_dequant_idct_add_uv_block)(short *q, short *dq, unsigned char *dst_u, unsigned char *dst_v, int stride, char *eobs);
+
+void vp8_dequant_idct_add_y_block_c(short *q, short *dq, unsigned char *dst, int stride, char *eobs);
+void vp8_dequant_idct_add_y_block_mmx(short *q, short *dq, unsigned char *dst, int stride, char *eobs);
+void vp8_dequant_idct_add_y_block_sse2(short *q, short *dq, unsigned char *dst, int stride, char *eobs);
+RTCD_EXTERN void (*vp8_dequant_idct_add_y_block)(short *q, short *dq, unsigned char *dst, int stride, char *eobs);
+
+void vp8_dequantize_b_c(struct blockd*, short *dqc);
+void vp8_dequantize_b_mmx(struct blockd*, short *dqc);
+RTCD_EXTERN void (*vp8_dequantize_b)(struct blockd*, short *dqc);
+
+void vp8_loop_filter_bh_c(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+void vp8_loop_filter_bh_mmx(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+void vp8_loop_filter_bh_sse2(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+RTCD_EXTERN void (*vp8_loop_filter_bh)(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+
+void vp8_loop_filter_bv_c(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+void vp8_loop_filter_bv_mmx(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+void vp8_loop_filter_bv_sse2(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+RTCD_EXTERN void (*vp8_loop_filter_bv)(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+
+void vp8_loop_filter_mbh_c(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+void vp8_loop_filter_mbh_mmx(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+void vp8_loop_filter_mbh_sse2(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+RTCD_EXTERN void (*vp8_loop_filter_mbh)(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+
+void vp8_loop_filter_mbv_c(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+void vp8_loop_filter_mbv_mmx(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+void vp8_loop_filter_mbv_sse2(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+RTCD_EXTERN void (*vp8_loop_filter_mbv)(unsigned char *y, unsigned char *u, unsigned char *v, int ystride, int uv_stride, struct loop_filter_info *lfi);
+
+void vp8_loop_filter_bhs_c(unsigned char *y, int ystride, const unsigned char *blimit);
+void vp8_loop_filter_bhs_mmx(unsigned char *y, int ystride, const unsigned char *blimit);
+void vp8_loop_filter_bhs_sse2(unsigned char *y, int ystride, const unsigned char *blimit);
+RTCD_EXTERN void (*vp8_loop_filter_simple_bh)(unsigned char *y, int ystride, const unsigned char *blimit);
+
+void vp8_loop_filter_bvs_c(unsigned char *y, int ystride, const unsigned char *blimit);
+void vp8_loop_filter_bvs_mmx(unsigned char *y, int ystride, const unsigned char *blimit);
+void vp8_loop_filter_bvs_sse2(unsigned char *y, int ystride, const unsigned char *blimit);
+RTCD_EXTERN void (*vp8_loop_filter_simple_bv)(unsigned char *y, int ystride, const unsigned char *blimit);
+
+void vp8_loop_filter_simple_horizontal_edge_c(unsigned char *y, int ystride, const unsigned char *blimit);
+void vp8_loop_filter_simple_horizontal_edge_mmx(unsigned char *y, int ystride, const unsigned char *blimit);
+void vp8_loop_filter_simple_horizontal_edge_sse2(unsigned char *y, int ystride, const unsigned char *blimit);
+RTCD_EXTERN void (*vp8_loop_filter_simple_mbh)(unsigned char *y, int ystride, const unsigned char *blimit);
+
+void vp8_loop_filter_simple_vertical_edge_c(unsigned char *y, int ystride, const unsigned char *blimit);
+void vp8_loop_filter_simple_vertical_edge_mmx(unsigned char *y, int ystride, const unsigned char *blimit);
+void vp8_loop_filter_simple_vertical_edge_sse2(unsigned char *y, int ystride, const unsigned char *blimit);
+RTCD_EXTERN void (*vp8_loop_filter_simple_mbv)(unsigned char *y, int ystride, const unsigned char *blimit);
+
+void vp8_short_idct4x4llm_c(short *input, unsigned char *pred, int pitch, unsigned char *dst, int dst_stride);
+void vp8_short_idct4x4llm_mmx(short *input, unsigned char *pred, int pitch, unsigned char *dst, int dst_stride);
+RTCD_EXTERN void (*vp8_short_idct4x4llm)(short *input, unsigned char *pred, int pitch, unsigned char *dst, int dst_stride);
+
+void vp8_short_inv_walsh4x4_c(short *input, short *output);
+void vp8_short_inv_walsh4x4_mmx(short *input, short *output);
+void vp8_short_inv_walsh4x4_sse2(short *input, short *output);
+RTCD_EXTERN void (*vp8_short_inv_walsh4x4)(short *input, short *output);
+
+void vp8_short_inv_walsh4x4_1_c(short *input, short *output);
+#define vp8_short_inv_walsh4x4_1 vp8_short_inv_walsh4x4_1_c
+
+void vp8_sixtap_predict16x16_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+void vp8_sixtap_predict16x16_mmx(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+void vp8_sixtap_predict16x16_sse2(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+void vp8_sixtap_predict16x16_ssse3(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+RTCD_EXTERN void (*vp8_sixtap_predict16x16)(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+
+void vp8_sixtap_predict4x4_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+void vp8_sixtap_predict4x4_mmx(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+void vp8_sixtap_predict4x4_ssse3(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+RTCD_EXTERN void (*vp8_sixtap_predict4x4)(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+
+void vp8_sixtap_predict8x4_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+void vp8_sixtap_predict8x4_mmx(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+void vp8_sixtap_predict8x4_sse2(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+void vp8_sixtap_predict8x4_ssse3(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+RTCD_EXTERN void (*vp8_sixtap_predict8x4)(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+
+void vp8_sixtap_predict8x8_c(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+void vp8_sixtap_predict8x8_mmx(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+void vp8_sixtap_predict8x8_sse2(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+void vp8_sixtap_predict8x8_ssse3(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+RTCD_EXTERN void (*vp8_sixtap_predict8x8)(unsigned char *src, int src_pitch, int xofst, int yofst, unsigned char *dst, int dst_pitch);
+
+void vp8_rtcd(void);
+
+#ifdef RTCD_C
+#include "vpx_ports/x86.h"
+static void setup_rtcd_internal(void)
+{
+    int flags = x86_simd_caps();
+
+    vp8_bilinear_predict16x16 = vp8_bilinear_predict16x16_c;
+    if (flags & HAS_MMX) vp8_bilinear_predict16x16 = vp8_bilinear_predict16x16_mmx;
+    if (flags & HAS_SSE2) vp8_bilinear_predict16x16 = vp8_bilinear_predict16x16_sse2;
+    if (flags & HAS_SSSE3) vp8_bilinear_predict16x16 = vp8_bilinear_predict16x16_ssse3;
+    vp8_bilinear_predict4x4 = vp8_bilinear_predict4x4_c;
+    if (flags & HAS_MMX) vp8_bilinear_predict4x4 = vp8_bilinear_predict4x4_mmx;
+    vp8_bilinear_predict8x4 = vp8_bilinear_predict8x4_c;
+    if (flags & HAS_MMX) vp8_bilinear_predict8x4 = vp8_bilinear_predict8x4_mmx;
+    vp8_bilinear_predict8x8 = vp8_bilinear_predict8x8_c;
+    if (flags & HAS_MMX) vp8_bilinear_predict8x8 = vp8_bilinear_predict8x8_mmx;
+    if (flags & HAS_SSE2) vp8_bilinear_predict8x8 = vp8_bilinear_predict8x8_sse2;
+    if (flags & HAS_SSSE3) vp8_bilinear_predict8x8 = vp8_bilinear_predict8x8_ssse3;
+    vp8_clear_system_state = vp8_clear_system_state_c;
+    if (flags & HAS_MMX) vp8_clear_system_state = vpx_reset_mmx_state;
+    vp8_copy_mem16x16 = vp8_copy_mem16x16_c;
+    if (flags & HAS_MMX) vp8_copy_mem16x16 = vp8_copy_mem16x16_mmx;
+    if (flags & HAS_SSE2) vp8_copy_mem16x16 = vp8_copy_mem16x16_sse2;
+    vp8_copy_mem8x4 = vp8_copy_mem8x4_c;
+    if (flags & HAS_MMX) vp8_copy_mem8x4 = vp8_copy_mem8x4_mmx;
+    vp8_copy_mem8x8 = vp8_copy_mem8x8_c;
+    if (flags & HAS_MMX) vp8_copy_mem8x8 = vp8_copy_mem8x8_mmx;
+    vp8_dc_only_idct_add = vp8_dc_only_idct_add_c;
+    if (flags & HAS_MMX) vp8_dc_only_idct_add = vp8_dc_only_idct_add_mmx;
+    vp8_dequant_idct_add = vp8_dequant_idct_add_c;
+    if (flags & HAS_MMX) vp8_dequant_idct_add = vp8_dequant_idct_add_mmx;
+    vp8_dequant_idct_add_uv_block = vp8_dequant_idct_add_uv_block_c;
+    if (flags & HAS_MMX) vp8_dequant_idct_add_uv_block = vp8_dequant_idct_add_uv_block_mmx;
+    if (flags & HAS_SSE2) vp8_dequant_idct_add_uv_block = vp8_dequant_idct_add_uv_block_sse2;
+    vp8_dequant_idct_add_y_block = vp8_dequant_idct_add_y_block_c;
+    if (flags & HAS_MMX) vp8_dequant_idct_add_y_block = vp8_dequant_idct_add_y_block_mmx;
+    if (flags & HAS_SSE2) vp8_dequant_idct_add_y_block = vp8_dequant_idct_add_y_block_sse2;
+    vp8_dequantize_b = vp8_dequantize_b_c;
+    if (flags & HAS_MMX) vp8_dequantize_b = vp8_dequantize_b_mmx;
+    vp8_loop_filter_bh = vp8_loop_filter_bh_c;
+    if (flags & HAS_MMX) vp8_loop_filter_bh = vp8_loop_filter_bh_mmx;
+    if (flags & HAS_SSE2) vp8_loop_filter_bh = vp8_loop_filter_bh_sse2;
+    vp8_loop_filter_bv = vp8_loop_filter_bv_c;
+    if (flags & HAS_MMX) vp8_loop_filter_bv = vp8_loop_filter_bv_mmx;
+    if (flags & HAS_SSE2) vp8_loop_filter_bv = vp8_loop_filter_bv_sse2;
+    vp8_loop_filter_mbh = vp8_loop_filter_mbh_c;
+    if (flags & HAS_MMX) vp8_loop_filter_mbh = vp8_loop_filter_mbh_mmx;
+    if (flags & HAS_SSE2) vp8_loop_filter_mbh = vp8_loop_filter_mbh_sse2;
+    vp8_loop_filter_mbv = vp8_loop_filter_mbv_c;
+    if (flags & HAS_MMX) vp8_loop_filter_mbv = vp8_loop_filter_mbv_mmx;
+    if (flags & HAS_SSE2) vp8_loop_filter_mbv = vp8_loop_filter_mbv_sse2;
+    vp8_loop_filter_simple_bh = vp8_loop_filter_bhs_c;
+    if (flags & HAS_MMX) vp8_loop_filter_simple_bh = vp8_loop_filter_bhs_mmx;
+    if (flags & HAS_SSE2) vp8_loop_filter_simple_bh = vp8_loop_filter_bhs_sse2;
+    vp8_loop_filter_simple_bv = vp8_loop_filter_bvs_c;
+    if (flags & HAS_MMX) vp8_loop_filter_simple_bv = vp8_loop_filter_bvs_mmx;
+    if (flags & HAS_SSE2) vp8_loop_filter_simple_bv = vp8_loop_filter_bvs_sse2;
+    vp8_loop_filter_simple_mbh = vp8_loop_filter_simple_horizontal_edge_c;
+    if (flags & HAS_MMX) vp8_loop_filter_simple_mbh = vp8_loop_filter_simple_horizontal_edge_mmx;
+    if (flags & HAS_SSE2) vp8_loop_filter_simple_mbh = vp8_loop_filter_simple_horizontal_edge_sse2;
+    vp8_loop_filter_simple_mbv = vp8_loop_filter_simple_vertical_edge_c;
+    if (flags & HAS_MMX) vp8_loop_filter_simple_mbv = vp8_loop_filter_simple_vertical_edge_mmx;
+    if (flags & HAS_SSE2) vp8_loop_filter_simple_mbv = vp8_loop_filter_simple_vertical_edge_sse2;
+    vp8_short_idct4x4llm = vp8_short_idct4x4llm_c;
+    if (flags & HAS_MMX) vp8_short_idct4x4llm = vp8_short_idct4x4llm_mmx;
+    vp8_short_inv_walsh4x4 = vp8_short_inv_walsh4x4_c;
+    if (flags & HAS_MMX) vp8_short_inv_walsh4x4 = vp8_short_inv_walsh4x4_mmx;
+    if (flags & HAS_SSE2) vp8_short_inv_walsh4x4 = vp8_short_inv_walsh4x4_sse2;
+    vp8_sixtap_predict16x16 = vp8_sixtap_predict16x16_c;
+    if (flags & HAS_MMX) vp8_sixtap_predict16x16 = vp8_sixtap_predict16x16_mmx;
+    if (flags & HAS_SSE2) vp8_sixtap_predict16x16 = vp8_sixtap_predict16x16_sse2;
+    if (flags & HAS_SSSE3) vp8_sixtap_predict16x16 = vp8_sixtap_predict16x16_ssse3;
+    vp8_sixtap_predict4x4 = vp8_sixtap_predict4x4_c;
+    if (flags & HAS_MMX) vp8_sixtap_predict4x4 = vp8_sixtap_predict4x4_mmx;
+    if (flags & HAS_SSSE3) vp8_sixtap_predict4x4 = vp8_sixtap_predict4x4_ssse3;
+    vp8_sixtap_predict8x4 = vp8_sixtap_predict8x4_c;
+    if (flags & HAS_MMX) vp8_sixtap_predict8x4 = vp8_sixtap_predict8x4_mmx;
+    if (flags & HAS_SSE2) vp8_sixtap_predict8x4 = vp8_sixtap_predict8x4_sse2;
+    if (flags & HAS_SSSE3) vp8_sixtap_predict8x4 = vp8_sixtap_predict8x4_ssse3;
+    vp8_sixtap_predict8x8 = vp8_sixtap_predict8x8_c;
+    if (flags & HAS_MMX) vp8_sixtap_predict8x8 = vp8_sixtap_predict8x8_mmx;
+    if (flags & HAS_SSE2) vp8_sixtap_predict8x8 = vp8_sixtap_predict8x8_sse2;
+    if (flags & HAS_SSSE3) vp8_sixtap_predict8x8 = vp8_sixtap_predict8x8_ssse3;
+}
+#endif
+
+#ifdef __cplusplus
+}  // extern "C"
+#endif
+
+#endif

thirdparty/libvpx/rtcd/vp9_rtcd_arm.h

@@ -0,0 +1,54 @@
+#ifndef VP9_RTCD_H_
+#define VP9_RTCD_H_
+
+#ifdef RTCD_C
+#define RTCD_EXTERN
+#else
+#define RTCD_EXTERN extern
+#endif
+
+/*
+ * VP9
+ */
+
+#include "vp9/common/vp9_common.h"
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+void vp9_iht16x16_256_add_c(const tran_low_t *input, uint8_t *output, int pitch, int tx_type);
+#define vp9_iht16x16_256_add vp9_iht16x16_256_add_c
+
+void vp9_iht4x4_16_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type);
+void vp9_iht4x4_16_add_neon(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type);
+RTCD_EXTERN void (*vp9_iht4x4_16_add)(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type);
+
+void vp9_iht8x8_64_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type);
+void vp9_iht8x8_64_add_neon(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type);
+RTCD_EXTERN void (*vp9_iht8x8_64_add)(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type);
+
+void vp9_rtcd(void);
+
+#ifdef RTCD_C
+#include "vpx_ports/arm.h"
+static void setup_rtcd_internal(void)
+{
+    int flags = arm_cpu_caps();
+
+    vp9_iht4x4_16_add = vp9_iht4x4_16_add_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vp9_iht4x4_16_add = vp9_iht4x4_16_add_neon;
+#endif
+    vp9_iht8x8_64_add = vp9_iht8x8_64_add_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vp9_iht8x8_64_add = vp9_iht8x8_64_add_neon;
+#endif
+}
+#endif
+
+#ifdef __cplusplus
+}  // extern "C"
+#endif
+
+#endif

thirdparty/libvpx/rtcd/vp9_rtcd_c.h

@@ -0,0 +1,41 @@
+#ifndef VP9_RTCD_H_
+#define VP9_RTCD_H_
+
+#ifdef RTCD_C
+#define RTCD_EXTERN
+#else
+#define RTCD_EXTERN extern
+#endif
+
+/*
+ * VP9
+ */
+
+#include "vp9/common/vp9_common.h"
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+void vp9_iht16x16_256_add_c(const tran_low_t *input, uint8_t *output, int pitch, int tx_type);
+#define vp9_iht16x16_256_add vp9_iht16x16_256_add_c
+
+void vp9_iht4x4_16_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type);
+#define vp9_iht4x4_16_add vp9_iht4x4_16_add_c
+
+void vp9_iht8x8_64_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type);
+#define vp9_iht8x8_64_add vp9_iht8x8_64_add_c
+
+void vp9_rtcd(void);
+
+#ifdef RTCD_C
+static void setup_rtcd_internal(void)
+{
+}
+#endif
+
+#ifdef __cplusplus
+}  // extern "C"
+#endif
+
+#endif

thirdparty/libvpx/rtcd/vp9_rtcd_x86.h

@@ -0,0 +1,55 @@
+#ifndef VP9_RTCD_H_
+#define VP9_RTCD_H_
+
+#ifdef RTCD_C
+#define RTCD_EXTERN
+#else
+#define RTCD_EXTERN extern
+#endif
+
+/*
+ * VP9
+ */
+
+#include "vp9/common/vp9_common.h"
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+void vp9_iht16x16_256_add_c(const tran_low_t *input, uint8_t *output, int pitch, int tx_type);
+void vp9_iht16x16_256_add_sse2(const tran_low_t *input, uint8_t *output, int pitch, int tx_type);
+RTCD_EXTERN void (*vp9_iht16x16_256_add)(const tran_low_t *input, uint8_t *output, int pitch, int tx_type);
+
+void vp9_iht4x4_16_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type);
+void vp9_iht4x4_16_add_sse2(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type);
+RTCD_EXTERN void (*vp9_iht4x4_16_add)(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type);
+
+void vp9_iht8x8_64_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type);
+void vp9_iht8x8_64_add_sse2(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type);
+RTCD_EXTERN void (*vp9_iht8x8_64_add)(const tran_low_t *input, uint8_t *dest, int dest_stride, int tx_type);
+
+void vp9_rtcd(void);
+
+#ifdef RTCD_C
+#include "vpx_ports/x86.h"
+static void setup_rtcd_internal(void)
+{
+    int flags = x86_simd_caps();
+
+    vp9_iht16x16_256_add = vp9_iht16x16_256_add_c;
+    if (flags & HAS_SSE2) vp9_iht16x16_256_add = vp9_iht16x16_256_add_sse2;
+
+    vp9_iht4x4_16_add = vp9_iht4x4_16_add_c;
+    if (flags & HAS_SSE2) vp9_iht4x4_16_add = vp9_iht4x4_16_add_sse2;
+
+    vp9_iht8x8_64_add = vp9_iht8x8_64_add_c;
+    if (flags & HAS_SSE2) vp9_iht8x8_64_add = vp9_iht8x8_64_add_sse2;
+}
+#endif
+
+#ifdef __cplusplus
+}  // extern "C"
+#endif
+
+#endif

thirdparty/libvpx/rtcd/vpx_dsp_rtcd_arm.h

@@ -0,0 +1,678 @@
+#ifndef VPX_DSP_RTCD_H_
+#define VPX_DSP_RTCD_H_
+
+#ifdef RTCD_C
+#define RTCD_EXTERN
+#else
+#define RTCD_EXTERN extern
+#endif
+
+/*
+ * DSP
+ */
+
+#include "vpx/vpx_integer.h"
+#include "vpx_dsp/vpx_dsp_common.h"
+
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+void vpx_convolve8_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+void vpx_convolve8_neon(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+RTCD_EXTERN void (*vpx_convolve8)(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+
+void vpx_convolve8_avg_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+void vpx_convolve8_avg_neon(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+RTCD_EXTERN void (*vpx_convolve8_avg)(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+
+void vpx_convolve8_avg_horiz_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+void vpx_convolve8_avg_horiz_neon(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+RTCD_EXTERN void (*vpx_convolve8_avg_horiz)(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+
+void vpx_convolve8_avg_vert_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+void vpx_convolve8_avg_vert_neon(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+RTCD_EXTERN void (*vpx_convolve8_avg_vert)(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+
+void vpx_convolve8_horiz_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+void vpx_convolve8_horiz_neon(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+RTCD_EXTERN void (*vpx_convolve8_horiz)(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+
+void vpx_convolve8_vert_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+void vpx_convolve8_vert_neon(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+RTCD_EXTERN void (*vpx_convolve8_vert)(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+
+void vpx_convolve_avg_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+void vpx_convolve_avg_neon(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+RTCD_EXTERN void (*vpx_convolve_avg)(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+
+void vpx_convolve_copy_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+void vpx_convolve_copy_neon(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+RTCD_EXTERN void (*vpx_convolve_copy)(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+
+void vpx_d117_predictor_16x16_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d117_predictor_16x16 vpx_d117_predictor_16x16_c
+
+void vpx_d117_predictor_32x32_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d117_predictor_32x32 vpx_d117_predictor_32x32_c
+
+void vpx_d117_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d117_predictor_4x4 vpx_d117_predictor_4x4_c
+
+void vpx_d117_predictor_8x8_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d117_predictor_8x8 vpx_d117_predictor_8x8_c
+
+void vpx_d135_predictor_16x16_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d135_predictor_16x16 vpx_d135_predictor_16x16_c
+
+void vpx_d135_predictor_32x32_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d135_predictor_32x32 vpx_d135_predictor_32x32_c
+
+void vpx_d135_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_d135_predictor_4x4_neon(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_d135_predictor_4x4)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_d135_predictor_8x8_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d135_predictor_8x8 vpx_d135_predictor_8x8_c
+
+void vpx_d153_predictor_16x16_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d153_predictor_16x16 vpx_d153_predictor_16x16_c
+
+void vpx_d153_predictor_32x32_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d153_predictor_32x32 vpx_d153_predictor_32x32_c
+
+void vpx_d153_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d153_predictor_4x4 vpx_d153_predictor_4x4_c
+
+void vpx_d153_predictor_8x8_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d153_predictor_8x8 vpx_d153_predictor_8x8_c
+
+void vpx_d207_predictor_16x16_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d207_predictor_16x16 vpx_d207_predictor_16x16_c
+
+void vpx_d207_predictor_32x32_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d207_predictor_32x32 vpx_d207_predictor_32x32_c
+
+void vpx_d207_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d207_predictor_4x4 vpx_d207_predictor_4x4_c
+
+void vpx_d207_predictor_8x8_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d207_predictor_8x8 vpx_d207_predictor_8x8_c
+
+void vpx_d207e_predictor_16x16_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d207e_predictor_16x16 vpx_d207e_predictor_16x16_c
+
+void vpx_d207e_predictor_32x32_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d207e_predictor_32x32 vpx_d207e_predictor_32x32_c
+
+void vpx_d207e_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d207e_predictor_4x4 vpx_d207e_predictor_4x4_c
+
+void vpx_d207e_predictor_8x8_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d207e_predictor_8x8 vpx_d207e_predictor_8x8_c
+
+void vpx_d45_predictor_16x16_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_d45_predictor_16x16_neon(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_d45_predictor_16x16)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_d45_predictor_32x32_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d45_predictor_32x32 vpx_d45_predictor_32x32_c
+
+void vpx_d45_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_d45_predictor_4x4_neon(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_d45_predictor_4x4)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_d45_predictor_8x8_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_d45_predictor_8x8_neon(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_d45_predictor_8x8)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_d45e_predictor_16x16_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d45e_predictor_16x16 vpx_d45e_predictor_16x16_c
+
+void vpx_d45e_predictor_32x32_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d45e_predictor_32x32 vpx_d45e_predictor_32x32_c
+
+void vpx_d45e_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d45e_predictor_4x4 vpx_d45e_predictor_4x4_c
+
+void vpx_d45e_predictor_8x8_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d45e_predictor_8x8 vpx_d45e_predictor_8x8_c
+
+void vpx_d63_predictor_16x16_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d63_predictor_16x16 vpx_d63_predictor_16x16_c
+
+void vpx_d63_predictor_32x32_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d63_predictor_32x32 vpx_d63_predictor_32x32_c
+
+void vpx_d63_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d63_predictor_4x4 vpx_d63_predictor_4x4_c
+
+void vpx_d63_predictor_8x8_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d63_predictor_8x8 vpx_d63_predictor_8x8_c
+
+void vpx_d63e_predictor_16x16_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d63e_predictor_16x16 vpx_d63e_predictor_16x16_c
+
+void vpx_d63e_predictor_32x32_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d63e_predictor_32x32 vpx_d63e_predictor_32x32_c
+
+void vpx_d63e_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d63e_predictor_4x4 vpx_d63e_predictor_4x4_c
+
+void vpx_d63e_predictor_8x8_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d63e_predictor_8x8 vpx_d63e_predictor_8x8_c
+
+void vpx_d63f_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d63f_predictor_4x4 vpx_d63f_predictor_4x4_c
+
+void vpx_dc_128_predictor_16x16_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_dc_128_predictor_16x16_neon(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_dc_128_predictor_16x16)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_dc_128_predictor_32x32_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_dc_128_predictor_32x32_neon(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_dc_128_predictor_32x32)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_dc_128_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_dc_128_predictor_4x4_neon(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_dc_128_predictor_4x4)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_dc_128_predictor_8x8_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_dc_128_predictor_8x8_neon(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_dc_128_predictor_8x8)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_dc_left_predictor_16x16_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_dc_left_predictor_16x16_neon(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_dc_left_predictor_16x16)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_dc_left_predictor_32x32_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_dc_left_predictor_32x32_neon(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_dc_left_predictor_32x32)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_dc_left_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_dc_left_predictor_4x4_neon(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_dc_left_predictor_4x4)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_dc_left_predictor_8x8_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_dc_left_predictor_8x8_neon(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_dc_left_predictor_8x8)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_dc_predictor_16x16_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_dc_predictor_16x16_neon(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_dc_predictor_16x16)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_dc_predictor_32x32_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_dc_predictor_32x32_neon(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_dc_predictor_32x32)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_dc_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_dc_predictor_4x4_neon(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_dc_predictor_4x4)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_dc_predictor_8x8_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_dc_predictor_8x8_neon(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_dc_predictor_8x8)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_dc_top_predictor_16x16_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_dc_top_predictor_16x16_neon(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_dc_top_predictor_16x16)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_dc_top_predictor_32x32_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_dc_top_predictor_32x32_neon(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_dc_top_predictor_32x32)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_dc_top_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_dc_top_predictor_4x4_neon(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_dc_top_predictor_4x4)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_dc_top_predictor_8x8_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_dc_top_predictor_8x8_neon(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_dc_top_predictor_8x8)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_h_predictor_16x16_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_h_predictor_16x16_neon(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_h_predictor_16x16)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_h_predictor_32x32_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_h_predictor_32x32_neon(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_h_predictor_32x32)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_h_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_h_predictor_4x4_neon(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_h_predictor_4x4)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_h_predictor_8x8_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_h_predictor_8x8_neon(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_h_predictor_8x8)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_he_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_he_predictor_4x4 vpx_he_predictor_4x4_c
+
+void vpx_idct16x16_10_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride);
+void vpx_idct16x16_10_add_neon(const tran_low_t *input, uint8_t *dest, int dest_stride);
+RTCD_EXTERN void (*vpx_idct16x16_10_add)(const tran_low_t *input, uint8_t *dest, int dest_stride);
+
+void vpx_idct16x16_1_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride);
+void vpx_idct16x16_1_add_neon(const tran_low_t *input, uint8_t *dest, int dest_stride);
+RTCD_EXTERN void (*vpx_idct16x16_1_add)(const tran_low_t *input, uint8_t *dest, int dest_stride);
+
+void vpx_idct16x16_256_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride);
+void vpx_idct16x16_256_add_neon(const tran_low_t *input, uint8_t *dest, int dest_stride);
+RTCD_EXTERN void (*vpx_idct16x16_256_add)(const tran_low_t *input, uint8_t *dest, int dest_stride);
+
+void vpx_idct32x32_1024_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride);
+void vpx_idct32x32_1024_add_neon(const tran_low_t *input, uint8_t *dest, int dest_stride);
+RTCD_EXTERN void (*vpx_idct32x32_1024_add)(const tran_low_t *input, uint8_t *dest, int dest_stride);
+
+void vpx_idct32x32_135_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride);
+void vpx_idct32x32_1024_add_neon(const tran_low_t *input, uint8_t *dest, int dest_stride);
+RTCD_EXTERN void (*vpx_idct32x32_135_add)(const tran_low_t *input, uint8_t *dest, int dest_stride);
+
+void vpx_idct32x32_1_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride);
+void vpx_idct32x32_1_add_neon(const tran_low_t *input, uint8_t *dest, int dest_stride);
+RTCD_EXTERN void (*vpx_idct32x32_1_add)(const tran_low_t *input, uint8_t *dest, int dest_stride);
+
+void vpx_idct32x32_34_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride);
+void vpx_idct32x32_1024_add_neon(const tran_low_t *input, uint8_t *dest, int dest_stride);
+RTCD_EXTERN void (*vpx_idct32x32_34_add)(const tran_low_t *input, uint8_t *dest, int dest_stride);
+
+void vpx_idct4x4_16_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride);
+void vpx_idct4x4_16_add_neon(const tran_low_t *input, uint8_t *dest, int dest_stride);
+RTCD_EXTERN void (*vpx_idct4x4_16_add)(const tran_low_t *input, uint8_t *dest, int dest_stride);
+
+void vpx_idct4x4_1_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride);
+void vpx_idct4x4_1_add_neon(const tran_low_t *input, uint8_t *dest, int dest_stride);
+RTCD_EXTERN void (*vpx_idct4x4_1_add)(const tran_low_t *input, uint8_t *dest, int dest_stride);
+
+void vpx_idct8x8_12_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride);
+void vpx_idct8x8_12_add_neon(const tran_low_t *input, uint8_t *dest, int dest_stride);
+RTCD_EXTERN void (*vpx_idct8x8_12_add)(const tran_low_t *input, uint8_t *dest, int dest_stride);
+
+void vpx_idct8x8_1_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride);
+void vpx_idct8x8_1_add_neon(const tran_low_t *input, uint8_t *dest, int dest_stride);
+RTCD_EXTERN void (*vpx_idct8x8_1_add)(const tran_low_t *input, uint8_t *dest, int dest_stride);
+
+void vpx_idct8x8_64_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride);
+void vpx_idct8x8_64_add_neon(const tran_low_t *input, uint8_t *dest, int dest_stride);
+RTCD_EXTERN void (*vpx_idct8x8_64_add)(const tran_low_t *input, uint8_t *dest, int dest_stride);
+
+void vpx_iwht4x4_16_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride);
+#define vpx_iwht4x4_16_add vpx_iwht4x4_16_add_c
+
+void vpx_iwht4x4_1_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride);
+#define vpx_iwht4x4_1_add vpx_iwht4x4_1_add_c
+
+void vpx_lpf_horizontal_4_c(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+void vpx_lpf_horizontal_4_neon(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+RTCD_EXTERN void (*vpx_lpf_horizontal_4)(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+
+void vpx_lpf_horizontal_4_dual_c(uint8_t *s, int pitch, const uint8_t *blimit0, const uint8_t *limit0, const uint8_t *thresh0, const uint8_t *blimit1, const uint8_t *limit1, const uint8_t *thresh1);
+void vpx_lpf_horizontal_4_dual_neon(uint8_t *s, int pitch, const uint8_t *blimit0, const uint8_t *limit0, const uint8_t *thresh0, const uint8_t *blimit1, const uint8_t *limit1, const uint8_t *thresh1);
+RTCD_EXTERN void (*vpx_lpf_horizontal_4_dual)(uint8_t *s, int pitch, const uint8_t *blimit0, const uint8_t *limit0, const uint8_t *thresh0, const uint8_t *blimit1, const uint8_t *limit1, const uint8_t *thresh1);
+
+void vpx_lpf_horizontal_8_c(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+void vpx_lpf_horizontal_8_neon(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+RTCD_EXTERN void (*vpx_lpf_horizontal_8)(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+
+void vpx_lpf_horizontal_8_dual_c(uint8_t *s, int pitch, const uint8_t *blimit0, const uint8_t *limit0, const uint8_t *thresh0, const uint8_t *blimit1, const uint8_t *limit1, const uint8_t *thresh1);
+void vpx_lpf_horizontal_8_dual_neon(uint8_t *s, int pitch, const uint8_t *blimit0, const uint8_t *limit0, const uint8_t *thresh0, const uint8_t *blimit1, const uint8_t *limit1, const uint8_t *thresh1);
+RTCD_EXTERN void (*vpx_lpf_horizontal_8_dual)(uint8_t *s, int pitch, const uint8_t *blimit0, const uint8_t *limit0, const uint8_t *thresh0, const uint8_t *blimit1, const uint8_t *limit1, const uint8_t *thresh1);
+
+void vpx_lpf_horizontal_edge_16_c(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+void vpx_lpf_horizontal_edge_16_neon(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+RTCD_EXTERN void (*vpx_lpf_horizontal_edge_16)(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+
+void vpx_lpf_horizontal_edge_8_c(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+void vpx_lpf_horizontal_edge_8_neon(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+RTCD_EXTERN void (*vpx_lpf_horizontal_edge_8)(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+
+void vpx_lpf_vertical_16_c(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+void vpx_lpf_vertical_16_neon(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+RTCD_EXTERN void (*vpx_lpf_vertical_16)(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+
+void vpx_lpf_vertical_16_dual_c(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+void vpx_lpf_vertical_16_dual_neon(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+RTCD_EXTERN void (*vpx_lpf_vertical_16_dual)(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+
+void vpx_lpf_vertical_4_c(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+void vpx_lpf_vertical_4_neon(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+RTCD_EXTERN void (*vpx_lpf_vertical_4)(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+
+void vpx_lpf_vertical_4_dual_c(uint8_t *s, int pitch, const uint8_t *blimit0, const uint8_t *limit0, const uint8_t *thresh0, const uint8_t *blimit1, const uint8_t *limit1, const uint8_t *thresh1);
+void vpx_lpf_vertical_4_dual_neon(uint8_t *s, int pitch, const uint8_t *blimit0, const uint8_t *limit0, const uint8_t *thresh0, const uint8_t *blimit1, const uint8_t *limit1, const uint8_t *thresh1);
+RTCD_EXTERN void (*vpx_lpf_vertical_4_dual)(uint8_t *s, int pitch, const uint8_t *blimit0, const uint8_t *limit0, const uint8_t *thresh0, const uint8_t *blimit1, const uint8_t *limit1, const uint8_t *thresh1);
+
+void vpx_lpf_vertical_8_c(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+void vpx_lpf_vertical_8_neon(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+RTCD_EXTERN void (*vpx_lpf_vertical_8)(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+
+void vpx_lpf_vertical_8_dual_c(uint8_t *s, int pitch, const uint8_t *blimit0, const uint8_t *limit0, const uint8_t *thresh0, const uint8_t *blimit1, const uint8_t *limit1, const uint8_t *thresh1);
+void vpx_lpf_vertical_8_dual_neon(uint8_t *s, int pitch, const uint8_t *blimit0, const uint8_t *limit0, const uint8_t *thresh0, const uint8_t *blimit1, const uint8_t *limit1, const uint8_t *thresh1);
+RTCD_EXTERN void (*vpx_lpf_vertical_8_dual)(uint8_t *s, int pitch, const uint8_t *blimit0, const uint8_t *limit0, const uint8_t *thresh0, const uint8_t *blimit1, const uint8_t *limit1, const uint8_t *thresh1);
+
+void vpx_scaled_2d_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+#define vpx_scaled_2d vpx_scaled_2d_c
+
+void vpx_scaled_avg_2d_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+#define vpx_scaled_avg_2d vpx_scaled_avg_2d_c
+
+void vpx_scaled_avg_horiz_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+#define vpx_scaled_avg_horiz vpx_scaled_avg_horiz_c
+
+void vpx_scaled_avg_vert_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+#define vpx_scaled_avg_vert vpx_scaled_avg_vert_c
+
+void vpx_scaled_horiz_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+#define vpx_scaled_horiz vpx_scaled_horiz_c
+
+void vpx_scaled_vert_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+#define vpx_scaled_vert vpx_scaled_vert_c
+
+void vpx_tm_predictor_16x16_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_tm_predictor_16x16_neon(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_tm_predictor_16x16)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_tm_predictor_32x32_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_tm_predictor_32x32_neon(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_tm_predictor_32x32)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_tm_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_tm_predictor_4x4_neon(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_tm_predictor_4x4)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_tm_predictor_8x8_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_tm_predictor_8x8_neon(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_tm_predictor_8x8)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_v_predictor_16x16_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_v_predictor_16x16_neon(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_v_predictor_16x16)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_v_predictor_32x32_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_v_predictor_32x32_neon(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_v_predictor_32x32)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_v_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_v_predictor_4x4_neon(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_v_predictor_4x4)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_v_predictor_8x8_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_v_predictor_8x8_neon(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_v_predictor_8x8)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_ve_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_ve_predictor_4x4 vpx_ve_predictor_4x4_c
+
+void vpx_dsp_rtcd(void);
+
+#ifdef RTCD_C
+#include "vpx_ports/arm.h"
+static void setup_rtcd_internal(void)
+{
+    int flags = arm_cpu_caps();
+
+    vpx_convolve8 = vpx_convolve8_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_convolve8 = vpx_convolve8_neon;
+#endif
+    vpx_convolve8_avg = vpx_convolve8_avg_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_convolve8_avg = vpx_convolve8_avg_neon;
+#endif
+    vpx_convolve8_avg_horiz = vpx_convolve8_avg_horiz_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_convolve8_avg_horiz = vpx_convolve8_avg_horiz_neon;
+#endif
+    vpx_convolve8_avg_vert = vpx_convolve8_avg_vert_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_convolve8_avg_vert = vpx_convolve8_avg_vert_neon;
+#endif
+    vpx_convolve8_horiz = vpx_convolve8_horiz_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_convolve8_horiz = vpx_convolve8_horiz_neon;
+#endif
+    vpx_convolve8_vert = vpx_convolve8_vert_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_convolve8_vert = vpx_convolve8_vert_neon;
+#endif
+    vpx_convolve_avg = vpx_convolve_avg_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_convolve_avg = vpx_convolve_avg_neon;
+#endif
+    vpx_convolve_copy = vpx_convolve_copy_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_convolve_copy = vpx_convolve_copy_neon;
+#endif
+    vpx_d135_predictor_4x4 = vpx_d135_predictor_4x4_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_d135_predictor_4x4 = vpx_d135_predictor_4x4_neon;
+#endif
+    vpx_d45_predictor_16x16 = vpx_d45_predictor_16x16_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_d45_predictor_16x16 = vpx_d45_predictor_16x16_neon;
+#endif
+    vpx_d45_predictor_4x4 = vpx_d45_predictor_4x4_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_d45_predictor_4x4 = vpx_d45_predictor_4x4_neon;
+#endif
+    vpx_d45_predictor_8x8 = vpx_d45_predictor_8x8_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_d45_predictor_8x8 = vpx_d45_predictor_8x8_neon;
+#endif
+    vpx_dc_128_predictor_16x16 = vpx_dc_128_predictor_16x16_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_dc_128_predictor_16x16 = vpx_dc_128_predictor_16x16_neon;
+#endif
+    vpx_dc_128_predictor_32x32 = vpx_dc_128_predictor_32x32_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_dc_128_predictor_32x32 = vpx_dc_128_predictor_32x32_neon;
+#endif
+    vpx_dc_128_predictor_4x4 = vpx_dc_128_predictor_4x4_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_dc_128_predictor_4x4 = vpx_dc_128_predictor_4x4_neon;
+#endif
+    vpx_dc_128_predictor_8x8 = vpx_dc_128_predictor_8x8_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_dc_128_predictor_8x8 = vpx_dc_128_predictor_8x8_neon;
+#endif
+    vpx_dc_left_predictor_16x16 = vpx_dc_left_predictor_16x16_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_dc_left_predictor_16x16 = vpx_dc_left_predictor_16x16_neon;
+#endif
+    vpx_dc_left_predictor_32x32 = vpx_dc_left_predictor_32x32_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_dc_left_predictor_32x32 = vpx_dc_left_predictor_32x32_neon;
+#endif
+    vpx_dc_left_predictor_4x4 = vpx_dc_left_predictor_4x4_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_dc_left_predictor_4x4 = vpx_dc_left_predictor_4x4_neon;
+#endif
+    vpx_dc_left_predictor_8x8 = vpx_dc_left_predictor_8x8_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_dc_left_predictor_8x8 = vpx_dc_left_predictor_8x8_neon;
+#endif
+    vpx_dc_predictor_16x16 = vpx_dc_predictor_16x16_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_dc_predictor_16x16 = vpx_dc_predictor_16x16_neon;
+#endif
+    vpx_dc_predictor_32x32 = vpx_dc_predictor_32x32_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_dc_predictor_32x32 = vpx_dc_predictor_32x32_neon;
+#endif
+    vpx_dc_predictor_4x4 = vpx_dc_predictor_4x4_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_dc_predictor_4x4 = vpx_dc_predictor_4x4_neon;
+#endif
+    vpx_dc_predictor_8x8 = vpx_dc_predictor_8x8_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_dc_predictor_8x8 = vpx_dc_predictor_8x8_neon;
+#endif
+    vpx_dc_top_predictor_16x16 = vpx_dc_top_predictor_16x16_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_dc_top_predictor_16x16 = vpx_dc_top_predictor_16x16_neon;
+#endif
+    vpx_dc_top_predictor_32x32 = vpx_dc_top_predictor_32x32_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_dc_top_predictor_32x32 = vpx_dc_top_predictor_32x32_neon;
+#endif
+    vpx_dc_top_predictor_4x4 = vpx_dc_top_predictor_4x4_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_dc_top_predictor_4x4 = vpx_dc_top_predictor_4x4_neon;
+#endif
+    vpx_dc_top_predictor_8x8 = vpx_dc_top_predictor_8x8_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_dc_top_predictor_8x8 = vpx_dc_top_predictor_8x8_neon;
+#endif
+    vpx_h_predictor_16x16 = vpx_h_predictor_16x16_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_h_predictor_16x16 = vpx_h_predictor_16x16_neon;
+#endif
+    vpx_h_predictor_32x32 = vpx_h_predictor_32x32_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_h_predictor_32x32 = vpx_h_predictor_32x32_neon;
+#endif
+    vpx_h_predictor_4x4 = vpx_h_predictor_4x4_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_h_predictor_4x4 = vpx_h_predictor_4x4_neon;
+#endif
+    vpx_h_predictor_8x8 = vpx_h_predictor_8x8_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_h_predictor_8x8 = vpx_h_predictor_8x8_neon;
+#endif
+    vpx_idct16x16_10_add = vpx_idct16x16_10_add_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_idct16x16_10_add = vpx_idct16x16_10_add_neon;
+#endif
+    vpx_idct16x16_1_add = vpx_idct16x16_1_add_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_idct16x16_1_add = vpx_idct16x16_1_add_neon;
+#endif
+    vpx_idct16x16_256_add = vpx_idct16x16_256_add_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_idct16x16_256_add = vpx_idct16x16_256_add_neon;
+#endif
+    vpx_idct32x32_1024_add = vpx_idct32x32_1024_add_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_idct32x32_1024_add = vpx_idct32x32_1024_add_neon;
+#endif
+    vpx_idct32x32_135_add = vpx_idct32x32_135_add_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_idct32x32_135_add = vpx_idct32x32_1024_add_neon;
+#endif
+    vpx_idct32x32_1_add = vpx_idct32x32_1_add_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_idct32x32_1_add = vpx_idct32x32_1_add_neon;
+#endif
+    vpx_idct32x32_34_add = vpx_idct32x32_34_add_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_idct32x32_34_add = vpx_idct32x32_1024_add_neon;
+#endif
+    vpx_idct4x4_16_add = vpx_idct4x4_16_add_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_idct4x4_16_add = vpx_idct4x4_16_add_neon;
+#endif
+    vpx_idct4x4_1_add = vpx_idct4x4_1_add_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_idct4x4_1_add = vpx_idct4x4_1_add_neon;
+#endif
+    vpx_idct8x8_12_add = vpx_idct8x8_12_add_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_idct8x8_12_add = vpx_idct8x8_12_add_neon;
+#endif
+    vpx_idct8x8_1_add = vpx_idct8x8_1_add_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_idct8x8_1_add = vpx_idct8x8_1_add_neon;
+#endif
+    vpx_idct8x8_64_add = vpx_idct8x8_64_add_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_idct8x8_64_add = vpx_idct8x8_64_add_neon;
+#endif
+    vpx_lpf_horizontal_4 = vpx_lpf_horizontal_4_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_lpf_horizontal_4 = vpx_lpf_horizontal_4_neon;
+#endif
+    vpx_lpf_horizontal_4_dual = vpx_lpf_horizontal_4_dual_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_lpf_horizontal_4_dual = vpx_lpf_horizontal_4_dual_neon;
+#endif
+    vpx_lpf_horizontal_8 = vpx_lpf_horizontal_8_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_lpf_horizontal_8 = vpx_lpf_horizontal_8_neon;
+#endif
+    vpx_lpf_horizontal_8_dual = vpx_lpf_horizontal_8_dual_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_lpf_horizontal_8_dual = vpx_lpf_horizontal_8_dual_neon;
+#endif
+    vpx_lpf_horizontal_edge_16 = vpx_lpf_horizontal_edge_16_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_lpf_horizontal_edge_16 = vpx_lpf_horizontal_edge_16_neon;
+#endif
+    vpx_lpf_horizontal_edge_8 = vpx_lpf_horizontal_edge_8_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_lpf_horizontal_edge_8 = vpx_lpf_horizontal_edge_8_neon;
+#endif
+    vpx_lpf_vertical_16 = vpx_lpf_vertical_16_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_lpf_vertical_16 = vpx_lpf_vertical_16_neon;
+#endif
+    vpx_lpf_vertical_16_dual = vpx_lpf_vertical_16_dual_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_lpf_vertical_16_dual = vpx_lpf_vertical_16_dual_neon;
+#endif
+    vpx_lpf_vertical_4 = vpx_lpf_vertical_4_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_lpf_vertical_4 = vpx_lpf_vertical_4_neon;
+#endif
+    vpx_lpf_vertical_4_dual = vpx_lpf_vertical_4_dual_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_lpf_vertical_4_dual = vpx_lpf_vertical_4_dual_neon;
+#endif
+    vpx_lpf_vertical_8 = vpx_lpf_vertical_8_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_lpf_vertical_8 = vpx_lpf_vertical_8_neon;
+#endif
+    vpx_lpf_vertical_8_dual = vpx_lpf_vertical_8_dual_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_lpf_vertical_8_dual = vpx_lpf_vertical_8_dual_neon;
+#endif
+    vpx_tm_predictor_16x16 = vpx_tm_predictor_16x16_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_tm_predictor_16x16 = vpx_tm_predictor_16x16_neon;
+#endif
+    vpx_tm_predictor_32x32 = vpx_tm_predictor_32x32_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_tm_predictor_32x32 = vpx_tm_predictor_32x32_neon;
+#endif
+    vpx_tm_predictor_4x4 = vpx_tm_predictor_4x4_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_tm_predictor_4x4 = vpx_tm_predictor_4x4_neon;
+#endif
+    vpx_tm_predictor_8x8 = vpx_tm_predictor_8x8_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_tm_predictor_8x8 = vpx_tm_predictor_8x8_neon;
+#endif
+    vpx_v_predictor_16x16 = vpx_v_predictor_16x16_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_v_predictor_16x16 = vpx_v_predictor_16x16_neon;
+#endif
+    vpx_v_predictor_32x32 = vpx_v_predictor_32x32_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_v_predictor_32x32 = vpx_v_predictor_32x32_neon;
+#endif
+    vpx_v_predictor_4x4 = vpx_v_predictor_4x4_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_v_predictor_4x4 = vpx_v_predictor_4x4_neon;
+#endif
+    vpx_v_predictor_8x8 = vpx_v_predictor_8x8_c;
+#if HAVE_NEON
+    if (flags & HAS_NEON) vpx_v_predictor_8x8 = vpx_v_predictor_8x8_neon;
+#endif
+}
+#endif
+
+#ifdef __cplusplus
+}  // extern "C"
+#endif
+
+#endif

thirdparty/libvpx/rtcd/vpx_dsp_rtcd_c.h

@@ -0,0 +1,355 @@
+#ifndef VPX_DSP_RTCD_H_
+#define VPX_DSP_RTCD_H_
+
+#ifdef RTCD_C
+#define RTCD_EXTERN
+#else
+#define RTCD_EXTERN extern
+#endif
+
+/*
+ * DSP
+ */
+
+#include "vpx/vpx_integer.h"
+#include "vpx_dsp/vpx_dsp_common.h"
+
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+void vpx_convolve8_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+#define vpx_convolve8 vpx_convolve8_c
+
+void vpx_convolve8_avg_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+#define vpx_convolve8_avg vpx_convolve8_avg_c
+
+void vpx_convolve8_avg_horiz_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+#define vpx_convolve8_avg_horiz vpx_convolve8_avg_horiz_c
+
+void vpx_convolve8_avg_vert_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+#define vpx_convolve8_avg_vert vpx_convolve8_avg_vert_c
+
+void vpx_convolve8_horiz_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+#define vpx_convolve8_horiz vpx_convolve8_horiz_c
+
+void vpx_convolve8_vert_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+#define vpx_convolve8_vert vpx_convolve8_vert_c
+
+void vpx_convolve_avg_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+#define vpx_convolve_avg vpx_convolve_avg_c
+
+void vpx_convolve_copy_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+#define vpx_convolve_copy vpx_convolve_copy_c
+
+void vpx_d117_predictor_16x16_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d117_predictor_16x16 vpx_d117_predictor_16x16_c
+
+void vpx_d117_predictor_32x32_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d117_predictor_32x32 vpx_d117_predictor_32x32_c
+
+void vpx_d117_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d117_predictor_4x4 vpx_d117_predictor_4x4_c
+
+void vpx_d117_predictor_8x8_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d117_predictor_8x8 vpx_d117_predictor_8x8_c
+
+void vpx_d135_predictor_16x16_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d135_predictor_16x16 vpx_d135_predictor_16x16_c
+
+void vpx_d135_predictor_32x32_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d135_predictor_32x32 vpx_d135_predictor_32x32_c
+
+void vpx_d135_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d135_predictor_4x4 vpx_d135_predictor_4x4_c
+
+void vpx_d135_predictor_8x8_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d135_predictor_8x8 vpx_d135_predictor_8x8_c
+
+void vpx_d153_predictor_16x16_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d153_predictor_16x16 vpx_d153_predictor_16x16_c
+
+void vpx_d153_predictor_32x32_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d153_predictor_32x32 vpx_d153_predictor_32x32_c
+
+void vpx_d153_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d153_predictor_4x4 vpx_d153_predictor_4x4_c
+
+void vpx_d153_predictor_8x8_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d153_predictor_8x8 vpx_d153_predictor_8x8_c
+
+void vpx_d207_predictor_16x16_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d207_predictor_16x16 vpx_d207_predictor_16x16_c
+
+void vpx_d207_predictor_32x32_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d207_predictor_32x32 vpx_d207_predictor_32x32_c
+
+void vpx_d207_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d207_predictor_4x4 vpx_d207_predictor_4x4_c
+
+void vpx_d207_predictor_8x8_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d207_predictor_8x8 vpx_d207_predictor_8x8_c
+
+void vpx_d207e_predictor_16x16_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d207e_predictor_16x16 vpx_d207e_predictor_16x16_c
+
+void vpx_d207e_predictor_32x32_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d207e_predictor_32x32 vpx_d207e_predictor_32x32_c
+
+void vpx_d207e_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d207e_predictor_4x4 vpx_d207e_predictor_4x4_c
+
+void vpx_d207e_predictor_8x8_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d207e_predictor_8x8 vpx_d207e_predictor_8x8_c
+
+void vpx_d45_predictor_16x16_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d45_predictor_16x16 vpx_d45_predictor_16x16_c
+
+void vpx_d45_predictor_32x32_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d45_predictor_32x32 vpx_d45_predictor_32x32_c
+
+void vpx_d45_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d45_predictor_4x4 vpx_d45_predictor_4x4_c
+
+void vpx_d45_predictor_8x8_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d45_predictor_8x8 vpx_d45_predictor_8x8_c
+
+void vpx_d45e_predictor_16x16_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d45e_predictor_16x16 vpx_d45e_predictor_16x16_c
+
+void vpx_d45e_predictor_32x32_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d45e_predictor_32x32 vpx_d45e_predictor_32x32_c
+
+void vpx_d45e_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d45e_predictor_4x4 vpx_d45e_predictor_4x4_c
+
+void vpx_d45e_predictor_8x8_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d45e_predictor_8x8 vpx_d45e_predictor_8x8_c
+
+void vpx_d63_predictor_16x16_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d63_predictor_16x16 vpx_d63_predictor_16x16_c
+
+void vpx_d63_predictor_32x32_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d63_predictor_32x32 vpx_d63_predictor_32x32_c
+
+void vpx_d63_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d63_predictor_4x4 vpx_d63_predictor_4x4_c
+
+void vpx_d63_predictor_8x8_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d63_predictor_8x8 vpx_d63_predictor_8x8_c
+
+void vpx_d63e_predictor_16x16_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d63e_predictor_16x16 vpx_d63e_predictor_16x16_c
+
+void vpx_d63e_predictor_32x32_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d63e_predictor_32x32 vpx_d63e_predictor_32x32_c
+
+void vpx_d63e_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d63e_predictor_4x4 vpx_d63e_predictor_4x4_c
+
+void vpx_d63e_predictor_8x8_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d63e_predictor_8x8 vpx_d63e_predictor_8x8_c
+
+void vpx_d63f_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d63f_predictor_4x4 vpx_d63f_predictor_4x4_c
+
+void vpx_dc_128_predictor_16x16_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_dc_128_predictor_16x16 vpx_dc_128_predictor_16x16_c
+
+void vpx_dc_128_predictor_32x32_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_dc_128_predictor_32x32 vpx_dc_128_predictor_32x32_c
+
+void vpx_dc_128_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_dc_128_predictor_4x4 vpx_dc_128_predictor_4x4_c
+
+void vpx_dc_128_predictor_8x8_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_dc_128_predictor_8x8 vpx_dc_128_predictor_8x8_c
+
+void vpx_dc_left_predictor_16x16_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_dc_left_predictor_16x16 vpx_dc_left_predictor_16x16_c
+
+void vpx_dc_left_predictor_32x32_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_dc_left_predictor_32x32 vpx_dc_left_predictor_32x32_c
+
+void vpx_dc_left_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_dc_left_predictor_4x4 vpx_dc_left_predictor_4x4_c
+
+void vpx_dc_left_predictor_8x8_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_dc_left_predictor_8x8 vpx_dc_left_predictor_8x8_c
+
+void vpx_dc_predictor_16x16_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_dc_predictor_16x16 vpx_dc_predictor_16x16_c
+
+void vpx_dc_predictor_32x32_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_dc_predictor_32x32 vpx_dc_predictor_32x32_c
+
+void vpx_dc_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_dc_predictor_4x4 vpx_dc_predictor_4x4_c
+
+void vpx_dc_predictor_8x8_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_dc_predictor_8x8 vpx_dc_predictor_8x8_c
+
+void vpx_dc_top_predictor_16x16_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_dc_top_predictor_16x16 vpx_dc_top_predictor_16x16_c
+
+void vpx_dc_top_predictor_32x32_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_dc_top_predictor_32x32 vpx_dc_top_predictor_32x32_c
+
+void vpx_dc_top_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_dc_top_predictor_4x4 vpx_dc_top_predictor_4x4_c
+
+void vpx_dc_top_predictor_8x8_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_dc_top_predictor_8x8 vpx_dc_top_predictor_8x8_c
+
+void vpx_h_predictor_16x16_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_h_predictor_16x16 vpx_h_predictor_16x16_c
+
+void vpx_h_predictor_32x32_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_h_predictor_32x32 vpx_h_predictor_32x32_c
+
+void vpx_h_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_h_predictor_4x4 vpx_h_predictor_4x4_c
+
+void vpx_h_predictor_8x8_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_h_predictor_8x8 vpx_h_predictor_8x8_c
+
+void vpx_he_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_he_predictor_4x4 vpx_he_predictor_4x4_c
+
+void vpx_idct16x16_10_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride);
+#define vpx_idct16x16_10_add vpx_idct16x16_10_add_c
+
+void vpx_idct16x16_1_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride);
+#define vpx_idct16x16_1_add vpx_idct16x16_1_add_c
+
+void vpx_idct16x16_256_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride);
+#define vpx_idct16x16_256_add vpx_idct16x16_256_add_c
+
+void vpx_idct32x32_1024_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride);
+#define vpx_idct32x32_1024_add vpx_idct32x32_1024_add_c
+
+void vpx_idct32x32_135_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride);
+#define vpx_idct32x32_135_add vpx_idct32x32_135_add_c
+
+void vpx_idct32x32_1_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride);
+#define vpx_idct32x32_1_add vpx_idct32x32_1_add_c
+
+void vpx_idct32x32_34_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride);
+#define vpx_idct32x32_34_add vpx_idct32x32_34_add_c
+
+void vpx_idct4x4_16_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride);
+#define vpx_idct4x4_16_add vpx_idct4x4_16_add_c
+
+void vpx_idct4x4_1_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride);
+#define vpx_idct4x4_1_add vpx_idct4x4_1_add_c
+
+void vpx_idct8x8_12_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride);
+#define vpx_idct8x8_12_add vpx_idct8x8_12_add_c
+
+void vpx_idct8x8_1_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride);
+#define vpx_idct8x8_1_add vpx_idct8x8_1_add_c
+
+void vpx_idct8x8_64_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride);
+#define vpx_idct8x8_64_add vpx_idct8x8_64_add_c
+
+void vpx_iwht4x4_16_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride);
+#define vpx_iwht4x4_16_add vpx_iwht4x4_16_add_c
+
+void vpx_iwht4x4_1_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride);
+#define vpx_iwht4x4_1_add vpx_iwht4x4_1_add_c
+
+void vpx_lpf_horizontal_4_c(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+#define vpx_lpf_horizontal_4 vpx_lpf_horizontal_4_c
+
+void vpx_lpf_horizontal_4_dual_c(uint8_t *s, int pitch, const uint8_t *blimit0, const uint8_t *limit0, const uint8_t *thresh0, const uint8_t *blimit1, const uint8_t *limit1, const uint8_t *thresh1);
+#define vpx_lpf_horizontal_4_dual vpx_lpf_horizontal_4_dual_c
+
+void vpx_lpf_horizontal_8_c(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+#define vpx_lpf_horizontal_8 vpx_lpf_horizontal_8_c
+
+void vpx_lpf_horizontal_8_dual_c(uint8_t *s, int pitch, const uint8_t *blimit0, const uint8_t *limit0, const uint8_t *thresh0, const uint8_t *blimit1, const uint8_t *limit1, const uint8_t *thresh1);
+#define vpx_lpf_horizontal_8_dual vpx_lpf_horizontal_8_dual_c
+
+void vpx_lpf_horizontal_edge_16_c(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+#define vpx_lpf_horizontal_edge_16 vpx_lpf_horizontal_edge_16_c
+
+void vpx_lpf_horizontal_edge_8_c(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+#define vpx_lpf_horizontal_edge_8 vpx_lpf_horizontal_edge_8_c
+
+void vpx_lpf_vertical_16_c(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+#define vpx_lpf_vertical_16 vpx_lpf_vertical_16_c
+
+void vpx_lpf_vertical_16_dual_c(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+#define vpx_lpf_vertical_16_dual vpx_lpf_vertical_16_dual_c
+
+void vpx_lpf_vertical_4_c(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+#define vpx_lpf_vertical_4 vpx_lpf_vertical_4_c
+
+void vpx_lpf_vertical_4_dual_c(uint8_t *s, int pitch, const uint8_t *blimit0, const uint8_t *limit0, const uint8_t *thresh0, const uint8_t *blimit1, const uint8_t *limit1, const uint8_t *thresh1);
+#define vpx_lpf_vertical_4_dual vpx_lpf_vertical_4_dual_c
+
+void vpx_lpf_vertical_8_c(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+#define vpx_lpf_vertical_8 vpx_lpf_vertical_8_c
+
+void vpx_lpf_vertical_8_dual_c(uint8_t *s, int pitch, const uint8_t *blimit0, const uint8_t *limit0, const uint8_t *thresh0, const uint8_t *blimit1, const uint8_t *limit1, const uint8_t *thresh1);
+#define vpx_lpf_vertical_8_dual vpx_lpf_vertical_8_dual_c
+
+void vpx_scaled_2d_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+#define vpx_scaled_2d vpx_scaled_2d_c
+
+void vpx_scaled_avg_2d_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+#define vpx_scaled_avg_2d vpx_scaled_avg_2d_c
+
+void vpx_scaled_avg_horiz_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+#define vpx_scaled_avg_horiz vpx_scaled_avg_horiz_c
+
+void vpx_scaled_avg_vert_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+#define vpx_scaled_avg_vert vpx_scaled_avg_vert_c
+
+void vpx_scaled_horiz_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+#define vpx_scaled_horiz vpx_scaled_horiz_c
+
+void vpx_scaled_vert_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+#define vpx_scaled_vert vpx_scaled_vert_c
+
+void vpx_tm_predictor_16x16_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_tm_predictor_16x16 vpx_tm_predictor_16x16_c
+
+void vpx_tm_predictor_32x32_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_tm_predictor_32x32 vpx_tm_predictor_32x32_c
+
+void vpx_tm_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_tm_predictor_4x4 vpx_tm_predictor_4x4_c
+
+void vpx_tm_predictor_8x8_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_tm_predictor_8x8 vpx_tm_predictor_8x8_c
+
+void vpx_v_predictor_16x16_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_v_predictor_16x16 vpx_v_predictor_16x16_c
+
+void vpx_v_predictor_32x32_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_v_predictor_32x32 vpx_v_predictor_32x32_c
+
+void vpx_v_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_v_predictor_4x4 vpx_v_predictor_4x4_c
+
+void vpx_v_predictor_8x8_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_v_predictor_8x8 vpx_v_predictor_8x8_c
+
+void vpx_ve_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_ve_predictor_4x4 vpx_ve_predictor_4x4_c
+
+void vpx_dsp_rtcd(void);
+
+#ifdef RTCD_C
+static void setup_rtcd_internal(void)
+{
+}
+#endif
+
+#ifdef __cplusplus
+}  // extern "C"
+#endif
+
+#endif

thirdparty/libvpx/rtcd/vpx_dsp_rtcd_x86.h

@@ -0,0 +1,614 @@
+#ifndef VPX_DSP_RTCD_H_
+#define VPX_DSP_RTCD_H_
+
+#ifdef RTCD_C
+#define RTCD_EXTERN
+#else
+#define RTCD_EXTERN extern
+#endif
+
+/*
+ * DSP
+ */
+
+#include "vpx/vpx_integer.h"
+#include "vpx_dsp/vpx_dsp_common.h"
+
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+void vpx_convolve8_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+void vpx_convolve8_sse2(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+void vpx_convolve8_ssse3(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+void vpx_convolve8_avx2(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+RTCD_EXTERN void (*vpx_convolve8)(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+
+void vpx_convolve8_avg_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+void vpx_convolve8_avg_sse2(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+void vpx_convolve8_avg_ssse3(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+RTCD_EXTERN void (*vpx_convolve8_avg)(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+
+void vpx_convolve8_avg_horiz_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+void vpx_convolve8_avg_horiz_sse2(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+void vpx_convolve8_avg_horiz_ssse3(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+RTCD_EXTERN void (*vpx_convolve8_avg_horiz)(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+
+void vpx_convolve8_avg_vert_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+void vpx_convolve8_avg_vert_sse2(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+void vpx_convolve8_avg_vert_ssse3(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+RTCD_EXTERN void (*vpx_convolve8_avg_vert)(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+
+void vpx_convolve8_horiz_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+void vpx_convolve8_horiz_sse2(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+void vpx_convolve8_horiz_ssse3(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+void vpx_convolve8_horiz_avx2(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+RTCD_EXTERN void (*vpx_convolve8_horiz)(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+
+void vpx_convolve8_vert_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+void vpx_convolve8_vert_sse2(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+void vpx_convolve8_vert_ssse3(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+void vpx_convolve8_vert_avx2(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+RTCD_EXTERN void (*vpx_convolve8_vert)(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+
+void vpx_convolve_avg_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+void vpx_convolve_avg_sse2(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+RTCD_EXTERN void (*vpx_convolve_avg)(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+
+void vpx_convolve_copy_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+void vpx_convolve_copy_sse2(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+RTCD_EXTERN void (*vpx_convolve_copy)(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+
+void vpx_d117_predictor_16x16_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d117_predictor_16x16 vpx_d117_predictor_16x16_c
+
+void vpx_d117_predictor_32x32_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d117_predictor_32x32 vpx_d117_predictor_32x32_c
+
+void vpx_d117_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d117_predictor_4x4 vpx_d117_predictor_4x4_c
+
+void vpx_d117_predictor_8x8_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d117_predictor_8x8 vpx_d117_predictor_8x8_c
+
+void vpx_d135_predictor_16x16_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d135_predictor_16x16 vpx_d135_predictor_16x16_c
+
+void vpx_d135_predictor_32x32_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d135_predictor_32x32 vpx_d135_predictor_32x32_c
+
+void vpx_d135_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d135_predictor_4x4 vpx_d135_predictor_4x4_c
+
+void vpx_d135_predictor_8x8_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d135_predictor_8x8 vpx_d135_predictor_8x8_c
+
+void vpx_d153_predictor_16x16_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_d153_predictor_16x16_ssse3(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_d153_predictor_16x16)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_d153_predictor_32x32_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_d153_predictor_32x32_ssse3(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_d153_predictor_32x32)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_d153_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_d153_predictor_4x4_ssse3(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_d153_predictor_4x4)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_d153_predictor_8x8_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_d153_predictor_8x8_ssse3(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_d153_predictor_8x8)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_d207_predictor_16x16_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_d207_predictor_16x16_ssse3(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_d207_predictor_16x16)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_d207_predictor_32x32_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_d207_predictor_32x32_ssse3(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_d207_predictor_32x32)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_d207_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_d207_predictor_4x4_sse2(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_d207_predictor_4x4)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_d207_predictor_8x8_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_d207_predictor_8x8_ssse3(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_d207_predictor_8x8)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_d207e_predictor_16x16_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d207e_predictor_16x16 vpx_d207e_predictor_16x16_c
+
+void vpx_d207e_predictor_32x32_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d207e_predictor_32x32 vpx_d207e_predictor_32x32_c
+
+void vpx_d207e_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d207e_predictor_4x4 vpx_d207e_predictor_4x4_c
+
+void vpx_d207e_predictor_8x8_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d207e_predictor_8x8 vpx_d207e_predictor_8x8_c
+
+void vpx_d45_predictor_16x16_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_d45_predictor_16x16_ssse3(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_d45_predictor_16x16)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_d45_predictor_32x32_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_d45_predictor_32x32_ssse3(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_d45_predictor_32x32)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_d45_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_d45_predictor_4x4_sse2(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_d45_predictor_4x4)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_d45_predictor_8x8_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_d45_predictor_8x8_sse2(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_d45_predictor_8x8)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_d45e_predictor_16x16_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d45e_predictor_16x16 vpx_d45e_predictor_16x16_c
+
+void vpx_d45e_predictor_32x32_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d45e_predictor_32x32 vpx_d45e_predictor_32x32_c
+
+void vpx_d45e_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d45e_predictor_4x4 vpx_d45e_predictor_4x4_c
+
+void vpx_d45e_predictor_8x8_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d45e_predictor_8x8 vpx_d45e_predictor_8x8_c
+
+void vpx_d63_predictor_16x16_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_d63_predictor_16x16_ssse3(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_d63_predictor_16x16)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_d63_predictor_32x32_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_d63_predictor_32x32_ssse3(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_d63_predictor_32x32)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_d63_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_d63_predictor_4x4_ssse3(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_d63_predictor_4x4)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_d63_predictor_8x8_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_d63_predictor_8x8_ssse3(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_d63_predictor_8x8)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_d63e_predictor_16x16_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d63e_predictor_16x16 vpx_d63e_predictor_16x16_c
+
+void vpx_d63e_predictor_32x32_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d63e_predictor_32x32 vpx_d63e_predictor_32x32_c
+
+void vpx_d63e_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d63e_predictor_4x4 vpx_d63e_predictor_4x4_c
+
+void vpx_d63e_predictor_8x8_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d63e_predictor_8x8 vpx_d63e_predictor_8x8_c
+
+void vpx_d63f_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_d63f_predictor_4x4 vpx_d63f_predictor_4x4_c
+
+void vpx_dc_128_predictor_16x16_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_dc_128_predictor_16x16_sse2(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_dc_128_predictor_16x16)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_dc_128_predictor_32x32_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_dc_128_predictor_32x32_sse2(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_dc_128_predictor_32x32)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_dc_128_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_dc_128_predictor_4x4_sse2(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_dc_128_predictor_4x4)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_dc_128_predictor_8x8_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_dc_128_predictor_8x8_sse2(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_dc_128_predictor_8x8)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_dc_left_predictor_16x16_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_dc_left_predictor_16x16_sse2(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_dc_left_predictor_16x16)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_dc_left_predictor_32x32_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_dc_left_predictor_32x32_sse2(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_dc_left_predictor_32x32)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_dc_left_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_dc_left_predictor_4x4_sse2(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_dc_left_predictor_4x4)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_dc_left_predictor_8x8_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_dc_left_predictor_8x8_sse2(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_dc_left_predictor_8x8)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_dc_predictor_16x16_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_dc_predictor_16x16_sse2(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_dc_predictor_16x16)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_dc_predictor_32x32_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_dc_predictor_32x32_sse2(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_dc_predictor_32x32)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_dc_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_dc_predictor_4x4_sse2(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_dc_predictor_4x4)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_dc_predictor_8x8_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_dc_predictor_8x8_sse2(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_dc_predictor_8x8)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_dc_top_predictor_16x16_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_dc_top_predictor_16x16_sse2(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_dc_top_predictor_16x16)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_dc_top_predictor_32x32_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_dc_top_predictor_32x32_sse2(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_dc_top_predictor_32x32)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_dc_top_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_dc_top_predictor_4x4_sse2(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_dc_top_predictor_4x4)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_dc_top_predictor_8x8_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_dc_top_predictor_8x8_sse2(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_dc_top_predictor_8x8)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_h_predictor_16x16_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_h_predictor_16x16_sse2(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_h_predictor_16x16)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_h_predictor_32x32_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_h_predictor_32x32_sse2(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_h_predictor_32x32)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_h_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_h_predictor_4x4_sse2(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_h_predictor_4x4)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_h_predictor_8x8_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_h_predictor_8x8_sse2(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_h_predictor_8x8)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_he_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_he_predictor_4x4 vpx_he_predictor_4x4_c
+
+void vpx_idct16x16_10_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride);
+void vpx_idct16x16_10_add_sse2(const tran_low_t *input, uint8_t *dest, int dest_stride);
+RTCD_EXTERN void (*vpx_idct16x16_10_add)(const tran_low_t *input, uint8_t *dest, int dest_stride);
+
+void vpx_idct16x16_1_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride);
+void vpx_idct16x16_1_add_sse2(const tran_low_t *input, uint8_t *dest, int dest_stride);
+RTCD_EXTERN void (*vpx_idct16x16_1_add)(const tran_low_t *input, uint8_t *dest, int dest_stride);
+
+void vpx_idct16x16_256_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride);
+void vpx_idct16x16_256_add_sse2(const tran_low_t *input, uint8_t *dest, int dest_stride);
+RTCD_EXTERN void (*vpx_idct16x16_256_add)(const tran_low_t *input, uint8_t *dest, int dest_stride);
+
+void vpx_idct32x32_1024_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride);
+void vpx_idct32x32_1024_add_sse2(const tran_low_t *input, uint8_t *dest, int dest_stride);
+RTCD_EXTERN void (*vpx_idct32x32_1024_add)(const tran_low_t *input, uint8_t *dest, int dest_stride);
+
+void vpx_idct32x32_135_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride);
+void vpx_idct32x32_1024_add_sse2(const tran_low_t *input, uint8_t *dest, int dest_stride);
+RTCD_EXTERN void (*vpx_idct32x32_135_add)(const tran_low_t *input, uint8_t *dest, int dest_stride);
+
+void vpx_idct32x32_1_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride);
+void vpx_idct32x32_1_add_sse2(const tran_low_t *input, uint8_t *dest, int dest_stride);
+RTCD_EXTERN void (*vpx_idct32x32_1_add)(const tran_low_t *input, uint8_t *dest, int dest_stride);
+
+void vpx_idct32x32_34_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride);
+void vpx_idct32x32_34_add_sse2(const tran_low_t *input, uint8_t *dest, int dest_stride);
+RTCD_EXTERN void (*vpx_idct32x32_34_add)(const tran_low_t *input, uint8_t *dest, int dest_stride);
+
+void vpx_idct4x4_16_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride);
+void vpx_idct4x4_16_add_sse2(const tran_low_t *input, uint8_t *dest, int dest_stride);
+RTCD_EXTERN void (*vpx_idct4x4_16_add)(const tran_low_t *input, uint8_t *dest, int dest_stride);
+
+void vpx_idct4x4_1_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride);
+void vpx_idct4x4_1_add_sse2(const tran_low_t *input, uint8_t *dest, int dest_stride);
+RTCD_EXTERN void (*vpx_idct4x4_1_add)(const tran_low_t *input, uint8_t *dest, int dest_stride);
+
+void vpx_idct8x8_12_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride);
+void vpx_idct8x8_12_add_sse2(const tran_low_t *input, uint8_t *dest, int dest_stride);
+RTCD_EXTERN void (*vpx_idct8x8_12_add)(const tran_low_t *input, uint8_t *dest, int dest_stride);
+
+void vpx_idct8x8_1_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride);
+void vpx_idct8x8_1_add_sse2(const tran_low_t *input, uint8_t *dest, int dest_stride);
+RTCD_EXTERN void (*vpx_idct8x8_1_add)(const tran_low_t *input, uint8_t *dest, int dest_stride);
+
+void vpx_idct8x8_64_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride);
+void vpx_idct8x8_64_add_sse2(const tran_low_t *input, uint8_t *dest, int dest_stride);
+RTCD_EXTERN void (*vpx_idct8x8_64_add)(const tran_low_t *input, uint8_t *dest, int dest_stride);
+
+void vpx_iwht4x4_16_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride);
+void vpx_iwht4x4_16_add_sse2(const tran_low_t *input, uint8_t *dest, int dest_stride);
+RTCD_EXTERN void (*vpx_iwht4x4_16_add)(const tran_low_t *input, uint8_t *dest, int dest_stride);
+
+void vpx_iwht4x4_1_add_c(const tran_low_t *input, uint8_t *dest, int dest_stride);
+#define vpx_iwht4x4_1_add vpx_iwht4x4_1_add_c
+
+void vpx_lpf_horizontal_4_c(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+void vpx_lpf_horizontal_4_sse2(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+RTCD_EXTERN void (*vpx_lpf_horizontal_4)(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+
+void vpx_lpf_horizontal_4_dual_c(uint8_t *s, int pitch, const uint8_t *blimit0, const uint8_t *limit0, const uint8_t *thresh0, const uint8_t *blimit1, const uint8_t *limit1, const uint8_t *thresh1);
+void vpx_lpf_horizontal_4_dual_sse2(uint8_t *s, int pitch, const uint8_t *blimit0, const uint8_t *limit0, const uint8_t *thresh0, const uint8_t *blimit1, const uint8_t *limit1, const uint8_t *thresh1);
+RTCD_EXTERN void (*vpx_lpf_horizontal_4_dual)(uint8_t *s, int pitch, const uint8_t *blimit0, const uint8_t *limit0, const uint8_t *thresh0, const uint8_t *blimit1, const uint8_t *limit1, const uint8_t *thresh1);
+
+void vpx_lpf_horizontal_8_c(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+void vpx_lpf_horizontal_8_sse2(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+RTCD_EXTERN void (*vpx_lpf_horizontal_8)(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+
+void vpx_lpf_horizontal_8_dual_c(uint8_t *s, int pitch, const uint8_t *blimit0, const uint8_t *limit0, const uint8_t *thresh0, const uint8_t *blimit1, const uint8_t *limit1, const uint8_t *thresh1);
+void vpx_lpf_horizontal_8_dual_sse2(uint8_t *s, int pitch, const uint8_t *blimit0, const uint8_t *limit0, const uint8_t *thresh0, const uint8_t *blimit1, const uint8_t *limit1, const uint8_t *thresh1);
+RTCD_EXTERN void (*vpx_lpf_horizontal_8_dual)(uint8_t *s, int pitch, const uint8_t *blimit0, const uint8_t *limit0, const uint8_t *thresh0, const uint8_t *blimit1, const uint8_t *limit1, const uint8_t *thresh1);
+
+void vpx_lpf_horizontal_edge_16_c(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+void vpx_lpf_horizontal_edge_16_sse2(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+void vpx_lpf_horizontal_edge_16_avx2(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+RTCD_EXTERN void (*vpx_lpf_horizontal_edge_16)(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+
+void vpx_lpf_horizontal_edge_8_c(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+void vpx_lpf_horizontal_edge_8_sse2(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+void vpx_lpf_horizontal_edge_8_avx2(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+RTCD_EXTERN void (*vpx_lpf_horizontal_edge_8)(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+
+void vpx_lpf_vertical_16_c(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+void vpx_lpf_vertical_16_sse2(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+RTCD_EXTERN void (*vpx_lpf_vertical_16)(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+
+void vpx_lpf_vertical_16_dual_c(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+void vpx_lpf_vertical_16_dual_sse2(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+RTCD_EXTERN void (*vpx_lpf_vertical_16_dual)(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+
+void vpx_lpf_vertical_4_c(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+void vpx_lpf_vertical_4_sse2(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+RTCD_EXTERN void (*vpx_lpf_vertical_4)(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+
+void vpx_lpf_vertical_4_dual_c(uint8_t *s, int pitch, const uint8_t *blimit0, const uint8_t *limit0, const uint8_t *thresh0, const uint8_t *blimit1, const uint8_t *limit1, const uint8_t *thresh1);
+void vpx_lpf_vertical_4_dual_sse2(uint8_t *s, int pitch, const uint8_t *blimit0, const uint8_t *limit0, const uint8_t *thresh0, const uint8_t *blimit1, const uint8_t *limit1, const uint8_t *thresh1);
+RTCD_EXTERN void (*vpx_lpf_vertical_4_dual)(uint8_t *s, int pitch, const uint8_t *blimit0, const uint8_t *limit0, const uint8_t *thresh0, const uint8_t *blimit1, const uint8_t *limit1, const uint8_t *thresh1);
+
+void vpx_lpf_vertical_8_c(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+void vpx_lpf_vertical_8_sse2(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+RTCD_EXTERN void (*vpx_lpf_vertical_8)(uint8_t *s, int pitch, const uint8_t *blimit, const uint8_t *limit, const uint8_t *thresh);
+
+void vpx_lpf_vertical_8_dual_c(uint8_t *s, int pitch, const uint8_t *blimit0, const uint8_t *limit0, const uint8_t *thresh0, const uint8_t *blimit1, const uint8_t *limit1, const uint8_t *thresh1);
+void vpx_lpf_vertical_8_dual_sse2(uint8_t *s, int pitch, const uint8_t *blimit0, const uint8_t *limit0, const uint8_t *thresh0, const uint8_t *blimit1, const uint8_t *limit1, const uint8_t *thresh1);
+RTCD_EXTERN void (*vpx_lpf_vertical_8_dual)(uint8_t *s, int pitch, const uint8_t *blimit0, const uint8_t *limit0, const uint8_t *thresh0, const uint8_t *blimit1, const uint8_t *limit1, const uint8_t *thresh1);
+
+void vpx_scaled_2d_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+void vpx_scaled_2d_ssse3(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+RTCD_EXTERN void (*vpx_scaled_2d)(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+
+void vpx_scaled_avg_2d_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+#define vpx_scaled_avg_2d vpx_scaled_avg_2d_c
+
+void vpx_scaled_avg_horiz_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+#define vpx_scaled_avg_horiz vpx_scaled_avg_horiz_c
+
+void vpx_scaled_avg_vert_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+#define vpx_scaled_avg_vert vpx_scaled_avg_vert_c
+
+void vpx_scaled_horiz_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+#define vpx_scaled_horiz vpx_scaled_horiz_c
+
+void vpx_scaled_vert_c(const uint8_t *src, ptrdiff_t src_stride, uint8_t *dst, ptrdiff_t dst_stride, const int16_t *filter_x, int x_step_q4, const int16_t *filter_y, int y_step_q4, int w, int h);
+#define vpx_scaled_vert vpx_scaled_vert_c
+
+void vpx_tm_predictor_16x16_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_tm_predictor_16x16_sse2(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_tm_predictor_16x16)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_tm_predictor_32x32_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_tm_predictor_32x32_sse2(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_tm_predictor_32x32)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_tm_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_tm_predictor_4x4_sse2(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_tm_predictor_4x4)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_tm_predictor_8x8_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_tm_predictor_8x8_sse2(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_tm_predictor_8x8)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_v_predictor_16x16_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_v_predictor_16x16_sse2(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_v_predictor_16x16)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_v_predictor_32x32_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_v_predictor_32x32_sse2(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_v_predictor_32x32)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_v_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_v_predictor_4x4_sse2(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_v_predictor_4x4)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_v_predictor_8x8_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+void vpx_v_predictor_8x8_sse2(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+RTCD_EXTERN void (*vpx_v_predictor_8x8)(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+
+void vpx_ve_predictor_4x4_c(uint8_t *dst, ptrdiff_t y_stride, const uint8_t *above, const uint8_t *left);
+#define vpx_ve_predictor_4x4 vpx_ve_predictor_4x4_c
+
+void vpx_dsp_rtcd(void);
+
+#ifdef RTCD_C
+#include "vpx_ports/x86.h"
+static void setup_rtcd_internal(void)
+{
+    int flags = x86_simd_caps();
+
+    vpx_convolve8 = vpx_convolve8_c;
+    if (flags & HAS_SSE2) vpx_convolve8 = vpx_convolve8_sse2;
+    if (flags & HAS_SSSE3) vpx_convolve8 = vpx_convolve8_ssse3;
+    if (flags & HAS_AVX2) vpx_convolve8 = vpx_convolve8_avx2;
+    vpx_convolve8_avg = vpx_convolve8_avg_c;
+    if (flags & HAS_SSE2) vpx_convolve8_avg = vpx_convolve8_avg_sse2;
+    if (flags & HAS_SSSE3) vpx_convolve8_avg = vpx_convolve8_avg_ssse3;
+    vpx_convolve8_avg_horiz = vpx_convolve8_avg_horiz_c;
+    if (flags & HAS_SSE2) vpx_convolve8_avg_horiz = vpx_convolve8_avg_horiz_sse2;
+    if (flags & HAS_SSSE3) vpx_convolve8_avg_horiz = vpx_convolve8_avg_horiz_ssse3;
+    vpx_convolve8_avg_vert = vpx_convolve8_avg_vert_c;
+    if (flags & HAS_SSE2) vpx_convolve8_avg_vert = vpx_convolve8_avg_vert_sse2;
+    if (flags & HAS_SSSE3) vpx_convolve8_avg_vert = vpx_convolve8_avg_vert_ssse3;
+    vpx_convolve8_horiz = vpx_convolve8_horiz_c;
+    if (flags & HAS_SSE2) vpx_convolve8_horiz = vpx_convolve8_horiz_sse2;
+    if (flags & HAS_SSSE3) vpx_convolve8_horiz = vpx_convolve8_horiz_ssse3;
+    if (flags & HAS_AVX2) vpx_convolve8_horiz = vpx_convolve8_horiz_avx2;
+    vpx_convolve8_vert = vpx_convolve8_vert_c;
+    if (flags & HAS_SSE2) vpx_convolve8_vert = vpx_convolve8_vert_sse2;
+    if (flags & HAS_SSSE3) vpx_convolve8_vert = vpx_convolve8_vert_ssse3;
+    if (flags & HAS_AVX2) vpx_convolve8_vert = vpx_convolve8_vert_avx2;
+    vpx_convolve_avg = vpx_convolve_avg_c;
+    if (flags & HAS_SSE2) vpx_convolve_avg = vpx_convolve_avg_sse2;
+    vpx_convolve_copy = vpx_convolve_copy_c;
+    if (flags & HAS_SSE2) vpx_convolve_copy = vpx_convolve_copy_sse2;
+    vpx_d153_predictor_16x16 = vpx_d153_predictor_16x16_c;
+    if (flags & HAS_SSSE3) vpx_d153_predictor_16x16 = vpx_d153_predictor_16x16_ssse3;
+    vpx_d153_predictor_32x32 = vpx_d153_predictor_32x32_c;
+    if (flags & HAS_SSSE3) vpx_d153_predictor_32x32 = vpx_d153_predictor_32x32_ssse3;
+    vpx_d153_predictor_4x4 = vpx_d153_predictor_4x4_c;
+    if (flags & HAS_SSSE3) vpx_d153_predictor_4x4 = vpx_d153_predictor_4x4_ssse3;
+    vpx_d153_predictor_8x8 = vpx_d153_predictor_8x8_c;
+    if (flags & HAS_SSSE3) vpx_d153_predictor_8x8 = vpx_d153_predictor_8x8_ssse3;
+    vpx_d207_predictor_16x16 = vpx_d207_predictor_16x16_c;
+    if (flags & HAS_SSSE3) vpx_d207_predictor_16x16 = vpx_d207_predictor_16x16_ssse3;
+    vpx_d207_predictor_32x32 = vpx_d207_predictor_32x32_c;
+    if (flags & HAS_SSSE3) vpx_d207_predictor_32x32 = vpx_d207_predictor_32x32_ssse3;
+    vpx_d207_predictor_4x4 = vpx_d207_predictor_4x4_c;
+    if (flags & HAS_SSE2) vpx_d207_predictor_4x4 = vpx_d207_predictor_4x4_sse2;
+    vpx_d207_predictor_8x8 = vpx_d207_predictor_8x8_c;
+    if (flags & HAS_SSSE3) vpx_d207_predictor_8x8 = vpx_d207_predictor_8x8_ssse3;
+    vpx_d45_predictor_16x16 = vpx_d45_predictor_16x16_c;
+    if (flags & HAS_SSSE3) vpx_d45_predictor_16x16 = vpx_d45_predictor_16x16_ssse3;
+    vpx_d45_predictor_32x32 = vpx_d45_predictor_32x32_c;
+    if (flags & HAS_SSSE3) vpx_d45_predictor_32x32 = vpx_d45_predictor_32x32_ssse3;
+    vpx_d45_predictor_4x4 = vpx_d45_predictor_4x4_c;
+    if (flags & HAS_SSE2) vpx_d45_predictor_4x4 = vpx_d45_predictor_4x4_sse2;
+    vpx_d45_predictor_8x8 = vpx_d45_predictor_8x8_c;
+    if (flags & HAS_SSE2) vpx_d45_predictor_8x8 = vpx_d45_predictor_8x8_sse2;
+    vpx_d63_predictor_16x16 = vpx_d63_predictor_16x16_c;
+    if (flags & HAS_SSSE3) vpx_d63_predictor_16x16 = vpx_d63_predictor_16x16_ssse3;
+    vpx_d63_predictor_32x32 = vpx_d63_predictor_32x32_c;
+    if (flags & HAS_SSSE3) vpx_d63_predictor_32x32 = vpx_d63_predictor_32x32_ssse3;
+    vpx_d63_predictor_4x4 = vpx_d63_predictor_4x4_c;
+    if (flags & HAS_SSSE3) vpx_d63_predictor_4x4 = vpx_d63_predictor_4x4_ssse3;
+    vpx_d63_predictor_8x8 = vpx_d63_predictor_8x8_c;
+    if (flags & HAS_SSSE3) vpx_d63_predictor_8x8 = vpx_d63_predictor_8x8_ssse3;
+    vpx_dc_128_predictor_16x16 = vpx_dc_128_predictor_16x16_c;
+    if (flags & HAS_SSE2) vpx_dc_128_predictor_16x16 = vpx_dc_128_predictor_16x16_sse2;
+    vpx_dc_128_predictor_32x32 = vpx_dc_128_predictor_32x32_c;
+    if (flags & HAS_SSE2) vpx_dc_128_predictor_32x32 = vpx_dc_128_predictor_32x32_sse2;
+    vpx_dc_128_predictor_4x4 = vpx_dc_128_predictor_4x4_c;
+    if (flags & HAS_SSE2) vpx_dc_128_predictor_4x4 = vpx_dc_128_predictor_4x4_sse2;
+    vpx_dc_128_predictor_8x8 = vpx_dc_128_predictor_8x8_c;
+    if (flags & HAS_SSE2) vpx_dc_128_predictor_8x8 = vpx_dc_128_predictor_8x8_sse2;
+    vpx_dc_left_predictor_16x16 = vpx_dc_left_predictor_16x16_c;
+    if (flags & HAS_SSE2) vpx_dc_left_predictor_16x16 = vpx_dc_left_predictor_16x16_sse2;
+    vpx_dc_left_predictor_32x32 = vpx_dc_left_predictor_32x32_c;
+    if (flags & HAS_SSE2) vpx_dc_left_predictor_32x32 = vpx_dc_left_predictor_32x32_sse2;
+    vpx_dc_left_predictor_4x4 = vpx_dc_left_predictor_4x4_c;
+    if (flags & HAS_SSE2) vpx_dc_left_predictor_4x4 = vpx_dc_left_predictor_4x4_sse2;
+    vpx_dc_left_predictor_8x8 = vpx_dc_left_predictor_8x8_c;
+    if (flags & HAS_SSE2) vpx_dc_left_predictor_8x8 = vpx_dc_left_predictor_8x8_sse2;
+    vpx_dc_predictor_16x16 = vpx_dc_predictor_16x16_c;
+    if (flags & HAS_SSE2) vpx_dc_predictor_16x16 = vpx_dc_predictor_16x16_sse2;
+    vpx_dc_predictor_32x32 = vpx_dc_predictor_32x32_c;
+    if (flags & HAS_SSE2) vpx_dc_predictor_32x32 = vpx_dc_predictor_32x32_sse2;
+    vpx_dc_predictor_4x4 = vpx_dc_predictor_4x4_c;
+    if (flags & HAS_SSE2) vpx_dc_predictor_4x4 = vpx_dc_predictor_4x4_sse2;
+    vpx_dc_predictor_8x8 = vpx_dc_predictor_8x8_c;
+    if (flags & HAS_SSE2) vpx_dc_predictor_8x8 = vpx_dc_predictor_8x8_sse2;
+    vpx_dc_top_predictor_16x16 = vpx_dc_top_predictor_16x16_c;
+    if (flags & HAS_SSE2) vpx_dc_top_predictor_16x16 = vpx_dc_top_predictor_16x16_sse2;
+    vpx_dc_top_predictor_32x32 = vpx_dc_top_predictor_32x32_c;
+    if (flags & HAS_SSE2) vpx_dc_top_predictor_32x32 = vpx_dc_top_predictor_32x32_sse2;
+    vpx_dc_top_predictor_4x4 = vpx_dc_top_predictor_4x4_c;
+    if (flags & HAS_SSE2) vpx_dc_top_predictor_4x4 = vpx_dc_top_predictor_4x4_sse2;
+    vpx_dc_top_predictor_8x8 = vpx_dc_top_predictor_8x8_c;
+    if (flags & HAS_SSE2) vpx_dc_top_predictor_8x8 = vpx_dc_top_predictor_8x8_sse2;
+    vpx_h_predictor_16x16 = vpx_h_predictor_16x16_c;
+    if (flags & HAS_SSE2) vpx_h_predictor_16x16 = vpx_h_predictor_16x16_sse2;
+    vpx_h_predictor_32x32 = vpx_h_predictor_32x32_c;
+    if (flags & HAS_SSE2) vpx_h_predictor_32x32 = vpx_h_predictor_32x32_sse2;
+    vpx_h_predictor_4x4 = vpx_h_predictor_4x4_c;
+    if (flags & HAS_SSE2) vpx_h_predictor_4x4 = vpx_h_predictor_4x4_sse2;
+    vpx_h_predictor_8x8 = vpx_h_predictor_8x8_c;
+    if (flags & HAS_SSE2) vpx_h_predictor_8x8 = vpx_h_predictor_8x8_sse2;
+    vpx_idct16x16_10_add = vpx_idct16x16_10_add_c;
+    if (flags & HAS_SSE2) vpx_idct16x16_10_add = vpx_idct16x16_10_add_sse2;
+    vpx_idct16x16_1_add = vpx_idct16x16_1_add_c;
+    if (flags & HAS_SSE2) vpx_idct16x16_1_add = vpx_idct16x16_1_add_sse2;
+    vpx_idct16x16_256_add = vpx_idct16x16_256_add_c;
+    if (flags & HAS_SSE2) vpx_idct16x16_256_add = vpx_idct16x16_256_add_sse2;
+    vpx_idct32x32_1024_add = vpx_idct32x32_1024_add_c;
+    if (flags & HAS_SSE2) vpx_idct32x32_1024_add = vpx_idct32x32_1024_add_sse2;
+    vpx_idct32x32_135_add = vpx_idct32x32_135_add_c;
+    if (flags & HAS_SSE2) vpx_idct32x32_135_add = vpx_idct32x32_1024_add_sse2;
+    vpx_idct32x32_1_add = vpx_idct32x32_1_add_c;
+    if (flags & HAS_SSE2) vpx_idct32x32_1_add = vpx_idct32x32_1_add_sse2;
+    vpx_idct32x32_34_add = vpx_idct32x32_34_add_c;
+    if (flags & HAS_SSE2) vpx_idct32x32_34_add = vpx_idct32x32_34_add_sse2;
+    vpx_idct4x4_16_add = vpx_idct4x4_16_add_c;
+    if (flags & HAS_SSE2) vpx_idct4x4_16_add = vpx_idct4x4_16_add_sse2;
+    vpx_idct4x4_1_add = vpx_idct4x4_1_add_c;
+    if (flags & HAS_SSE2) vpx_idct4x4_1_add = vpx_idct4x4_1_add_sse2;
+    vpx_idct8x8_12_add = vpx_idct8x8_12_add_c;
+    if (flags & HAS_SSE2) vpx_idct8x8_12_add = vpx_idct8x8_12_add_sse2;
+    vpx_idct8x8_1_add = vpx_idct8x8_1_add_c;
+    if (flags & HAS_SSE2) vpx_idct8x8_1_add = vpx_idct8x8_1_add_sse2;
+    vpx_idct8x8_64_add = vpx_idct8x8_64_add_c;
+    if (flags & HAS_SSE2) vpx_idct8x8_64_add = vpx_idct8x8_64_add_sse2;
+    vpx_iwht4x4_16_add = vpx_iwht4x4_16_add_c;
+    if (flags & HAS_SSE2) vpx_iwht4x4_16_add = vpx_iwht4x4_16_add_sse2;
+    vpx_lpf_horizontal_4 = vpx_lpf_horizontal_4_c;
+    if (flags & HAS_SSE2) vpx_lpf_horizontal_4 = vpx_lpf_horizontal_4_sse2;
+    vpx_lpf_horizontal_4_dual = vpx_lpf_horizontal_4_dual_c;
+    if (flags & HAS_SSE2) vpx_lpf_horizontal_4_dual = vpx_lpf_horizontal_4_dual_sse2;
+    vpx_lpf_horizontal_8 = vpx_lpf_horizontal_8_c;
+    if (flags & HAS_SSE2) vpx_lpf_horizontal_8 = vpx_lpf_horizontal_8_sse2;
+    vpx_lpf_horizontal_8_dual = vpx_lpf_horizontal_8_dual_c;
+    if (flags & HAS_SSE2) vpx_lpf_horizontal_8_dual = vpx_lpf_horizontal_8_dual_sse2;
+    vpx_lpf_horizontal_edge_16 = vpx_lpf_horizontal_edge_16_c;
+    if (flags & HAS_SSE2) vpx_lpf_horizontal_edge_16 = vpx_lpf_horizontal_edge_16_sse2;
+    if (flags & HAS_AVX2) vpx_lpf_horizontal_edge_16 = vpx_lpf_horizontal_edge_16_avx2;
+    vpx_lpf_horizontal_edge_8 = vpx_lpf_horizontal_edge_8_c;
+    if (flags & HAS_SSE2) vpx_lpf_horizontal_edge_8 = vpx_lpf_horizontal_edge_8_sse2;
+    if (flags & HAS_AVX2) vpx_lpf_horizontal_edge_8 = vpx_lpf_horizontal_edge_8_avx2;
+    vpx_lpf_vertical_16 = vpx_lpf_vertical_16_c;
+    if (flags & HAS_SSE2) vpx_lpf_vertical_16 = vpx_lpf_vertical_16_sse2;
+    vpx_lpf_vertical_16_dual = vpx_lpf_vertical_16_dual_c;
+    if (flags & HAS_SSE2) vpx_lpf_vertical_16_dual = vpx_lpf_vertical_16_dual_sse2;
+    vpx_lpf_vertical_4 = vpx_lpf_vertical_4_c;
+    if (flags & HAS_SSE2) vpx_lpf_vertical_4 = vpx_lpf_vertical_4_sse2;
+    vpx_lpf_vertical_4_dual = vpx_lpf_vertical_4_dual_c;
+    if (flags & HAS_SSE2) vpx_lpf_vertical_4_dual = vpx_lpf_vertical_4_dual_sse2;
+    vpx_lpf_vertical_8 = vpx_lpf_vertical_8_c;
+    if (flags & HAS_SSE2) vpx_lpf_vertical_8 = vpx_lpf_vertical_8_sse2;
+    vpx_lpf_vertical_8_dual = vpx_lpf_vertical_8_dual_c;
+    if (flags & HAS_SSE2) vpx_lpf_vertical_8_dual = vpx_lpf_vertical_8_dual_sse2;
+    vpx_scaled_2d = vpx_scaled_2d_c;
+    if (flags & HAS_SSSE3) vpx_scaled_2d = vpx_scaled_2d_ssse3;
+    vpx_tm_predictor_16x16 = vpx_tm_predictor_16x16_c;
+    if (flags & HAS_SSE2) vpx_tm_predictor_16x16 = vpx_tm_predictor_16x16_sse2;
+    vpx_tm_predictor_32x32 = vpx_tm_predictor_32x32_c;
+    if (flags & HAS_SSE2) vpx_tm_predictor_32x32 = vpx_tm_predictor_32x32_sse2;
+    vpx_tm_predictor_4x4 = vpx_tm_predictor_4x4_c;
+    if (flags & HAS_SSE2) vpx_tm_predictor_4x4 = vpx_tm_predictor_4x4_sse2;
+    vpx_tm_predictor_8x8 = vpx_tm_predictor_8x8_c;
+    if (flags & HAS_SSE2) vpx_tm_predictor_8x8 = vpx_tm_predictor_8x8_sse2;
+    vpx_v_predictor_16x16 = vpx_v_predictor_16x16_c;
+    if (flags & HAS_SSE2) vpx_v_predictor_16x16 = vpx_v_predictor_16x16_sse2;
+    vpx_v_predictor_32x32 = vpx_v_predictor_32x32_c;
+    if (flags & HAS_SSE2) vpx_v_predictor_32x32 = vpx_v_predictor_32x32_sse2;
+    vpx_v_predictor_4x4 = vpx_v_predictor_4x4_c;
+    if (flags & HAS_SSE2) vpx_v_predictor_4x4 = vpx_v_predictor_4x4_sse2;
+    vpx_v_predictor_8x8 = vpx_v_predictor_8x8_c;
+    if (flags & HAS_SSE2) vpx_v_predictor_8x8 = vpx_v_predictor_8x8_sse2;
+}
+#endif
+
+#ifdef __cplusplus
+}  // extern "C"
+#endif
+
+#endif

thirdparty/libvpx/vp8_rtcd.h

@@ -0,0 +1,9 @@
+#include "vpx_config.h"
+
+#if defined(WEBM_X86ASM) && (ARCH_X86 || ARCH_X86_64)
+	#include "rtcd/vp8_rtcd_x86.h"
+#elif defined(WEBM_ARMASM) && ARCH_ARM
+	#include "rtcd/vp8_rtcd_arm.h"
+#else
+	#include "rtcd/vp8_rtcd_c.h"
+#endif

thirdparty/libvpx/vp9_rtcd.h

@@ -0,0 +1,9 @@
+#include "vpx_config.h"
+
+#if defined(WEBM_X86ASM) && (ARCH_X86 || ARCH_X86_64)
+	#include "rtcd/vp9_rtcd_x86.h"
+#elif defined(WEBM_ARMASM) && ARCH_ARM
+	#include "rtcd/vp9_rtcd_arm.h"
+#else
+	#include "rtcd/vp9_rtcd_c.h"
+#endif

thirdparty/libvpx/vpx_config.asm

@@ -0,0 +1,65 @@
+%ifdef X86_32
+	ARCH_X86 equ 1
+	ARCH_X86_64 equ 0
+%elifdef X86_64
+	ARCH_X86 equ 0
+	ARCH_X86_64 equ 1
+%endif
+
+HAVE_VPX_PORTS equ 1
+CONFIG_DEPENDENCY_TRACKING equ 0
+CONFIG_EXTERNAL_BUILD equ 0
+CONFIG_INSTALL_DOCS equ 0
+CONFIG_INSTALL_BINS equ 0
+CONFIG_INSTALL_LIBS equ 0
+CONFIG_INSTALL_SRCS equ 0
+CONFIG_USE_X86INC equ 1
+CONFIG_DEBUG equ 0
+CONFIG_GPROF equ 0
+CONFIG_GCOV equ 0
+CONFIG_RVCT equ 0
+CONFIG_PIC equ 1 ;TODO: autodetect
+CONFIG_CODEC_SRCS equ 0
+CONFIG_DEBUG_LIBS equ 0
+CONFIG_DEQUANT_TOKENS equ 0
+CONFIG_DC_RECON equ 0
+CONFIG_RUNTIME_CPU_DETECT equ 1
+CONFIG_POSTPROC equ 0
+CONFIG_VP9_POSTPROC equ 0
+CONFIG_MULTITHREAD equ 1
+CONFIG_INTERNAL_STATS equ 0
+CONFIG_VP8_ENCODER equ 0
+CONFIG_VP8_DECODER equ 1
+CONFIG_VP9_ENCODER equ 0
+CONFIG_VP9_DECODER equ 1
+CONFIG_VP8 equ 1
+CONFIG_VP9 equ 1
+CONFIG_ENCODERS equ 0
+CONFIG_DECODERS equ 1
+CONFIG_STATIC_MSVCRT equ 0
+CONFIG_SPATIAL_RESAMPLING equ 0
+CONFIG_REALTIME_ONLY equ 0
+CONFIG_ONTHEFLY_BITPACKING equ 0
+CONFIG_ERROR_CONCEALMENT equ 0
+CONFIG_SHARED equ 0
+CONFIG_STATIC equ 0
+CONFIG_SMALL equ 0
+CONFIG_POSTPROC_VISUALIZER equ 0
+CONFIG_OS_SUPPORT equ 1
+CONFIG_UNIT_TESTS equ 0
+CONFIG_WEBM_IO equ 0
+CONFIG_LIBYUV equ 0
+CONFIG_DECODE_PERF_TESTS equ 0
+CONFIG_ENCODE_PERF_TESTS equ 0
+CONFIG_MULTI_RES_ENCODING equ 0
+CONFIG_TEMPORAL_DENOISING equ 1
+CONFIG_VP9_TEMPORAL_DENOISING equ 0
+CONFIG_COEFFICIENT_RANGE_CHECKING equ 0
+CONFIG_VP9_HIGHBITDEPTH equ 0
+CONFIG_BETTER_HW_COMPATIBILITY equ 0
+CONFIG_EXPERIMENTAL equ 0
+CONFIG_SIZE_LIMIT equ 0
+CONFIG_SPATIAL_SVC equ 0
+CONFIG_FP_MB_STATS equ 0
+CONFIG_EMULATE_HARDWARE equ 0
+CONFIG_MISC_FIXES equ 0

thirdparty/libvpx/vpx_config.h

@@ -0,0 +1,124 @@
+/* Copyright (c) 2011 The WebM project authors. All Rights Reserved. */
+/*  */
+/* Use of this source code is governed by a BSD-style license */
+/* that can be found in the LICENSE file in the root of the source */
+/* tree. An additional intellectual property rights grant can be found */
+/* in the file PATENTS.  All contributing project authors may */
+/* be found in the AUTHORS file in the root of the source tree. */
+/* This file automatically generated by configure. Do not edit! */
+#ifndef VPX_CONFIG_H
+#define VPX_CONFIG_H
+#define RESTRICT
+#define INLINE inline
+
+#define HAVE_MIPS32 0
+#define HAVE_MEDIA 0
+
+#if defined(__i386) || defined(__i386__) || defined(_M_IX86)
+	#define ARCH_X86 1
+	#define ARCH_X86_64 0
+
+	#define ARCH_ARM 0
+	#define HAVE_NEON 0
+	#define HAVE_NEON_ASM 0
+
+	#define HAVE_MMX 1
+	#define HAVE_SSE2 1
+	#define HAVE_SSSE3 1
+	#define HAVE_AVX2 1
+#elif defined(__x86_64) || defined(__x86_64__) || defined(__amd64) || defined(_M_X64)
+	#define ARCH_X86 0
+	#define ARCH_X86_64 1
+
+	#define ARCH_ARM 0
+	#define HAVE_NEON 0
+	#define HAVE_NEON_ASM 0
+
+	#define HAVE_MMX 1
+	#define HAVE_SSE2 1
+	#define HAVE_SSSE3 1
+	#define HAVE_AVX2 1
+#elif defined(__arm__) || defined(__TARGET_ARCH_ARM) || defined(_M_ARM) || defined(__aarch64__)
+	#define ARCH_X86 0
+	#define ARCH_X86_64 0
+
+	#define ARCH_ARM 1
+	#define HAVE_NEON 1
+	#define HAVE_NEON_ASM 1
+#else
+	#define ARCH_X86 0
+	#define ARCH_X86_64 0
+
+	#define ARCH_ARM 0
+	#define HAVE_NEON 0
+	#define HAVE_NEON_ASM 0
+#endif
+
+#define CONFIG_BIG_ENDIAN 0 //TODO: Autodetect
+
+#ifdef _WIN32
+	#define HAVE_PTHREAD_H 0
+	#define HAVE_UNISTD_H 0
+#else
+	#define HAVE_PTHREAD_H 1
+	#define HAVE_UNISTD_H 1
+#endif
+
+/**/
+
+#define HAVE_VPX_PORTS 1
+#define CONFIG_DEPENDENCY_TRACKING 0
+#define CONFIG_EXTERNAL_BUILD 0
+#define CONFIG_INSTALL_DOCS 0
+#define CONFIG_INSTALL_BINS 0
+#define CONFIG_INSTALL_LIBS 0
+#define CONFIG_INSTALL_SRCS 0
+#define CONFIG_DEBUG 0
+#define CONFIG_GPROF 0
+#define CONFIG_GCOV 0
+#define CONFIG_RVCT 0
+#define CONFIG_CODEC_SRCS 0
+#define CONFIG_DEBUG_LIBS 0
+#define CONFIG_DEQUANT_TOKENS 0
+#define CONFIG_DC_RECON 0
+#define CONFIG_RUNTIME_CPU_DETECT 1
+#define CONFIG_POSTPROC 0
+#define CONFIG_VP9_POSTPROC 0
+#define CONFIG_MULTITHREAD 1
+#define CONFIG_INTERNAL_STATS 0
+#define CONFIG_VP8_ENCODER 0
+#define CONFIG_VP8_DECODER 1
+#define CONFIG_VP9_ENCODER 0
+#define CONFIG_VP9_DECODER 1
+#define CONFIG_VP8 1
+#define CONFIG_VP9 1
+#define CONFIG_ENCODERS 0
+#define CONFIG_DECODERS 1
+#define CONFIG_STATIC_MSVCRT 0
+#define CONFIG_SPATIAL_RESAMPLING 0
+#define CONFIG_REALTIME_ONLY 0
+#define CONFIG_ONTHEFLY_BITPACKING 0
+#define CONFIG_ERROR_CONCEALMENT 0
+#define CONFIG_SHARED 0
+#define CONFIG_STATIC 0
+#define CONFIG_SMALL 0
+#define CONFIG_POSTPROC_VISUALIZER 0
+#define CONFIG_OS_SUPPORT 1
+#define CONFIG_UNIT_TESTS 0
+#define CONFIG_WEBM_IO 0
+#define CONFIG_LIBYUV 0
+#define CONFIG_DECODE_PERF_TESTS 0
+#define CONFIG_ENCODE_PERF_TESTS 0
+#define CONFIG_MULTI_RES_ENCODING 0
+#define CONFIG_TEMPORAL_DENOISING 0
+#define CONFIG_VP9_TEMPORAL_DENOISING 0
+#define CONFIG_COEFFICIENT_RANGE_CHECKING 0
+#define CONFIG_VP9_HIGHBITDEPTH 0
+#define CONFIG_BETTER_HW_COMPATIBILITY 0
+#define CONFIG_EXPERIMENTAL 0
+#define CONFIG_SIZE_LIMIT 0
+#define CONFIG_SPATIAL_SVC 0
+#define CONFIG_FP_MB_STATS 0
+#define CONFIG_EMULATE_HARDWARE 0
+#define CONFIG_MISC_FIXES 0
+#endif /* VPX_CONFIG_H */

thirdparty/libvpx/vpx_dsp/arm/armasm_ms/intrapred_neon_asm.asm

@@ -1,3 +1,5 @@
+; This file was created from a .asm file
+;  using the ads2armasm_ms.pl script.
 ;
 ;  Copyright (c) 2014 The WebM project authors. All Rights Reserved.
 ;
@@ -20,11 +22,10 @@
     EXPORT  |vpx_tm_predictor_8x8_neon|
     EXPORT  |vpx_tm_predictor_16x16_neon|
     EXPORT  |vpx_tm_predictor_32x32_neon|
-    ARM
+    
-    REQUIRE8
+    
-    PRESERVE8
 
-    AREA ||.text||, CODE, READONLY, ALIGN=2
+    AREA |.text|, CODE, READONLY, ALIGN=2
 
 ;void vpx_v_predictor_4x4_neon(uint8_t *dst, ptrdiff_t y_stride,
 ;                              const uint8_t *above,
@@ -42,6 +43,7 @@
     vst1.32             {d0[0]}, [r0], r1
     bx                  lr
     ENDP                ; |vpx_v_predictor_4x4_neon|
+    ALIGN 4
 
 ;void vpx_v_predictor_8x8_neon(uint8_t *dst, ptrdiff_t y_stride,
 ;                              const uint8_t *above,
@@ -63,6 +65,7 @@
     vst1.8              {d0}, [r0], r1
     bx                  lr
     ENDP                ; |vpx_v_predictor_8x8_neon|
+    ALIGN 4
 
 ;void vpx_v_predictor_16x16_neon(uint8_t *dst, ptrdiff_t y_stride,
 ;                                const uint8_t *above,
@@ -92,6 +95,7 @@
     vst1.8              {q0}, [r0], r1
     bx                  lr
     ENDP                ; |vpx_v_predictor_16x16_neon|
+    ALIGN 4
 
 ;void vpx_v_predictor_32x32_neon(uint8_t *dst, ptrdiff_t y_stride,
 ;                                const uint8_t *above,
@@ -125,6 +129,7 @@ loop_v
     bgt                 loop_v
     bx                  lr
     ENDP                ; |vpx_v_predictor_32x32_neon|
+    ALIGN 4
 
 ;void vpx_h_predictor_4x4_neon(uint8_t *dst, ptrdiff_t y_stride,
 ;                              const uint8_t *above,
@@ -146,6 +151,7 @@ loop_v
     vst1.32             {d0[0]}, [r0], r1
     bx                  lr
     ENDP                ; |vpx_h_predictor_4x4_neon|
+    ALIGN 4
 
 ;void vpx_h_predictor_8x8_neon(uint8_t *dst, ptrdiff_t y_stride,
 ;                              const uint8_t *above,
@@ -175,6 +181,7 @@ loop_v
     vst1.64             {d0}, [r0], r1
     bx                  lr
     ENDP                ; |vpx_h_predictor_8x8_neon|
+    ALIGN 4
 
 ;void vpx_h_predictor_16x16_neon(uint8_t *dst, ptrdiff_t y_stride,
 ;                                const uint8_t *above,
@@ -220,6 +227,7 @@ loop_v
     vst1.8              {q0}, [r0], r1
     bx                  lr
     ENDP                ; |vpx_h_predictor_16x16_neon|
+    ALIGN 4
 
 ;void vpx_h_predictor_32x32_neon(uint8_t *dst, ptrdiff_t y_stride,
 ;                                const uint8_t *above,
@@ -286,6 +294,7 @@ loop_h
     bgt                 loop_h
     bx                  lr
     ENDP                ; |vpx_h_predictor_32x32_neon|
+    ALIGN 4
 
 ;void vpx_tm_predictor_4x4_neon (uint8_t *dst, ptrdiff_t y_stride,
 ;                                const uint8_t *above,
@@ -332,6 +341,7 @@ loop_h
     vst1.32             {d1[0]}, [r0], r1
     bx                  lr
     ENDP                ; |vpx_tm_predictor_4x4_neon|
+    ALIGN 4
 
 ;void vpx_tm_predictor_8x8_neon (uint8_t *dst, ptrdiff_t y_stride,
 ;                                const uint8_t *above,
@@ -404,6 +414,7 @@ loop_h
 
     bx                  lr
     ENDP                ; |vpx_tm_predictor_8x8_neon|
+    ALIGN 4
 
 ;void vpx_tm_predictor_16x16_neon (uint8_t *dst, ptrdiff_t y_stride,
 ;                                const uint8_t *above,
@@ -497,6 +508,7 @@ loop_16x16_neon
 
     bx                  lr
     ENDP                ; |vpx_tm_predictor_16x16_neon|
+    ALIGN 4
 
 ;void vpx_tm_predictor_32x32_neon (uint8_t *dst, ptrdiff_t y_stride,
 ;                                  const uint8_t *above,
@@ -626,5 +638,6 @@ loop_32x32_neon
 
     bx                  lr
     ENDP                ; |vpx_tm_predictor_32x32_neon|
+    ALIGN 4
 
     END

thirdparty/libvpx/vpx_dsp/arm/armasm_ms/loopfilter_mb_neon.asm

@@ -1,3 +1,5 @@
+; This file was created from a .asm file
+;  using the ads2armasm_ms.pl script.
 ;
 ;  Copyright (c) 2013 The WebM project authors. All Rights Reserved.
 ;
@@ -11,9 +13,8 @@
     EXPORT  |vpx_lpf_horizontal_edge_8_neon|
     EXPORT  |vpx_lpf_horizontal_edge_16_neon|
     EXPORT  |vpx_lpf_vertical_16_neon|
-    ARM
 
-    AREA ||.text||, CODE, READONLY, ALIGN=2
+    AREA |.text|, CODE, READONLY, ALIGN=2
 
 ; void mb_lpf_horizontal_edge(uint8_t *s, int p,
 ;                             const uint8_t *blimit,
@@ -117,6 +118,7 @@ h_next
     pop         {r4-r8, pc}
 
     ENDP        ; |mb_lpf_horizontal_edge|
+    ALIGN 4
 
 ; void vpx_lpf_horizontal_edge_8_neon(uint8_t *s, int pitch,
 ;                                     const uint8_t *blimit,
@@ -131,6 +133,7 @@ h_next
     mov r12, #1
     b mb_lpf_horizontal_edge
     ENDP        ; |vpx_lpf_horizontal_edge_8_neon|
+    ALIGN 4
 
 ; void vpx_lpf_horizontal_edge_16_neon(uint8_t *s, int pitch,
 ;                                      const uint8_t *blimit,
@@ -145,6 +148,7 @@ h_next
     mov r12, #2
     b mb_lpf_horizontal_edge
     ENDP        ; |vpx_lpf_horizontal_edge_16_neon|
+    ALIGN 4
 
 ; void vpx_lpf_vertical_16_neon(uint8_t *s, int p,
 ;                               const uint8_t *blimit,
@@ -309,6 +313,7 @@ v_end
     pop         {r4-r8, pc}
 
     ENDP        ; |vpx_lpf_vertical_16_neon|
+    ALIGN 4
 
 ; void vpx_wide_mbfilter_neon();
 ; This is a helper function for the loopfilters. The invidual functions do the
@@ -631,5 +636,6 @@ v_end
 
     bx          lr
     ENDP        ; |vpx_wide_mbfilter_neon|
+    ALIGN 4
 
     END

thirdparty/libvpx/vpx_dsp/arm/armasm_ms/save_reg_neon.asm

@@ -1,3 +1,5 @@
+; This file was created from a .asm file
+;  using the ads2armasm_ms.pl script.
 ;
 ;  Copyright (c) 2010 The WebM project authors. All Rights Reserved.
 ;
@@ -12,11 +14,10 @@
     EXPORT  |vpx_push_neon|
     EXPORT  |vpx_pop_neon|
 
-    ARM
+    
-    REQUIRE8
+    
-    PRESERVE8
 
-    AREA ||.text||, CODE, READONLY, ALIGN=2
+    AREA |.text|, CODE, READONLY, ALIGN=2
 
 |vpx_push_neon| PROC
     vst1.i64            {d8, d9, d10, d11}, [r0]!
@@ -24,6 +25,7 @@
     bx              lr
 
     ENDP
+    ALIGN 4
 
 |vpx_pop_neon| PROC
     vld1.i64            {d8, d9, d10, d11}, [r0]!
@@ -31,6 +33,7 @@
     bx              lr
 
     ENDP
+    ALIGN 4
 
     END
 

thirdparty/libvpx/vpx_dsp/arm/gas/intrapred_neon_asm.s

@@ -0,0 +1,658 @@
+@ This file was created from a .asm file
+@  using the ads2gas.pl script.
+	.equ DO1STROUNDING, 0
+@
+@  Copyright (c) 2014 The WebM project authors. All Rights Reserved.
+@
+@  Use of this source code is governed by a BSD-style license
+@  that can be found in the LICENSE file in the root of the source
+@  tree. An additional intellectual property rights grant can be found
+@  in the file PATENTS.  All contributing project authors may
+@  be found in the AUTHORS file in the root of the source tree.
+@
+
+    .global vpx_v_predictor_4x4_neon 
+	.type vpx_v_predictor_4x4_neon, function
+    .global vpx_v_predictor_8x8_neon 
+	.type vpx_v_predictor_8x8_neon, function
+    .global vpx_v_predictor_16x16_neon 
+	.type vpx_v_predictor_16x16_neon, function
+    .global vpx_v_predictor_32x32_neon 
+	.type vpx_v_predictor_32x32_neon, function
+    .global vpx_h_predictor_4x4_neon 
+	.type vpx_h_predictor_4x4_neon, function
+    .global vpx_h_predictor_8x8_neon 
+	.type vpx_h_predictor_8x8_neon, function
+    .global vpx_h_predictor_16x16_neon 
+	.type vpx_h_predictor_16x16_neon, function
+    .global vpx_h_predictor_32x32_neon 
+	.type vpx_h_predictor_32x32_neon, function
+    .global vpx_tm_predictor_4x4_neon 
+	.type vpx_tm_predictor_4x4_neon, function
+    .global vpx_tm_predictor_8x8_neon 
+	.type vpx_tm_predictor_8x8_neon, function
+    .global vpx_tm_predictor_16x16_neon 
+	.type vpx_tm_predictor_16x16_neon, function
+    .global vpx_tm_predictor_32x32_neon 
+	.type vpx_tm_predictor_32x32_neon, function
+   .arm
+   .eabi_attribute 24, 1 @Tag_ABI_align_needed
+   .eabi_attribute 25, 1 @Tag_ABI_align_preserved
+
+.text
+.p2align 2
+
+@void vpx_v_predictor_4x4_neon(uint8_t *dst, ptrdiff_t y_stride,
+@                              const uint8_t *above,
+@                              const uint8_t *left)
+@ r0  uint8_t *dst
+@ r1  ptrdiff_t y_stride
+@ r2  const uint8_t *above
+@ r3  const uint8_t *left
+
+_vpx_v_predictor_4x4_neon:
+	vpx_v_predictor_4x4_neon: @ PROC
+    vld1.32             {d0[0]}, [r2]
+    vst1.32             {d0[0]}, [r0], r1
+    vst1.32             {d0[0]}, [r0], r1
+    vst1.32             {d0[0]}, [r0], r1
+    vst1.32             {d0[0]}, [r0], r1
+    bx                  lr
+	.size vpx_v_predictor_4x4_neon, .-vpx_v_predictor_4x4_neon    @ ENDP                @ |vpx_v_predictor_4x4_neon|
+
+@void vpx_v_predictor_8x8_neon(uint8_t *dst, ptrdiff_t y_stride,
+@                              const uint8_t *above,
+@                              const uint8_t *left)
+@ r0  uint8_t *dst
+@ r1  ptrdiff_t y_stride
+@ r2  const uint8_t *above
+@ r3  const uint8_t *left
+
+_vpx_v_predictor_8x8_neon:
+	vpx_v_predictor_8x8_neon: @ PROC
+    vld1.8              {d0}, [r2]
+    vst1.8              {d0}, [r0], r1
+    vst1.8              {d0}, [r0], r1
+    vst1.8              {d0}, [r0], r1
+    vst1.8              {d0}, [r0], r1
+    vst1.8              {d0}, [r0], r1
+    vst1.8              {d0}, [r0], r1
+    vst1.8              {d0}, [r0], r1
+    vst1.8              {d0}, [r0], r1
+    bx                  lr
+	.size vpx_v_predictor_8x8_neon, .-vpx_v_predictor_8x8_neon    @ ENDP                @ |vpx_v_predictor_8x8_neon|
+
+@void vpx_v_predictor_16x16_neon(uint8_t *dst, ptrdiff_t y_stride,
+@                                const uint8_t *above,
+@                                const uint8_t *left)
+@ r0  uint8_t *dst
+@ r1  ptrdiff_t y_stride
+@ r2  const uint8_t *above
+@ r3  const uint8_t *left
+
+_vpx_v_predictor_16x16_neon:
+	vpx_v_predictor_16x16_neon: @ PROC
+    vld1.8              {q0}, [r2]
+    vst1.8              {q0}, [r0], r1
+    vst1.8              {q0}, [r0], r1
+    vst1.8              {q0}, [r0], r1
+    vst1.8              {q0}, [r0], r1
+    vst1.8              {q0}, [r0], r1
+    vst1.8              {q0}, [r0], r1
+    vst1.8              {q0}, [r0], r1
+    vst1.8              {q0}, [r0], r1
+    vst1.8              {q0}, [r0], r1
+    vst1.8              {q0}, [r0], r1
+    vst1.8              {q0}, [r0], r1
+    vst1.8              {q0}, [r0], r1
+    vst1.8              {q0}, [r0], r1
+    vst1.8              {q0}, [r0], r1
+    vst1.8              {q0}, [r0], r1
+    vst1.8              {q0}, [r0], r1
+    bx                  lr
+	.size vpx_v_predictor_16x16_neon, .-vpx_v_predictor_16x16_neon    @ ENDP                @ |vpx_v_predictor_16x16_neon|
+
+@void vpx_v_predictor_32x32_neon(uint8_t *dst, ptrdiff_t y_stride,
+@                                const uint8_t *above,
+@                                const uint8_t *left)
+@ r0  uint8_t *dst
+@ r1  ptrdiff_t y_stride
+@ r2  const uint8_t *above
+@ r3  const uint8_t *left
+
+_vpx_v_predictor_32x32_neon:
+	vpx_v_predictor_32x32_neon: @ PROC
+    vld1.8              {q0, q1}, [r2]
+    mov                 r2, #2
+loop_v:
+    vst1.8              {q0, q1}, [r0], r1
+    vst1.8              {q0, q1}, [r0], r1
+    vst1.8              {q0, q1}, [r0], r1
+    vst1.8              {q0, q1}, [r0], r1
+    vst1.8              {q0, q1}, [r0], r1
+    vst1.8              {q0, q1}, [r0], r1
+    vst1.8              {q0, q1}, [r0], r1
+    vst1.8              {q0, q1}, [r0], r1
+    vst1.8              {q0, q1}, [r0], r1
+    vst1.8              {q0, q1}, [r0], r1
+    vst1.8              {q0, q1}, [r0], r1
+    vst1.8              {q0, q1}, [r0], r1
+    vst1.8              {q0, q1}, [r0], r1
+    vst1.8              {q0, q1}, [r0], r1
+    vst1.8              {q0, q1}, [r0], r1
+    vst1.8              {q0, q1}, [r0], r1
+    subs                r2, r2, #1
+    bgt                 loop_v
+    bx                  lr
+	.size vpx_v_predictor_32x32_neon, .-vpx_v_predictor_32x32_neon    @ ENDP                @ |vpx_v_predictor_32x32_neon|
+
+@void vpx_h_predictor_4x4_neon(uint8_t *dst, ptrdiff_t y_stride,
+@                              const uint8_t *above,
+@                              const uint8_t *left)
+@ r0  uint8_t *dst
+@ r1  ptrdiff_t y_stride
+@ r2  const uint8_t *above
+@ r3  const uint8_t *left
+
+_vpx_h_predictor_4x4_neon:
+	vpx_h_predictor_4x4_neon: @ PROC
+    vld1.32             {d1[0]}, [r3]
+    vdup.8              d0, d1[0]
+    vst1.32             {d0[0]}, [r0], r1
+    vdup.8              d0, d1[1]
+    vst1.32             {d0[0]}, [r0], r1
+    vdup.8              d0, d1[2]
+    vst1.32             {d0[0]}, [r0], r1
+    vdup.8              d0, d1[3]
+    vst1.32             {d0[0]}, [r0], r1
+    bx                  lr
+	.size vpx_h_predictor_4x4_neon, .-vpx_h_predictor_4x4_neon    @ ENDP                @ |vpx_h_predictor_4x4_neon|
+
+@void vpx_h_predictor_8x8_neon(uint8_t *dst, ptrdiff_t y_stride,
+@                              const uint8_t *above,
+@                              const uint8_t *left)
+@ r0  uint8_t *dst
+@ r1  ptrdiff_t y_stride
+@ r2  const uint8_t *above
+@ r3  const uint8_t *left
+
+_vpx_h_predictor_8x8_neon:
+	vpx_h_predictor_8x8_neon: @ PROC
+    vld1.64             {d1}, [r3]
+    vdup.8              d0, d1[0]
+    vst1.64             {d0}, [r0], r1
+    vdup.8              d0, d1[1]
+    vst1.64             {d0}, [r0], r1
+    vdup.8              d0, d1[2]
+    vst1.64             {d0}, [r0], r1
+    vdup.8              d0, d1[3]
+    vst1.64             {d0}, [r0], r1
+    vdup.8              d0, d1[4]
+    vst1.64             {d0}, [r0], r1
+    vdup.8              d0, d1[5]
+    vst1.64             {d0}, [r0], r1
+    vdup.8              d0, d1[6]
+    vst1.64             {d0}, [r0], r1
+    vdup.8              d0, d1[7]
+    vst1.64             {d0}, [r0], r1
+    bx                  lr
+	.size vpx_h_predictor_8x8_neon, .-vpx_h_predictor_8x8_neon    @ ENDP                @ |vpx_h_predictor_8x8_neon|
+
+@void vpx_h_predictor_16x16_neon(uint8_t *dst, ptrdiff_t y_stride,
+@                                const uint8_t *above,
+@                                const uint8_t *left)
+@ r0  uint8_t *dst
+@ r1  ptrdiff_t y_stride
+@ r2  const uint8_t *above
+@ r3  const uint8_t *left
+
+_vpx_h_predictor_16x16_neon:
+	vpx_h_predictor_16x16_neon: @ PROC
+    vld1.8              {q1}, [r3]
+    vdup.8              q0, d2[0]
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d2[1]
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d2[2]
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d2[3]
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d2[4]
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d2[5]
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d2[6]
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d2[7]
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d3[0]
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d3[1]
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d3[2]
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d3[3]
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d3[4]
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d3[5]
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d3[6]
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d3[7]
+    vst1.8              {q0}, [r0], r1
+    bx                  lr
+	.size vpx_h_predictor_16x16_neon, .-vpx_h_predictor_16x16_neon    @ ENDP                @ |vpx_h_predictor_16x16_neon|
+
+@void vpx_h_predictor_32x32_neon(uint8_t *dst, ptrdiff_t y_stride,
+@                                const uint8_t *above,
+@                                const uint8_t *left)
+@ r0  uint8_t *dst
+@ r1  ptrdiff_t y_stride
+@ r2  const uint8_t *above
+@ r3  const uint8_t *left
+
+_vpx_h_predictor_32x32_neon:
+	vpx_h_predictor_32x32_neon: @ PROC
+    sub                 r1, r1, #16
+    mov                 r2, #2
+loop_h:
+    vld1.8              {q1}, [r3]!
+    vdup.8              q0, d2[0]
+    vst1.8              {q0}, [r0]!
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d2[1]
+    vst1.8              {q0}, [r0]!
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d2[2]
+    vst1.8              {q0}, [r0]!
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d2[3]
+    vst1.8              {q0}, [r0]!
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d2[4]
+    vst1.8              {q0}, [r0]!
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d2[5]
+    vst1.8              {q0}, [r0]!
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d2[6]
+    vst1.8              {q0}, [r0]!
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d2[7]
+    vst1.8              {q0}, [r0]!
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d3[0]
+    vst1.8              {q0}, [r0]!
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d3[1]
+    vst1.8              {q0}, [r0]!
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d3[2]
+    vst1.8              {q0}, [r0]!
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d3[3]
+    vst1.8              {q0}, [r0]!
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d3[4]
+    vst1.8              {q0}, [r0]!
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d3[5]
+    vst1.8              {q0}, [r0]!
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d3[6]
+    vst1.8              {q0}, [r0]!
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d3[7]
+    vst1.8              {q0}, [r0]!
+    vst1.8              {q0}, [r0], r1
+    subs                r2, r2, #1
+    bgt                 loop_h
+    bx                  lr
+	.size vpx_h_predictor_32x32_neon, .-vpx_h_predictor_32x32_neon    @ ENDP                @ |vpx_h_predictor_32x32_neon|
+
+@void vpx_tm_predictor_4x4_neon (uint8_t *dst, ptrdiff_t y_stride,
+@                                const uint8_t *above,
+@                                const uint8_t *left)
+@ r0  uint8_t *dst
+@ r1  ptrdiff_t y_stride
+@ r2  const uint8_t *above
+@ r3  const uint8_t *left
+
+_vpx_tm_predictor_4x4_neon:
+	vpx_tm_predictor_4x4_neon: @ PROC
+    @ Load ytop_left = above[-1];
+    sub                 r12, r2, #1
+    vld1.u8             {d0[]}, [r12]
+
+    @ Load above 4 pixels
+    vld1.32             {d2[0]}, [r2]
+
+    @ Compute above - ytop_left
+    vsubl.u8            q3, d2, d0
+
+    @ Load left row by row and compute left + (above - ytop_left)
+    @ 1st row and 2nd row
+    vld1.u8             {d2[]}, [r3]!
+    vld1.u8             {d4[]}, [r3]!
+    vmovl.u8            q1, d2
+    vmovl.u8            q2, d4
+    vadd.s16            q1, q1, q3
+    vadd.s16            q2, q2, q3
+    vqmovun.s16         d0, q1
+    vqmovun.s16         d1, q2
+    vst1.32             {d0[0]}, [r0], r1
+    vst1.32             {d1[0]}, [r0], r1
+
+    @ 3rd row and 4th row
+    vld1.u8             {d2[]}, [r3]!
+    vld1.u8             {d4[]}, [r3]
+    vmovl.u8            q1, d2
+    vmovl.u8            q2, d4
+    vadd.s16            q1, q1, q3
+    vadd.s16            q2, q2, q3
+    vqmovun.s16         d0, q1
+    vqmovun.s16         d1, q2
+    vst1.32             {d0[0]}, [r0], r1
+    vst1.32             {d1[0]}, [r0], r1
+    bx                  lr
+	.size vpx_tm_predictor_4x4_neon, .-vpx_tm_predictor_4x4_neon    @ ENDP                @ |vpx_tm_predictor_4x4_neon|
+
+@void vpx_tm_predictor_8x8_neon (uint8_t *dst, ptrdiff_t y_stride,
+@                                const uint8_t *above,
+@                                const uint8_t *left)
+@ r0  uint8_t *dst
+@ r1  ptrdiff_t y_stride
+@ r2  const uint8_t *above
+@ r3  const uint8_t *left
+
+_vpx_tm_predictor_8x8_neon:
+	vpx_tm_predictor_8x8_neon: @ PROC
+    @ Load ytop_left = above[-1];
+    sub                 r12, r2, #1
+    vld1.8              {d0[]}, [r12]
+
+    @ preload 8 left
+    vld1.8              {d30}, [r3]
+
+    @ Load above 8 pixels
+    vld1.64             {d2}, [r2]
+
+    vmovl.u8            q10, d30
+
+    @ Compute above - ytop_left
+    vsubl.u8            q3, d2, d0
+
+    @ Load left row by row and compute left + (above - ytop_left)
+    @ 1st row and 2nd row
+    vdup.16             q0, d20[0]
+    vdup.16             q1, d20[1]
+    vadd.s16            q0, q3, q0
+    vadd.s16            q1, q3, q1
+
+    @ 3rd row and 4th row
+    vdup.16             q8, d20[2]
+    vdup.16             q9, d20[3]
+    vadd.s16            q8, q3, q8
+    vadd.s16            q9, q3, q9
+
+    vqmovun.s16         d0, q0
+    vqmovun.s16         d1, q1
+    vqmovun.s16         d2, q8
+    vqmovun.s16         d3, q9
+
+    vst1.64             {d0}, [r0], r1
+    vst1.64             {d1}, [r0], r1
+    vst1.64             {d2}, [r0], r1
+    vst1.64             {d3}, [r0], r1
+
+    @ 5th row and 6th row
+    vdup.16             q0, d21[0]
+    vdup.16             q1, d21[1]
+    vadd.s16            q0, q3, q0
+    vadd.s16            q1, q3, q1
+
+    @ 7th row and 8th row
+    vdup.16             q8, d21[2]
+    vdup.16             q9, d21[3]
+    vadd.s16            q8, q3, q8
+    vadd.s16            q9, q3, q9
+
+    vqmovun.s16         d0, q0
+    vqmovun.s16         d1, q1
+    vqmovun.s16         d2, q8
+    vqmovun.s16         d3, q9
+
+    vst1.64             {d0}, [r0], r1
+    vst1.64             {d1}, [r0], r1
+    vst1.64             {d2}, [r0], r1
+    vst1.64             {d3}, [r0], r1
+
+    bx                  lr
+	.size vpx_tm_predictor_8x8_neon, .-vpx_tm_predictor_8x8_neon    @ ENDP                @ |vpx_tm_predictor_8x8_neon|
+
+@void vpx_tm_predictor_16x16_neon (uint8_t *dst, ptrdiff_t y_stride,
+@                                const uint8_t *above,
+@                                const uint8_t *left)
+@ r0  uint8_t *dst
+@ r1  ptrdiff_t y_stride
+@ r2  const uint8_t *above
+@ r3  const uint8_t *left
+
+_vpx_tm_predictor_16x16_neon:
+	vpx_tm_predictor_16x16_neon: @ PROC
+    @ Load ytop_left = above[-1];
+    sub                 r12, r2, #1
+    vld1.8              {d0[]}, [r12]
+
+    @ Load above 8 pixels
+    vld1.8              {q1}, [r2]
+
+    @ preload 8 left into r12
+    vld1.8              {d18}, [r3]!
+
+    @ Compute above - ytop_left
+    vsubl.u8            q2, d2, d0
+    vsubl.u8            q3, d3, d0
+
+    vmovl.u8            q10, d18
+
+    @ Load left row by row and compute left + (above - ytop_left)
+    @ Process 8 rows in each single loop and loop 2 times to process 16 rows.
+    mov                 r2, #2
+
+loop_16x16_neon:
+    @ Process two rows.
+    vdup.16             q0, d20[0]
+    vdup.16             q8, d20[1]
+    vadd.s16            q1, q0, q2
+    vadd.s16            q0, q0, q3
+    vadd.s16            q11, q8, q2
+    vadd.s16            q8, q8, q3
+    vqmovun.s16         d2, q1
+    vqmovun.s16         d3, q0
+    vqmovun.s16         d22, q11
+    vqmovun.s16         d23, q8
+    vdup.16             q0, d20[2]                  @ proload next 2 rows data
+    vdup.16             q8, d20[3]
+    vst1.64             {d2,d3}, [r0], r1
+    vst1.64             {d22,d23}, [r0], r1
+
+    @ Process two rows.
+    vadd.s16            q1, q0, q2
+    vadd.s16            q0, q0, q3
+    vadd.s16            q11, q8, q2
+    vadd.s16            q8, q8, q3
+    vqmovun.s16         d2, q1
+    vqmovun.s16         d3, q0
+    vqmovun.s16         d22, q11
+    vqmovun.s16         d23, q8
+    vdup.16             q0, d21[0]                  @ proload next 2 rows data
+    vdup.16             q8, d21[1]
+    vst1.64             {d2,d3}, [r0], r1
+    vst1.64             {d22,d23}, [r0], r1
+
+    vadd.s16            q1, q0, q2
+    vadd.s16            q0, q0, q3
+    vadd.s16            q11, q8, q2
+    vadd.s16            q8, q8, q3
+    vqmovun.s16         d2, q1
+    vqmovun.s16         d3, q0
+    vqmovun.s16         d22, q11
+    vqmovun.s16         d23, q8
+    vdup.16             q0, d21[2]                  @ proload next 2 rows data
+    vdup.16             q8, d21[3]
+    vst1.64             {d2,d3}, [r0], r1
+    vst1.64             {d22,d23}, [r0], r1
+
+
+    vadd.s16            q1, q0, q2
+    vadd.s16            q0, q0, q3
+    vadd.s16            q11, q8, q2
+    vadd.s16            q8, q8, q3
+    vqmovun.s16         d2, q1
+    vqmovun.s16         d3, q0
+    vqmovun.s16         d22, q11
+    vqmovun.s16         d23, q8
+    vld1.8              {d18}, [r3]!                  @ preload 8 left into r12
+    vmovl.u8            q10, d18
+    vst1.64             {d2,d3}, [r0], r1
+    vst1.64             {d22,d23}, [r0], r1
+
+    subs                r2, r2, #1
+    bgt                 loop_16x16_neon
+
+    bx                  lr
+	.size vpx_tm_predictor_16x16_neon, .-vpx_tm_predictor_16x16_neon    @ ENDP                @ |vpx_tm_predictor_16x16_neon|
+
+@void vpx_tm_predictor_32x32_neon (uint8_t *dst, ptrdiff_t y_stride,
+@                                  const uint8_t *above,
+@                                  const uint8_t *left)
+@ r0  uint8_t *dst
+@ r1  ptrdiff_t y_stride
+@ r2  const uint8_t *above
+@ r3  const uint8_t *left
+
+_vpx_tm_predictor_32x32_neon:
+	vpx_tm_predictor_32x32_neon: @ PROC
+    @ Load ytop_left = above[-1];
+    sub                 r12, r2, #1
+    vld1.8              {d0[]}, [r12]
+
+    @ Load above 32 pixels
+    vld1.8              {q1}, [r2]!
+    vld1.8              {q2}, [r2]
+
+    @ preload 8 left pixels
+    vld1.8              {d26}, [r3]!
+
+    @ Compute above - ytop_left
+    vsubl.u8            q8, d2, d0
+    vsubl.u8            q9, d3, d0
+    vsubl.u8            q10, d4, d0
+    vsubl.u8            q11, d5, d0
+
+    vmovl.u8            q3, d26
+
+    @ Load left row by row and compute left + (above - ytop_left)
+    @ Process 8 rows in each single loop and loop 4 times to process 32 rows.
+    mov                 r2, #4
+
+loop_32x32_neon:
+    @ Process two rows.
+    vdup.16             q0, d6[0]
+    vdup.16             q2, d6[1]
+    vadd.s16            q12, q0, q8
+    vadd.s16            q13, q0, q9
+    vadd.s16            q14, q0, q10
+    vadd.s16            q15, q0, q11
+    vqmovun.s16         d0, q12
+    vqmovun.s16         d1, q13
+    vadd.s16            q12, q2, q8
+    vadd.s16            q13, q2, q9
+    vqmovun.s16         d2, q14
+    vqmovun.s16         d3, q15
+    vadd.s16            q14, q2, q10
+    vadd.s16            q15, q2, q11
+    vst1.64             {d0-d3}, [r0], r1
+    vqmovun.s16         d24, q12
+    vqmovun.s16         d25, q13
+    vqmovun.s16         d26, q14
+    vqmovun.s16         d27, q15
+    vdup.16             q1, d6[2]
+    vdup.16             q2, d6[3]
+    vst1.64             {d24-d27}, [r0], r1
+
+    @ Process two rows.
+    vadd.s16            q12, q1, q8
+    vadd.s16            q13, q1, q9
+    vadd.s16            q14, q1, q10
+    vadd.s16            q15, q1, q11
+    vqmovun.s16         d0, q12
+    vqmovun.s16         d1, q13
+    vadd.s16            q12, q2, q8
+    vadd.s16            q13, q2, q9
+    vqmovun.s16         d2, q14
+    vqmovun.s16         d3, q15
+    vadd.s16            q14, q2, q10
+    vadd.s16            q15, q2, q11
+    vst1.64             {d0-d3}, [r0], r1
+    vqmovun.s16         d24, q12
+    vqmovun.s16         d25, q13
+    vqmovun.s16         d26, q14
+    vqmovun.s16         d27, q15
+    vdup.16             q0, d7[0]
+    vdup.16             q2, d7[1]
+    vst1.64             {d24-d27}, [r0], r1
+
+    @ Process two rows.
+    vadd.s16            q12, q0, q8
+    vadd.s16            q13, q0, q9
+    vadd.s16            q14, q0, q10
+    vadd.s16            q15, q0, q11
+    vqmovun.s16         d0, q12
+    vqmovun.s16         d1, q13
+    vadd.s16            q12, q2, q8
+    vadd.s16            q13, q2, q9
+    vqmovun.s16         d2, q14
+    vqmovun.s16         d3, q15
+    vadd.s16            q14, q2, q10
+    vadd.s16            q15, q2, q11
+    vst1.64             {d0-d3}, [r0], r1
+    vqmovun.s16         d24, q12
+    vqmovun.s16         d25, q13
+    vqmovun.s16         d26, q14
+    vqmovun.s16         d27, q15
+    vdup.16             q0, d7[2]
+    vdup.16             q2, d7[3]
+    vst1.64             {d24-d27}, [r0], r1
+
+    @ Process two rows.
+    vadd.s16            q12, q0, q8
+    vadd.s16            q13, q0, q9
+    vadd.s16            q14, q0, q10
+    vadd.s16            q15, q0, q11
+    vqmovun.s16         d0, q12
+    vqmovun.s16         d1, q13
+    vadd.s16            q12, q2, q8
+    vadd.s16            q13, q2, q9
+    vqmovun.s16         d2, q14
+    vqmovun.s16         d3, q15
+    vadd.s16            q14, q2, q10
+    vadd.s16            q15, q2, q11
+    vst1.64             {d0-d3}, [r0], r1
+    vqmovun.s16         d24, q12
+    vqmovun.s16         d25, q13
+    vld1.8              {d0}, [r3]!                   @ preload 8 left pixels
+    vqmovun.s16         d26, q14
+    vqmovun.s16         d27, q15
+    vmovl.u8            q3, d0
+    vst1.64             {d24-d27}, [r0], r1
+
+    subs                r2, r2, #1
+    bgt                 loop_32x32_neon
+
+    bx                  lr
+	.size vpx_tm_predictor_32x32_neon, .-vpx_tm_predictor_32x32_neon    @ ENDP                @ |vpx_tm_predictor_32x32_neon|
+
+	.section	.note.GNU-stack,"",%progbits

thirdparty/libvpx/vpx_dsp/arm/gas/loopfilter_mb_neon.s

@@ -0,0 +1,647 @@
+@ This file was created from a .asm file
+@  using the ads2gas.pl script.
+	.equ DO1STROUNDING, 0
+@
+@  Copyright (c) 2013 The WebM project authors. All Rights Reserved.
+@
+@  Use of this source code is governed by a BSD-style license
+@  that can be found in the LICENSE file in the root of the source
+@  tree. An additional intellectual property rights grant can be found
+@  in the file PATENTS.  All contributing project authors may
+@  be found in the AUTHORS file in the root of the source tree.
+@
+
+    .global vpx_lpf_horizontal_edge_8_neon 
+	.type vpx_lpf_horizontal_edge_8_neon, function
+    .global vpx_lpf_horizontal_edge_16_neon 
+	.type vpx_lpf_horizontal_edge_16_neon, function
+    .global vpx_lpf_vertical_16_neon 
+	.type vpx_lpf_vertical_16_neon, function
+   .arm
+
+.text
+.p2align 2
+
+@ void mb_lpf_horizontal_edge(uint8_t *s, int p,
+@                             const uint8_t *blimit,
+@                             const uint8_t *limit,
+@                             const uint8_t *thresh,
+@                             int count)
+@ r0    uint8_t *s,
+@ r1    int p, /* pitch */
+@ r2    const uint8_t *blimit,
+@ r3    const uint8_t *limit,
+@ sp    const uint8_t *thresh,
+@ r12   int count
+_mb_lpf_horizontal_edge:
+	mb_lpf_horizontal_edge: @ PROC
+    push        {r4-r8, lr}
+    vpush       {d8-d15}
+    ldr         r4, [sp, #88]              @ load thresh
+
+h_count:
+    vld1.8      {d16[]}, [r2]              @ load *blimit
+    vld1.8      {d17[]}, [r3]              @ load *limit
+    vld1.8      {d18[]}, [r4]              @ load *thresh
+
+    sub         r8, r0, r1, lsl #3         @ move src pointer down by 8 lines
+
+    vld1.u8     {d0}, [r8,:64], r1          @ p7
+    vld1.u8     {d1}, [r8,:64], r1          @ p6
+    vld1.u8     {d2}, [r8,:64], r1          @ p5
+    vld1.u8     {d3}, [r8,:64], r1          @ p4
+    vld1.u8     {d4}, [r8,:64], r1          @ p3
+    vld1.u8     {d5}, [r8,:64], r1          @ p2
+    vld1.u8     {d6}, [r8,:64], r1          @ p1
+    vld1.u8     {d7}, [r8,:64], r1          @ p0
+    vld1.u8     {d8}, [r8,:64], r1          @ q0
+    vld1.u8     {d9}, [r8,:64], r1          @ q1
+    vld1.u8     {d10}, [r8,:64], r1         @ q2
+    vld1.u8     {d11}, [r8,:64], r1         @ q3
+    vld1.u8     {d12}, [r8,:64], r1         @ q4
+    vld1.u8     {d13}, [r8,:64], r1         @ q5
+    vld1.u8     {d14}, [r8,:64], r1         @ q6
+    vld1.u8     {d15}, [r8,:64], r1         @ q7
+
+    bl          vpx_wide_mbfilter_neon
+
+    tst         r7, #1
+    beq         h_mbfilter
+
+    @ flat && mask were not set for any of the channels. Just store the values
+    @ from filter.
+    sub         r8, r0, r1, lsl #1
+
+    vst1.u8     {d25}, [r8,:64], r1         @ store op1
+    vst1.u8     {d24}, [r8,:64], r1         @ store op0
+    vst1.u8     {d23}, [r8,:64], r1         @ store oq0
+    vst1.u8     {d26}, [r8,:64], r1         @ store oq1
+
+    b           h_next
+
+h_mbfilter:
+    tst         r7, #2
+    beq         h_wide_mbfilter
+
+    @ flat2 was not set for any of the channels. Just store the values from
+    @ mbfilter.
+    sub         r8, r0, r1, lsl #1
+    sub         r8, r8, r1
+
+    vst1.u8     {d18}, [r8,:64], r1         @ store op2
+    vst1.u8     {d19}, [r8,:64], r1         @ store op1
+    vst1.u8     {d20}, [r8,:64], r1         @ store op0
+    vst1.u8     {d21}, [r8,:64], r1         @ store oq0
+    vst1.u8     {d22}, [r8,:64], r1         @ store oq1
+    vst1.u8     {d23}, [r8,:64], r1         @ store oq2
+
+    b           h_next
+
+h_wide_mbfilter:
+    sub         r8, r0, r1, lsl #3
+    add         r8, r8, r1
+
+    vst1.u8     {d16}, [r8,:64], r1         @ store op6
+    vst1.u8     {d24}, [r8,:64], r1         @ store op5
+    vst1.u8     {d25}, [r8,:64], r1         @ store op4
+    vst1.u8     {d26}, [r8,:64], r1         @ store op3
+    vst1.u8     {d27}, [r8,:64], r1         @ store op2
+    vst1.u8     {d18}, [r8,:64], r1         @ store op1
+    vst1.u8     {d19}, [r8,:64], r1         @ store op0
+    vst1.u8     {d20}, [r8,:64], r1         @ store oq0
+    vst1.u8     {d21}, [r8,:64], r1         @ store oq1
+    vst1.u8     {d22}, [r8,:64], r1         @ store oq2
+    vst1.u8     {d23}, [r8,:64], r1         @ store oq3
+    vst1.u8     {d1}, [r8,:64], r1          @ store oq4
+    vst1.u8     {d2}, [r8,:64], r1          @ store oq5
+    vst1.u8     {d3}, [r8,:64], r1          @ store oq6
+
+h_next:
+    add         r0, r0, #8
+    subs        r12, r12, #1
+    bne         h_count
+
+    vpop        {d8-d15}
+    pop         {r4-r8, pc}
+
+	.size mb_lpf_horizontal_edge, .-mb_lpf_horizontal_edge    @ ENDP        @ |mb_lpf_horizontal_edge|
+
+@ void vpx_lpf_horizontal_edge_8_neon(uint8_t *s, int pitch,
+@                                     const uint8_t *blimit,
+@                                     const uint8_t *limit,
+@                                     const uint8_t *thresh)
+@ r0    uint8_t *s,
+@ r1    int pitch,
+@ r2    const uint8_t *blimit,
+@ r3    const uint8_t *limit,
+@ sp    const uint8_t *thresh
+_vpx_lpf_horizontal_edge_8_neon:
+	vpx_lpf_horizontal_edge_8_neon: @ PROC
+    mov r12, #1
+    b mb_lpf_horizontal_edge
+	.size vpx_lpf_horizontal_edge_8_neon, .-vpx_lpf_horizontal_edge_8_neon    @ ENDP        @ |vpx_lpf_horizontal_edge_8_neon|
+
+@ void vpx_lpf_horizontal_edge_16_neon(uint8_t *s, int pitch,
+@                                      const uint8_t *blimit,
+@                                      const uint8_t *limit,
+@                                      const uint8_t *thresh)
+@ r0    uint8_t *s,
+@ r1    int pitch,
+@ r2    const uint8_t *blimit,
+@ r3    const uint8_t *limit,
+@ sp    const uint8_t *thresh
+_vpx_lpf_horizontal_edge_16_neon:
+	vpx_lpf_horizontal_edge_16_neon: @ PROC
+    mov r12, #2
+    b mb_lpf_horizontal_edge
+	.size vpx_lpf_horizontal_edge_16_neon, .-vpx_lpf_horizontal_edge_16_neon    @ ENDP        @ |vpx_lpf_horizontal_edge_16_neon|
+
+@ void vpx_lpf_vertical_16_neon(uint8_t *s, int p,
+@                               const uint8_t *blimit,
+@                               const uint8_t *limit,
+@                               const uint8_t *thresh)
+@ r0    uint8_t *s,
+@ r1    int p, /* pitch */
+@ r2    const uint8_t *blimit,
+@ r3    const uint8_t *limit,
+@ sp    const uint8_t *thresh,
+_vpx_lpf_vertical_16_neon:
+	vpx_lpf_vertical_16_neon: @ PROC
+    push        {r4-r8, lr}
+    vpush       {d8-d15}
+    ldr         r4, [sp, #88]              @ load thresh
+
+    vld1.8      {d16[]}, [r2]              @ load *blimit
+    vld1.8      {d17[]}, [r3]              @ load *limit
+    vld1.8      {d18[]}, [r4]              @ load *thresh
+
+    sub         r8, r0, #8
+
+    vld1.8      {d0}, [r8,:64], r1
+    vld1.8      {d8}, [r0,:64], r1
+    vld1.8      {d1}, [r8,:64], r1
+    vld1.8      {d9}, [r0,:64], r1
+    vld1.8      {d2}, [r8,:64], r1
+    vld1.8      {d10}, [r0,:64], r1
+    vld1.8      {d3}, [r8,:64], r1
+    vld1.8      {d11}, [r0,:64], r1
+    vld1.8      {d4}, [r8,:64], r1
+    vld1.8      {d12}, [r0,:64], r1
+    vld1.8      {d5}, [r8,:64], r1
+    vld1.8      {d13}, [r0,:64], r1
+    vld1.8      {d6}, [r8,:64], r1
+    vld1.8      {d14}, [r0,:64], r1
+    vld1.8      {d7}, [r8,:64], r1
+    vld1.8      {d15}, [r0,:64], r1
+
+    sub         r0, r0, r1, lsl #3
+
+    vtrn.32     q0, q2
+    vtrn.32     q1, q3
+    vtrn.32     q4, q6
+    vtrn.32     q5, q7
+
+    vtrn.16     q0, q1
+    vtrn.16     q2, q3
+    vtrn.16     q4, q5
+    vtrn.16     q6, q7
+
+    vtrn.8      d0, d1
+    vtrn.8      d2, d3
+    vtrn.8      d4, d5
+    vtrn.8      d6, d7
+
+    vtrn.8      d8, d9
+    vtrn.8      d10, d11
+    vtrn.8      d12, d13
+    vtrn.8      d14, d15
+
+    bl          vpx_wide_mbfilter_neon
+
+    tst         r7, #1
+    beq         v_mbfilter
+
+    @ flat && mask were not set for any of the channels. Just store the values
+    @ from filter.
+    sub         r8, r0, #2
+
+    vswp        d23, d25
+
+    vst4.8      {d23[0], d24[0], d25[0], d26[0]}, [r8], r1
+    vst4.8      {d23[1], d24[1], d25[1], d26[1]}, [r8], r1
+    vst4.8      {d23[2], d24[2], d25[2], d26[2]}, [r8], r1
+    vst4.8      {d23[3], d24[3], d25[3], d26[3]}, [r8], r1
+    vst4.8      {d23[4], d24[4], d25[4], d26[4]}, [r8], r1
+    vst4.8      {d23[5], d24[5], d25[5], d26[5]}, [r8], r1
+    vst4.8      {d23[6], d24[6], d25[6], d26[6]}, [r8], r1
+    vst4.8      {d23[7], d24[7], d25[7], d26[7]}, [r8], r1
+
+    b           v_end
+
+v_mbfilter:
+    tst         r7, #2
+    beq         v_wide_mbfilter
+
+    @ flat2 was not set for any of the channels. Just store the values from
+    @ mbfilter.
+    sub         r8, r0, #3
+
+    vst3.8      {d18[0], d19[0], d20[0]}, [r8], r1
+    vst3.8      {d21[0], d22[0], d23[0]}, [r0], r1
+    vst3.8      {d18[1], d19[1], d20[1]}, [r8], r1
+    vst3.8      {d21[1], d22[1], d23[1]}, [r0], r1
+    vst3.8      {d18[2], d19[2], d20[2]}, [r8], r1
+    vst3.8      {d21[2], d22[2], d23[2]}, [r0], r1
+    vst3.8      {d18[3], d19[3], d20[3]}, [r8], r1
+    vst3.8      {d21[3], d22[3], d23[3]}, [r0], r1
+    vst3.8      {d18[4], d19[4], d20[4]}, [r8], r1
+    vst3.8      {d21[4], d22[4], d23[4]}, [r0], r1
+    vst3.8      {d18[5], d19[5], d20[5]}, [r8], r1
+    vst3.8      {d21[5], d22[5], d23[5]}, [r0], r1
+    vst3.8      {d18[6], d19[6], d20[6]}, [r8], r1
+    vst3.8      {d21[6], d22[6], d23[6]}, [r0], r1
+    vst3.8      {d18[7], d19[7], d20[7]}, [r8], r1
+    vst3.8      {d21[7], d22[7], d23[7]}, [r0], r1
+
+    b           v_end
+
+v_wide_mbfilter:
+    sub         r8, r0, #8
+
+    vtrn.32     d0,  d26
+    vtrn.32     d16, d27
+    vtrn.32     d24, d18
+    vtrn.32     d25, d19
+
+    vtrn.16     d0,  d24
+    vtrn.16     d16, d25
+    vtrn.16     d26, d18
+    vtrn.16     d27, d19
+
+    vtrn.8      d0,  d16
+    vtrn.8      d24, d25
+    vtrn.8      d26, d27
+    vtrn.8      d18, d19
+
+    vtrn.32     d20, d1
+    vtrn.32     d21, d2
+    vtrn.32     d22, d3
+    vtrn.32     d23, d15
+
+    vtrn.16     d20, d22
+    vtrn.16     d21, d23
+    vtrn.16     d1,  d3
+    vtrn.16     d2,  d15
+
+    vtrn.8      d20, d21
+    vtrn.8      d22, d23
+    vtrn.8      d1,  d2
+    vtrn.8      d3,  d15
+
+    vst1.8      {d0}, [r8,:64], r1
+    vst1.8      {d20}, [r0,:64], r1
+    vst1.8      {d16}, [r8,:64], r1
+    vst1.8      {d21}, [r0,:64], r1
+    vst1.8      {d24}, [r8,:64], r1
+    vst1.8      {d22}, [r0,:64], r1
+    vst1.8      {d25}, [r8,:64], r1
+    vst1.8      {d23}, [r0,:64], r1
+    vst1.8      {d26}, [r8,:64], r1
+    vst1.8      {d1}, [r0,:64], r1
+    vst1.8      {d27}, [r8,:64], r1
+    vst1.8      {d2}, [r0,:64], r1
+    vst1.8      {d18}, [r8,:64], r1
+    vst1.8      {d3}, [r0,:64], r1
+    vst1.8      {d19}, [r8,:64], r1
+    vst1.8      {d15}, [r0,:64], r1
+
+v_end:
+    vpop        {d8-d15}
+    pop         {r4-r8, pc}
+
+	.size vpx_lpf_vertical_16_neon, .-vpx_lpf_vertical_16_neon    @ ENDP        @ |vpx_lpf_vertical_16_neon|
+
+@ void vpx_wide_mbfilter_neon();
+@ This is a helper function for the loopfilters. The invidual functions do the
+@ necessary load, transpose (if necessary) and store.
+@
+@ r0-r3 PRESERVE
+@ d16    blimit
+@ d17    limit
+@ d18    thresh
+@ d0    p7
+@ d1    p6
+@ d2    p5
+@ d3    p4
+@ d4    p3
+@ d5    p2
+@ d6    p1
+@ d7    p0
+@ d8    q0
+@ d9    q1
+@ d10   q2
+@ d11   q3
+@ d12   q4
+@ d13   q5
+@ d14   q6
+@ d15   q7
+_vpx_wide_mbfilter_neon:
+	vpx_wide_mbfilter_neon: @ PROC
+    mov         r7, #0
+
+    @ filter_mask
+    vabd.u8     d19, d4, d5                @ abs(p3 - p2)
+    vabd.u8     d20, d5, d6                @ abs(p2 - p1)
+    vabd.u8     d21, d6, d7                @ abs(p1 - p0)
+    vabd.u8     d22, d9, d8                @ abs(q1 - q0)
+    vabd.u8     d23, d10, d9               @ abs(q2 - q1)
+    vabd.u8     d24, d11, d10              @ abs(q3 - q2)
+
+    @ only compare the largest value to limit
+    vmax.u8     d19, d19, d20              @ max(abs(p3 - p2), abs(p2 - p1))
+    vmax.u8     d20, d21, d22              @ max(abs(p1 - p0), abs(q1 - q0))
+    vmax.u8     d23, d23, d24              @ max(abs(q2 - q1), abs(q3 - q2))
+    vmax.u8     d19, d19, d20
+
+    vabd.u8     d24, d7, d8                @ abs(p0 - q0)
+
+    vmax.u8     d19, d19, d23
+
+    vabd.u8     d23, d6, d9                @ a = abs(p1 - q1)
+    vqadd.u8    d24, d24, d24              @ b = abs(p0 - q0) * 2
+
+    @ abs () > limit
+    vcge.u8     d19, d17, d19
+
+    @ flatmask4
+    vabd.u8     d25, d7, d5                @ abs(p0 - p2)
+    vabd.u8     d26, d8, d10               @ abs(q0 - q2)
+    vabd.u8     d27, d4, d7                @ abs(p3 - p0)
+    vabd.u8     d28, d11, d8               @ abs(q3 - q0)
+
+    @ only compare the largest value to thresh
+    vmax.u8     d25, d25, d26              @ max(abs(p0 - p2), abs(q0 - q2))
+    vmax.u8     d26, d27, d28              @ max(abs(p3 - p0), abs(q3 - q0))
+    vmax.u8     d25, d25, d26
+    vmax.u8     d20, d20, d25
+
+    vshr.u8     d23, d23, #1               @ a = a / 2
+    vqadd.u8    d24, d24, d23              @ a = b + a
+
+    vmov.u8     d30, #1
+    vcge.u8     d24, d16, d24              @ (a > blimit * 2 + limit) * -1
+
+    vcge.u8     d20, d30, d20              @ flat
+
+    vand        d19, d19, d24              @ mask
+
+    @ hevmask
+    vcgt.u8     d21, d21, d18              @ (abs(p1 - p0) > thresh)*-1
+    vcgt.u8     d22, d22, d18              @ (abs(q1 - q0) > thresh)*-1
+    vorr        d21, d21, d22              @ hev
+
+    vand        d16, d20, d19              @ flat && mask
+    vmov        r5, r6, d16
+
+    @ flatmask5(1, p7, p6, p5, p4, p0, q0, q4, q5, q6, q7)
+    vabd.u8     d22, d3, d7                @ abs(p4 - p0)
+    vabd.u8     d23, d12, d8               @ abs(q4 - q0)
+    vabd.u8     d24, d7, d2                @ abs(p0 - p5)
+    vabd.u8     d25, d8, d13               @ abs(q0 - q5)
+    vabd.u8     d26, d1, d7                @ abs(p6 - p0)
+    vabd.u8     d27, d14, d8               @ abs(q6 - q0)
+    vabd.u8     d28, d0, d7                @ abs(p7 - p0)
+    vabd.u8     d29, d15, d8               @ abs(q7 - q0)
+
+    @ only compare the largest value to thresh
+    vmax.u8     d22, d22, d23              @ max(abs(p4 - p0), abs(q4 - q0))
+    vmax.u8     d23, d24, d25              @ max(abs(p0 - p5), abs(q0 - q5))
+    vmax.u8     d24, d26, d27              @ max(abs(p6 - p0), abs(q6 - q0))
+    vmax.u8     d25, d28, d29              @ max(abs(p7 - p0), abs(q7 - q0))
+
+    vmax.u8     d26, d22, d23
+    vmax.u8     d27, d24, d25
+    vmax.u8     d23, d26, d27
+
+    vcge.u8     d18, d30, d23              @ flat2
+
+    vmov.u8     d22, #0x80
+
+    orrs        r5, r5, r6                 @ Check for 0
+    orreq       r7, r7, #1                 @ Only do filter branch
+
+    vand        d17, d18, d16              @ flat2 && flat && mask
+    vmov        r5, r6, d17
+
+    @ mbfilter() function
+
+    @ filter() function
+    @ convert to signed
+    veor        d23, d8, d22               @ qs0
+    veor        d24, d7, d22               @ ps0
+    veor        d25, d6, d22               @ ps1
+    veor        d26, d9, d22               @ qs1
+
+    vmov.u8     d27, #3
+
+    vsub.s8     d28, d23, d24              @ ( qs0 - ps0)
+    vqsub.s8    d29, d25, d26              @ filter = clamp(ps1-qs1)
+    vmull.s8    q15, d28, d27              @ 3 * ( qs0 - ps0)
+    vand        d29, d29, d21              @ filter &= hev
+    vaddw.s8    q15, q15, d29              @ filter + 3 * (qs0 - ps0)
+    vmov.u8     d29, #4
+
+    @ filter = clamp(filter + 3 * ( qs0 - ps0))
+    vqmovn.s16  d28, q15
+
+    vand        d28, d28, d19              @ filter &= mask
+
+    vqadd.s8    d30, d28, d27              @ filter2 = clamp(filter+3)
+    vqadd.s8    d29, d28, d29              @ filter1 = clamp(filter+4)
+    vshr.s8     d30, d30, #3               @ filter2 >>= 3
+    vshr.s8     d29, d29, #3               @ filter1 >>= 3
+
+
+    vqadd.s8    d24, d24, d30              @ op0 = clamp(ps0 + filter2)
+    vqsub.s8    d23, d23, d29              @ oq0 = clamp(qs0 - filter1)
+
+    @ outer tap adjustments: ++filter1 >> 1
+    vrshr.s8    d29, d29, #1
+    vbic        d29, d29, d21              @ filter &= ~hev
+
+    vqadd.s8    d25, d25, d29              @ op1 = clamp(ps1 + filter)
+    vqsub.s8    d26, d26, d29              @ oq1 = clamp(qs1 - filter)
+
+    veor        d24, d24, d22              @ *f_op0 = u^0x80
+    veor        d23, d23, d22              @ *f_oq0 = u^0x80
+    veor        d25, d25, d22              @ *f_op1 = u^0x80
+    veor        d26, d26, d22              @ *f_oq1 = u^0x80
+
+    tst         r7, #1
+    bxne        lr
+
+    orrs        r5, r5, r6                 @ Check for 0
+    orreq       r7, r7, #2                 @ Only do mbfilter branch
+
+    @ mbfilter flat && mask branch
+    @ TODO(fgalligan): Can I decrease the cycles shifting to consective d's
+    @ and using vibt on the q's?
+    vmov.u8     d29, #2
+    vaddl.u8    q15, d7, d8                @ op2 = p0 + q0
+    vmlal.u8    q15, d4, d27               @ op2 = p0 + q0 + p3 * 3
+    vmlal.u8    q15, d5, d29               @ op2 = p0 + q0 + p3 * 3 + p2 * 2
+    vaddl.u8    q10, d4, d5
+    vaddw.u8    q15, d6                    @ op2=p1 + p0 + q0 + p3 * 3 + p2 *2
+    vaddl.u8    q14, d6, d9
+    vqrshrn.u16 d18, q15, #3               @ r_op2
+
+    vsub.i16    q15, q10
+    vaddl.u8    q10, d4, d6
+    vadd.i16    q15, q14
+    vaddl.u8    q14, d7, d10
+    vqrshrn.u16 d19, q15, #3               @ r_op1
+
+    vsub.i16    q15, q10
+    vadd.i16    q15, q14
+    vaddl.u8    q14, d8, d11
+    vqrshrn.u16 d20, q15, #3               @ r_op0
+
+    vsubw.u8    q15, d4                    @ oq0 = op0 - p3
+    vsubw.u8    q15, d7                    @ oq0 -= p0
+    vadd.i16    q15, q14
+    vaddl.u8    q14, d9, d11
+    vqrshrn.u16 d21, q15, #3               @ r_oq0
+
+    vsubw.u8    q15, d5                    @ oq1 = oq0 - p2
+    vsubw.u8    q15, d8                    @ oq1 -= q0
+    vadd.i16    q15, q14
+    vaddl.u8    q14, d10, d11
+    vqrshrn.u16 d22, q15, #3               @ r_oq1
+
+    vsubw.u8    q15, d6                    @ oq2 = oq0 - p1
+    vsubw.u8    q15, d9                    @ oq2 -= q1
+    vadd.i16    q15, q14
+    vqrshrn.u16 d27, q15, #3               @ r_oq2
+
+    @ Filter does not set op2 or oq2, so use p2 and q2.
+    vbif        d18, d5, d16               @ t_op2 |= p2 & ~(flat & mask)
+    vbif        d19, d25, d16              @ t_op1 |= f_op1 & ~(flat & mask)
+    vbif        d20, d24, d16              @ t_op0 |= f_op0 & ~(flat & mask)
+    vbif        d21, d23, d16              @ t_oq0 |= f_oq0 & ~(flat & mask)
+    vbif        d22, d26, d16              @ t_oq1 |= f_oq1 & ~(flat & mask)
+
+    vbit        d23, d27, d16              @ t_oq2 |= r_oq2 & (flat & mask)
+    vbif        d23, d10, d16              @ t_oq2 |= q2 & ~(flat & mask)
+
+    tst         r7, #2
+    bxne        lr
+
+    @ wide_mbfilter flat2 && flat && mask branch
+    vmov.u8     d16, #7
+    vaddl.u8    q15, d7, d8                @ op6 = p0 + q0
+    vaddl.u8    q12, d2, d3
+    vaddl.u8    q13, d4, d5
+    vaddl.u8    q14, d1, d6
+    vmlal.u8    q15, d0, d16               @ op6 += p7 * 3
+    vadd.i16    q12, q13
+    vadd.i16    q15, q14
+    vaddl.u8    q14, d2, d9
+    vadd.i16    q15, q12
+    vaddl.u8    q12, d0, d1
+    vaddw.u8    q15, d1
+    vaddl.u8    q13, d0, d2
+    vadd.i16    q14, q15, q14
+    vqrshrn.u16 d16, q15, #4               @ w_op6
+
+    vsub.i16    q15, q14, q12
+    vaddl.u8    q14, d3, d10
+    vqrshrn.u16 d24, q15, #4               @ w_op5
+
+    vsub.i16    q15, q13
+    vaddl.u8    q13, d0, d3
+    vadd.i16    q15, q14
+    vaddl.u8    q14, d4, d11
+    vqrshrn.u16 d25, q15, #4               @ w_op4
+
+    vadd.i16    q15, q14
+    vaddl.u8    q14, d0, d4
+    vsub.i16    q15, q13
+    vsub.i16    q14, q15, q14
+    vqrshrn.u16 d26, q15, #4               @ w_op3
+
+    vaddw.u8    q15, q14, d5               @ op2 += p2
+    vaddl.u8    q14, d0, d5
+    vaddw.u8    q15, d12                   @ op2 += q4
+    vbif        d26, d4, d17               @ op3 |= p3 & ~(f2 & f & m)
+    vqrshrn.u16 d27, q15, #4               @ w_op2
+
+    vsub.i16    q15, q14
+    vaddl.u8    q14, d0, d6
+    vaddw.u8    q15, d6                    @ op1 += p1
+    vaddw.u8    q15, d13                   @ op1 += q5
+    vbif        d27, d18, d17              @ op2 |= t_op2 & ~(f2 & f & m)
+    vqrshrn.u16 d18, q15, #4               @ w_op1
+
+    vsub.i16    q15, q14
+    vaddl.u8    q14, d0, d7
+    vaddw.u8    q15, d7                    @ op0 += p0
+    vaddw.u8    q15, d14                   @ op0 += q6
+    vbif        d18, d19, d17              @ op1 |= t_op1 & ~(f2 & f & m)
+    vqrshrn.u16 d19, q15, #4               @ w_op0
+
+    vsub.i16    q15, q14
+    vaddl.u8    q14, d1, d8
+    vaddw.u8    q15, d8                    @ oq0 += q0
+    vaddw.u8    q15, d15                   @ oq0 += q7
+    vbif        d19, d20, d17              @ op0 |= t_op0 & ~(f2 & f & m)
+    vqrshrn.u16 d20, q15, #4               @ w_oq0
+
+    vsub.i16    q15, q14
+    vaddl.u8    q14, d2, d9
+    vaddw.u8    q15, d9                    @ oq1 += q1
+    vaddl.u8    q4, d10, d15
+    vaddw.u8    q15, d15                   @ oq1 += q7
+    vbif        d20, d21, d17              @ oq0 |= t_oq0 & ~(f2 & f & m)
+    vqrshrn.u16 d21, q15, #4               @ w_oq1
+
+    vsub.i16    q15, q14
+    vaddl.u8    q14, d3, d10
+    vadd.i16    q15, q4
+    vaddl.u8    q4, d11, d15
+    vbif        d21, d22, d17              @ oq1 |= t_oq1 & ~(f2 & f & m)
+    vqrshrn.u16 d22, q15, #4               @ w_oq2
+
+    vsub.i16    q15, q14
+    vaddl.u8    q14, d4, d11
+    vadd.i16    q15, q4
+    vaddl.u8    q4, d12, d15
+    vbif        d22, d23, d17              @ oq2 |= t_oq2 & ~(f2 & f & m)
+    vqrshrn.u16 d23, q15, #4               @ w_oq3
+
+    vsub.i16    q15, q14
+    vaddl.u8    q14, d5, d12
+    vadd.i16    q15, q4
+    vaddl.u8    q4, d13, d15
+    vbif        d16, d1, d17               @ op6 |= p6 & ~(f2 & f & m)
+    vqrshrn.u16 d1, q15, #4                @ w_oq4
+
+    vsub.i16    q15, q14
+    vaddl.u8    q14, d6, d13
+    vadd.i16    q15, q4
+    vaddl.u8    q4, d14, d15
+    vbif        d24, d2, d17               @ op5 |= p5 & ~(f2 & f & m)
+    vqrshrn.u16 d2, q15, #4                @ w_oq5
+
+    vsub.i16    q15, q14
+    vbif        d25, d3, d17               @ op4 |= p4 & ~(f2 & f & m)
+    vadd.i16    q15, q4
+    vbif        d23, d11, d17              @ oq3 |= q3 & ~(f2 & f & m)
+    vqrshrn.u16 d3, q15, #4                @ w_oq6
+    vbif        d1, d12, d17               @ oq4 |= q4 & ~(f2 & f & m)
+    vbif        d2, d13, d17               @ oq5 |= q5 & ~(f2 & f & m)
+    vbif        d3, d14, d17               @ oq6 |= q6 & ~(f2 & f & m)
+
+    bx          lr
+	.size vpx_wide_mbfilter_neon, .-vpx_wide_mbfilter_neon    @ ENDP        @ |vpx_wide_mbfilter_neon|
+
+	.section	.note.GNU-stack,"",%progbits

thirdparty/libvpx/vpx_dsp/arm/gas/save_reg_neon.s

@@ -0,0 +1,44 @@
+@ This file was created from a .asm file
+@  using the ads2gas.pl script.
+	.equ DO1STROUNDING, 0
+@
+@  Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+@
+@  Use of this source code is governed by a BSD-style license
+@  that can be found in the LICENSE file in the root of the source
+@  tree. An additional intellectual property rights grant can be found
+@  in the file PATENTS.  All contributing project authors may
+@  be found in the AUTHORS file in the root of the source tree.
+@
+
+
+    .global vpx_push_neon 
+	.type vpx_push_neon, function
+    .global vpx_pop_neon 
+	.type vpx_pop_neon, function
+
+   .arm
+   .eabi_attribute 24, 1 @Tag_ABI_align_needed
+   .eabi_attribute 25, 1 @Tag_ABI_align_preserved
+
+.text
+.p2align 2
+
+_vpx_push_neon:
+	vpx_push_neon: @ PROC
+    vst1.i64            {d8, d9, d10, d11}, [r0]!
+    vst1.i64            {d12, d13, d14, d15}, [r0]!
+    bx              lr
+
+	.size vpx_push_neon, .-vpx_push_neon    @ ENDP
+
+_vpx_pop_neon:
+	vpx_pop_neon: @ PROC
+    vld1.i64            {d8, d9, d10, d11}, [r0]!
+    vld1.i64            {d12, d13, d14, d15}, [r0]!
+    bx              lr
+
+	.size vpx_pop_neon, .-vpx_pop_neon    @ ENDP
+
+
+	.section	.note.GNU-stack,"",%progbits

thirdparty/libvpx/vpx_dsp/arm/gas_apple/intrapred_neon_asm.s

@@ -0,0 +1,660 @@
+@ This file was created from a .asm file
+@  using the ads2gas_apple.pl script.
+
+	.set WIDE_REFERENCE, 0
+	.set ARCHITECTURE, 5
+	.set DO1STROUNDING, 0
+ @
+ @  Copyright (c) 2014 The WebM project authors. All Rights Reserved.
+ @
+ @  Use of this source code is governed by a BSD-style license
+ @  that can be found in the LICENSE file in the root of the source
+ @  tree. An additional intellectual property rights grant can be found
+ @  in the file PATENTS.  All contributing project authors may
+ @  be found in the AUTHORS file in the root of the source tree.
+ @
+
+    .globl _vpx_v_predictor_4x4_neon
+	.globl vpx_v_predictor_4x4_neon
+    .globl _vpx_v_predictor_8x8_neon
+	.globl vpx_v_predictor_8x8_neon
+    .globl _vpx_v_predictor_16x16_neon
+	.globl vpx_v_predictor_16x16_neon
+    .globl _vpx_v_predictor_32x32_neon
+	.globl vpx_v_predictor_32x32_neon
+    .globl _vpx_h_predictor_4x4_neon
+	.globl vpx_h_predictor_4x4_neon
+    .globl _vpx_h_predictor_8x8_neon
+	.globl vpx_h_predictor_8x8_neon
+    .globl _vpx_h_predictor_16x16_neon
+	.globl vpx_h_predictor_16x16_neon
+    .globl _vpx_h_predictor_32x32_neon
+	.globl vpx_h_predictor_32x32_neon
+    .globl _vpx_tm_predictor_4x4_neon
+	.globl vpx_tm_predictor_4x4_neon
+    .globl _vpx_tm_predictor_8x8_neon
+	.globl vpx_tm_predictor_8x8_neon
+    .globl _vpx_tm_predictor_16x16_neon
+	.globl vpx_tm_predictor_16x16_neon
+    .globl _vpx_tm_predictor_32x32_neon
+	.globl vpx_tm_predictor_32x32_neon
+   @ ARM
+   @ 
+   @ PRESERVE8
+
+.text
+.p2align 2
+
+ @void vpx_v_predictor_4x4_neon(uint8_t *dst, ptrdiff_t y_stride,
+ @                              const uint8_t *above,
+ @                              const uint8_t *left)
+ @ r0  uint8_t *dst
+ @ r1  ptrdiff_t y_stride
+ @ r2  const uint8_t *above
+ @ r3  const uint8_t *left
+
+_vpx_v_predictor_4x4_neon:
+	vpx_v_predictor_4x4_neon: @
+    vld1.32             {d0[0]}, [r2]
+    vst1.32             {d0[0]}, [r0], r1
+    vst1.32             {d0[0]}, [r0], r1
+    vst1.32             {d0[0]}, [r0], r1
+    vst1.32             {d0[0]}, [r0], r1
+    bx                  lr
+    @                 @ |vpx_v_predictor_4x4_neon|
+
+ @void vpx_v_predictor_8x8_neon(uint8_t *dst, ptrdiff_t y_stride,
+ @                              const uint8_t *above,
+ @                              const uint8_t *left)
+ @ r0  uint8_t *dst
+ @ r1  ptrdiff_t y_stride
+ @ r2  const uint8_t *above
+ @ r3  const uint8_t *left
+
+_vpx_v_predictor_8x8_neon:
+	vpx_v_predictor_8x8_neon: @
+    vld1.8              {d0}, [r2]
+    vst1.8              {d0}, [r0], r1
+    vst1.8              {d0}, [r0], r1
+    vst1.8              {d0}, [r0], r1
+    vst1.8              {d0}, [r0], r1
+    vst1.8              {d0}, [r0], r1
+    vst1.8              {d0}, [r0], r1
+    vst1.8              {d0}, [r0], r1
+    vst1.8              {d0}, [r0], r1
+    bx                  lr
+    @                 @ |vpx_v_predictor_8x8_neon|
+
+ @void vpx_v_predictor_16x16_neon(uint8_t *dst, ptrdiff_t y_stride,
+ @                                const uint8_t *above,
+ @                                const uint8_t *left)
+ @ r0  uint8_t *dst
+ @ r1  ptrdiff_t y_stride
+ @ r2  const uint8_t *above
+ @ r3  const uint8_t *left
+
+_vpx_v_predictor_16x16_neon:
+	vpx_v_predictor_16x16_neon: @
+    vld1.8              {q0}, [r2]
+    vst1.8              {q0}, [r0], r1
+    vst1.8              {q0}, [r0], r1
+    vst1.8              {q0}, [r0], r1
+    vst1.8              {q0}, [r0], r1
+    vst1.8              {q0}, [r0], r1
+    vst1.8              {q0}, [r0], r1
+    vst1.8              {q0}, [r0], r1
+    vst1.8              {q0}, [r0], r1
+    vst1.8              {q0}, [r0], r1
+    vst1.8              {q0}, [r0], r1
+    vst1.8              {q0}, [r0], r1
+    vst1.8              {q0}, [r0], r1
+    vst1.8              {q0}, [r0], r1
+    vst1.8              {q0}, [r0], r1
+    vst1.8              {q0}, [r0], r1
+    vst1.8              {q0}, [r0], r1
+    bx                  lr
+    @                 @ |vpx_v_predictor_16x16_neon|
+
+ @void vpx_v_predictor_32x32_neon(uint8_t *dst, ptrdiff_t y_stride,
+ @                                const uint8_t *above,
+ @                                const uint8_t *left)
+ @ r0  uint8_t *dst
+ @ r1  ptrdiff_t y_stride
+ @ r2  const uint8_t *above
+ @ r3  const uint8_t *left
+
+_vpx_v_predictor_32x32_neon:
+	vpx_v_predictor_32x32_neon: @
+    vld1.8              {q0, q1}, [r2]
+    mov                 r2, #2
+loop_v:
+    vst1.8              {q0, q1}, [r0], r1
+    vst1.8              {q0, q1}, [r0], r1
+    vst1.8              {q0, q1}, [r0], r1
+    vst1.8              {q0, q1}, [r0], r1
+    vst1.8              {q0, q1}, [r0], r1
+    vst1.8              {q0, q1}, [r0], r1
+    vst1.8              {q0, q1}, [r0], r1
+    vst1.8              {q0, q1}, [r0], r1
+    vst1.8              {q0, q1}, [r0], r1
+    vst1.8              {q0, q1}, [r0], r1
+    vst1.8              {q0, q1}, [r0], r1
+    vst1.8              {q0, q1}, [r0], r1
+    vst1.8              {q0, q1}, [r0], r1
+    vst1.8              {q0, q1}, [r0], r1
+    vst1.8              {q0, q1}, [r0], r1
+    vst1.8              {q0, q1}, [r0], r1
+    subs                r2, r2, #1
+    bgt                 loop_v
+    bx                  lr
+    @                 @ |vpx_v_predictor_32x32_neon|
+
+ @void vpx_h_predictor_4x4_neon(uint8_t *dst, ptrdiff_t y_stride,
+ @                              const uint8_t *above,
+ @                              const uint8_t *left)
+ @ r0  uint8_t *dst
+ @ r1  ptrdiff_t y_stride
+ @ r2  const uint8_t *above
+ @ r3  const uint8_t *left
+
+_vpx_h_predictor_4x4_neon:
+	vpx_h_predictor_4x4_neon: @
+    vld1.32             {d1[0]}, [r3]
+    vdup.8              d0, d1[0]
+    vst1.32             {d0[0]}, [r0], r1
+    vdup.8              d0, d1[1]
+    vst1.32             {d0[0]}, [r0], r1
+    vdup.8              d0, d1[2]
+    vst1.32             {d0[0]}, [r0], r1
+    vdup.8              d0, d1[3]
+    vst1.32             {d0[0]}, [r0], r1
+    bx                  lr
+    @                 @ |vpx_h_predictor_4x4_neon|
+
+ @void vpx_h_predictor_8x8_neon(uint8_t *dst, ptrdiff_t y_stride,
+ @                              const uint8_t *above,
+ @                              const uint8_t *left)
+ @ r0  uint8_t *dst
+ @ r1  ptrdiff_t y_stride
+ @ r2  const uint8_t *above
+ @ r3  const uint8_t *left
+
+_vpx_h_predictor_8x8_neon:
+	vpx_h_predictor_8x8_neon: @
+    vld1.64             {d1}, [r3]
+    vdup.8              d0, d1[0]
+    vst1.64             {d0}, [r0], r1
+    vdup.8              d0, d1[1]
+    vst1.64             {d0}, [r0], r1
+    vdup.8              d0, d1[2]
+    vst1.64             {d0}, [r0], r1
+    vdup.8              d0, d1[3]
+    vst1.64             {d0}, [r0], r1
+    vdup.8              d0, d1[4]
+    vst1.64             {d0}, [r0], r1
+    vdup.8              d0, d1[5]
+    vst1.64             {d0}, [r0], r1
+    vdup.8              d0, d1[6]
+    vst1.64             {d0}, [r0], r1
+    vdup.8              d0, d1[7]
+    vst1.64             {d0}, [r0], r1
+    bx                  lr
+    @                 @ |vpx_h_predictor_8x8_neon|
+
+ @void vpx_h_predictor_16x16_neon(uint8_t *dst, ptrdiff_t y_stride,
+ @                                const uint8_t *above,
+ @                                const uint8_t *left)
+ @ r0  uint8_t *dst
+ @ r1  ptrdiff_t y_stride
+ @ r2  const uint8_t *above
+ @ r3  const uint8_t *left
+
+_vpx_h_predictor_16x16_neon:
+	vpx_h_predictor_16x16_neon: @
+    vld1.8              {q1}, [r3]
+    vdup.8              q0, d2[0]
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d2[1]
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d2[2]
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d2[3]
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d2[4]
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d2[5]
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d2[6]
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d2[7]
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d3[0]
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d3[1]
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d3[2]
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d3[3]
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d3[4]
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d3[5]
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d3[6]
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d3[7]
+    vst1.8              {q0}, [r0], r1
+    bx                  lr
+    @                 @ |vpx_h_predictor_16x16_neon|
+
+ @void vpx_h_predictor_32x32_neon(uint8_t *dst, ptrdiff_t y_stride,
+ @                                const uint8_t *above,
+ @                                const uint8_t *left)
+ @ r0  uint8_t *dst
+ @ r1  ptrdiff_t y_stride
+ @ r2  const uint8_t *above
+ @ r3  const uint8_t *left
+
+_vpx_h_predictor_32x32_neon:
+	vpx_h_predictor_32x32_neon: @
+    sub                 r1, r1, #16
+    mov                 r2, #2
+loop_h:
+    vld1.8              {q1}, [r3]!
+    vdup.8              q0, d2[0]
+    vst1.8              {q0}, [r0]!
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d2[1]
+    vst1.8              {q0}, [r0]!
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d2[2]
+    vst1.8              {q0}, [r0]!
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d2[3]
+    vst1.8              {q0}, [r0]!
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d2[4]
+    vst1.8              {q0}, [r0]!
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d2[5]
+    vst1.8              {q0}, [r0]!
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d2[6]
+    vst1.8              {q0}, [r0]!
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d2[7]
+    vst1.8              {q0}, [r0]!
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d3[0]
+    vst1.8              {q0}, [r0]!
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d3[1]
+    vst1.8              {q0}, [r0]!
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d3[2]
+    vst1.8              {q0}, [r0]!
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d3[3]
+    vst1.8              {q0}, [r0]!
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d3[4]
+    vst1.8              {q0}, [r0]!
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d3[5]
+    vst1.8              {q0}, [r0]!
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d3[6]
+    vst1.8              {q0}, [r0]!
+    vst1.8              {q0}, [r0], r1
+    vdup.8              q0, d3[7]
+    vst1.8              {q0}, [r0]!
+    vst1.8              {q0}, [r0], r1
+    subs                r2, r2, #1
+    bgt                 loop_h
+    bx                  lr
+    @                 @ |vpx_h_predictor_32x32_neon|
+
+ @void vpx_tm_predictor_4x4_neon (uint8_t *dst, ptrdiff_t y_stride,
+ @                                const uint8_t *above,
+ @                                const uint8_t *left)
+ @ r0  uint8_t *dst
+ @ r1  ptrdiff_t y_stride
+ @ r2  const uint8_t *above
+ @ r3  const uint8_t *left
+
+_vpx_tm_predictor_4x4_neon:
+	vpx_tm_predictor_4x4_neon: @
+     @ Load ytop_left = above[-1] @
+    sub                 r12, r2, #1
+    vld1.u8             {d0[]}, [r12]
+
+     @ Load above 4 pixels
+    vld1.32             {d2[0]}, [r2]
+
+     @ Compute above - ytop_left
+    vsubl.u8            q3, d2, d0
+
+     @ Load left row by row and compute left + (above - ytop_left)
+     @ 1st row and 2nd row
+    vld1.u8             {d2[]}, [r3]!
+    vld1.u8             {d4[]}, [r3]!
+    vmovl.u8            q1, d2
+    vmovl.u8            q2, d4
+    vadd.s16            q1, q1, q3
+    vadd.s16            q2, q2, q3
+    vqmovun.s16         d0, q1
+    vqmovun.s16         d1, q2
+    vst1.32             {d0[0]}, [r0], r1
+    vst1.32             {d1[0]}, [r0], r1
+
+     @ 3rd row and 4th row
+    vld1.u8             {d2[]}, [r3]!
+    vld1.u8             {d4[]}, [r3]
+    vmovl.u8            q1, d2
+    vmovl.u8            q2, d4
+    vadd.s16            q1, q1, q3
+    vadd.s16            q2, q2, q3
+    vqmovun.s16         d0, q1
+    vqmovun.s16         d1, q2
+    vst1.32             {d0[0]}, [r0], r1
+    vst1.32             {d1[0]}, [r0], r1
+    bx                  lr
+    @                 @ |vpx_tm_predictor_4x4_neon|
+
+ @void vpx_tm_predictor_8x8_neon (uint8_t *dst, ptrdiff_t y_stride,
+ @                                const uint8_t *above,
+ @                                const uint8_t *left)
+ @ r0  uint8_t *dst
+ @ r1  ptrdiff_t y_stride
+ @ r2  const uint8_t *above
+ @ r3  const uint8_t *left
+
+_vpx_tm_predictor_8x8_neon:
+	vpx_tm_predictor_8x8_neon: @
+     @ Load ytop_left = above[-1] @
+    sub                 r12, r2, #1
+    vld1.8              {d0[]}, [r12]
+
+     @ preload 8 left
+    vld1.8              {d30}, [r3]
+
+     @ Load above 8 pixels
+    vld1.64             {d2}, [r2]
+
+    vmovl.u8            q10, d30
+
+     @ Compute above - ytop_left
+    vsubl.u8            q3, d2, d0
+
+     @ Load left row by row and compute left + (above - ytop_left)
+     @ 1st row and 2nd row
+    vdup.16             q0, d20[0]
+    vdup.16             q1, d20[1]
+    vadd.s16            q0, q3, q0
+    vadd.s16            q1, q3, q1
+
+     @ 3rd row and 4th row
+    vdup.16             q8, d20[2]
+    vdup.16             q9, d20[3]
+    vadd.s16            q8, q3, q8
+    vadd.s16            q9, q3, q9
+
+    vqmovun.s16         d0, q0
+    vqmovun.s16         d1, q1
+    vqmovun.s16         d2, q8
+    vqmovun.s16         d3, q9
+
+    vst1.64             {d0}, [r0], r1
+    vst1.64             {d1}, [r0], r1
+    vst1.64             {d2}, [r0], r1
+    vst1.64             {d3}, [r0], r1
+
+     @ 5th row and 6th row
+    vdup.16             q0, d21[0]
+    vdup.16             q1, d21[1]
+    vadd.s16            q0, q3, q0
+    vadd.s16            q1, q3, q1
+
+     @ 7th row and 8th row
+    vdup.16             q8, d21[2]
+    vdup.16             q9, d21[3]
+    vadd.s16            q8, q3, q8
+    vadd.s16            q9, q3, q9
+
+    vqmovun.s16         d0, q0
+    vqmovun.s16         d1, q1
+    vqmovun.s16         d2, q8
+    vqmovun.s16         d3, q9
+
+    vst1.64             {d0}, [r0], r1
+    vst1.64             {d1}, [r0], r1
+    vst1.64             {d2}, [r0], r1
+    vst1.64             {d3}, [r0], r1
+
+    bx                  lr
+    @                 @ |vpx_tm_predictor_8x8_neon|
+
+ @void vpx_tm_predictor_16x16_neon (uint8_t *dst, ptrdiff_t y_stride,
+ @                                const uint8_t *above,
+ @                                const uint8_t *left)
+ @ r0  uint8_t *dst
+ @ r1  ptrdiff_t y_stride
+ @ r2  const uint8_t *above
+ @ r3  const uint8_t *left
+
+_vpx_tm_predictor_16x16_neon:
+	vpx_tm_predictor_16x16_neon: @
+     @ Load ytop_left = above[-1] @
+    sub                 r12, r2, #1
+    vld1.8              {d0[]}, [r12]
+
+     @ Load above 8 pixels
+    vld1.8              {q1}, [r2]
+
+     @ preload 8 left into r12
+    vld1.8              {d18}, [r3]!
+
+     @ Compute above - ytop_left
+    vsubl.u8            q2, d2, d0
+    vsubl.u8            q3, d3, d0
+
+    vmovl.u8            q10, d18
+
+     @ Load left row by row and compute left + (above - ytop_left)
+     @ Process 8 rows in each single loop and loop 2 times to process 16 rows.
+    mov                 r2, #2
+
+loop_16x16_neon:
+     @ Process two rows.
+    vdup.16             q0, d20[0]
+    vdup.16             q8, d20[1]
+    vadd.s16            q1, q0, q2
+    vadd.s16            q0, q0, q3
+    vadd.s16            q11, q8, q2
+    vadd.s16            q8, q8, q3
+    vqmovun.s16         d2, q1
+    vqmovun.s16         d3, q0
+    vqmovun.s16         d22, q11
+    vqmovun.s16         d23, q8
+    vdup.16             q0, d20[2]                   @ proload next 2 rows data
+    vdup.16             q8, d20[3]
+    vst1.64             {d2,d3}, [r0], r1
+    vst1.64             {d22,d23}, [r0], r1
+
+     @ Process two rows.
+    vadd.s16            q1, q0, q2
+    vadd.s16            q0, q0, q3
+    vadd.s16            q11, q8, q2
+    vadd.s16            q8, q8, q3
+    vqmovun.s16         d2, q1
+    vqmovun.s16         d3, q0
+    vqmovun.s16         d22, q11
+    vqmovun.s16         d23, q8
+    vdup.16             q0, d21[0]                   @ proload next 2 rows data
+    vdup.16             q8, d21[1]
+    vst1.64             {d2,d3}, [r0], r1
+    vst1.64             {d22,d23}, [r0], r1
+
+    vadd.s16            q1, q0, q2
+    vadd.s16            q0, q0, q3
+    vadd.s16            q11, q8, q2
+    vadd.s16            q8, q8, q3
+    vqmovun.s16         d2, q1
+    vqmovun.s16         d3, q0
+    vqmovun.s16         d22, q11
+    vqmovun.s16         d23, q8
+    vdup.16             q0, d21[2]                   @ proload next 2 rows data
+    vdup.16             q8, d21[3]
+    vst1.64             {d2,d3}, [r0], r1
+    vst1.64             {d22,d23}, [r0], r1
+
+
+    vadd.s16            q1, q0, q2
+    vadd.s16            q0, q0, q3
+    vadd.s16            q11, q8, q2
+    vadd.s16            q8, q8, q3
+    vqmovun.s16         d2, q1
+    vqmovun.s16         d3, q0
+    vqmovun.s16         d22, q11
+    vqmovun.s16         d23, q8
+    vld1.8              {d18}, [r3]!                   @ preload 8 left into r12
+    vmovl.u8            q10, d18
+    vst1.64             {d2,d3}, [r0], r1
+    vst1.64             {d22,d23}, [r0], r1
+
+    subs                r2, r2, #1
+    bgt                 loop_16x16_neon
+
+    bx                  lr
+    @                 @ |vpx_tm_predictor_16x16_neon|
+
+ @void vpx_tm_predictor_32x32_neon (uint8_t *dst, ptrdiff_t y_stride,
+ @                                  const uint8_t *above,
+ @                                  const uint8_t *left)
+ @ r0  uint8_t *dst
+ @ r1  ptrdiff_t y_stride
+ @ r2  const uint8_t *above
+ @ r3  const uint8_t *left
+
+_vpx_tm_predictor_32x32_neon:
+	vpx_tm_predictor_32x32_neon: @
+     @ Load ytop_left = above[-1] @
+    sub                 r12, r2, #1
+    vld1.8              {d0[]}, [r12]
+
+     @ Load above 32 pixels
+    vld1.8              {q1}, [r2]!
+    vld1.8              {q2}, [r2]
+
+     @ preload 8 left pixels
+    vld1.8              {d26}, [r3]!
+
+     @ Compute above - ytop_left
+    vsubl.u8            q8, d2, d0
+    vsubl.u8            q9, d3, d0
+    vsubl.u8            q10, d4, d0
+    vsubl.u8            q11, d5, d0
+
+    vmovl.u8            q3, d26
+
+     @ Load left row by row and compute left + (above - ytop_left)
+     @ Process 8 rows in each single loop and loop 4 times to process 32 rows.
+    mov                 r2, #4
+
+loop_32x32_neon:
+     @ Process two rows.
+    vdup.16             q0, d6[0]
+    vdup.16             q2, d6[1]
+    vadd.s16            q12, q0, q8
+    vadd.s16            q13, q0, q9
+    vadd.s16            q14, q0, q10
+    vadd.s16            q15, q0, q11
+    vqmovun.s16         d0, q12
+    vqmovun.s16         d1, q13
+    vadd.s16            q12, q2, q8
+    vadd.s16            q13, q2, q9
+    vqmovun.s16         d2, q14
+    vqmovun.s16         d3, q15
+    vadd.s16            q14, q2, q10
+    vadd.s16            q15, q2, q11
+    vst1.64             {d0-d3}, [r0], r1
+    vqmovun.s16         d24, q12
+    vqmovun.s16         d25, q13
+    vqmovun.s16         d26, q14
+    vqmovun.s16         d27, q15
+    vdup.16             q1, d6[2]
+    vdup.16             q2, d6[3]
+    vst1.64             {d24-d27}, [r0], r1
+
+     @ Process two rows.
+    vadd.s16            q12, q1, q8
+    vadd.s16            q13, q1, q9
+    vadd.s16            q14, q1, q10
+    vadd.s16            q15, q1, q11
+    vqmovun.s16         d0, q12
+    vqmovun.s16         d1, q13
+    vadd.s16            q12, q2, q8
+    vadd.s16            q13, q2, q9
+    vqmovun.s16         d2, q14
+    vqmovun.s16         d3, q15
+    vadd.s16            q14, q2, q10
+    vadd.s16            q15, q2, q11
+    vst1.64             {d0-d3}, [r0], r1
+    vqmovun.s16         d24, q12
+    vqmovun.s16         d25, q13
+    vqmovun.s16         d26, q14
+    vqmovun.s16         d27, q15
+    vdup.16             q0, d7[0]
+    vdup.16             q2, d7[1]
+    vst1.64             {d24-d27}, [r0], r1
+
+     @ Process two rows.
+    vadd.s16            q12, q0, q8
+    vadd.s16            q13, q0, q9
+    vadd.s16            q14, q0, q10
+    vadd.s16            q15, q0, q11
+    vqmovun.s16         d0, q12
+    vqmovun.s16         d1, q13
+    vadd.s16            q12, q2, q8
+    vadd.s16            q13, q2, q9
+    vqmovun.s16         d2, q14
+    vqmovun.s16         d3, q15
+    vadd.s16            q14, q2, q10
+    vadd.s16            q15, q2, q11
+    vst1.64             {d0-d3}, [r0], r1
+    vqmovun.s16         d24, q12
+    vqmovun.s16         d25, q13
+    vqmovun.s16         d26, q14
+    vqmovun.s16         d27, q15
+    vdup.16             q0, d7[2]
+    vdup.16             q2, d7[3]
+    vst1.64             {d24-d27}, [r0], r1
+
+     @ Process two rows.
+    vadd.s16            q12, q0, q8
+    vadd.s16            q13, q0, q9
+    vadd.s16            q14, q0, q10
+    vadd.s16            q15, q0, q11
+    vqmovun.s16         d0, q12
+    vqmovun.s16         d1, q13
+    vadd.s16            q12, q2, q8
+    vadd.s16            q13, q2, q9
+    vqmovun.s16         d2, q14
+    vqmovun.s16         d3, q15
+    vadd.s16            q14, q2, q10
+    vadd.s16            q15, q2, q11
+    vst1.64             {d0-d3}, [r0], r1
+    vqmovun.s16         d24, q12
+    vqmovun.s16         d25, q13
+    vld1.8              {d0}, [r3]!                    @ preload 8 left pixels
+    vqmovun.s16         d26, q14
+    vqmovun.s16         d27, q15
+    vmovl.u8            q3, d0
+    vst1.64             {d24-d27}, [r0], r1
+
+    subs                r2, r2, #1
+    bgt                 loop_32x32_neon
+
+    bx                  lr
+    @                 @ |vpx_tm_predictor_32x32_neon|
+

thirdparty/libvpx/vpx_dsp/arm/gas_apple/loopfilter_mb_neon.s

@@ -0,0 +1,649 @@
+@ This file was created from a .asm file
+@  using the ads2gas_apple.pl script.
+
+	.set WIDE_REFERENCE, 0
+	.set ARCHITECTURE, 5
+	.set DO1STROUNDING, 0
+ @
+ @  Copyright (c) 2013 The WebM project authors. All Rights Reserved.
+ @
+ @  Use of this source code is governed by a BSD-style license
+ @  that can be found in the LICENSE file in the root of the source
+ @  tree. An additional intellectual property rights grant can be found
+ @  in the file PATENTS.  All contributing project authors may
+ @  be found in the AUTHORS file in the root of the source tree.
+ @
+
+    .globl _vpx_lpf_horizontal_edge_8_neon
+	.globl vpx_lpf_horizontal_edge_8_neon
+    .globl _vpx_lpf_horizontal_edge_16_neon
+	.globl vpx_lpf_horizontal_edge_16_neon
+    .globl _vpx_lpf_vertical_16_neon
+	.globl vpx_lpf_vertical_16_neon
+   @ ARM
+
+.text
+.p2align 2
+
+ @ void mb_lpf_horizontal_edge(uint8_t *s, int p,
+ @                             const uint8_t *blimit,
+ @                             const uint8_t *limit,
+ @                             const uint8_t *thresh,
+ @                             int count)
+ @ r0    uint8_t *s,
+ @ r1    int p, /* pitch */
+ @ r2    const uint8_t *blimit,
+ @ r3    const uint8_t *limit,
+ @ sp    const uint8_t *thresh,
+ @ r12   int count
+_mb_lpf_horizontal_edge:
+	mb_lpf_horizontal_edge: @
+    push        {r4-r8, lr}
+    vpush       {d8-d15}
+    ldr         r4, [sp, #88]               @ load thresh
+
+h_count:
+    vld1.8      {d16[]}, [r2]               @ load *blimit
+    vld1.8      {d17[]}, [r3]               @ load *limit
+    vld1.8      {d18[]}, [r4]               @ load *thresh
+
+    sub         r8, r0, r1, lsl #3          @ move src pointer down by 8 lines
+
+    vld1.u8     {d0}, [r8,:64], r1           @ p7
+    vld1.u8     {d1}, [r8,:64], r1           @ p6
+    vld1.u8     {d2}, [r8,:64], r1           @ p5
+    vld1.u8     {d3}, [r8,:64], r1           @ p4
+    vld1.u8     {d4}, [r8,:64], r1           @ p3
+    vld1.u8     {d5}, [r8,:64], r1           @ p2
+    vld1.u8     {d6}, [r8,:64], r1           @ p1
+    vld1.u8     {d7}, [r8,:64], r1           @ p0
+    vld1.u8     {d8}, [r8,:64], r1           @ q0
+    vld1.u8     {d9}, [r8,:64], r1           @ q1
+    vld1.u8     {d10}, [r8,:64], r1          @ q2
+    vld1.u8     {d11}, [r8,:64], r1          @ q3
+    vld1.u8     {d12}, [r8,:64], r1          @ q4
+    vld1.u8     {d13}, [r8,:64], r1          @ q5
+    vld1.u8     {d14}, [r8,:64], r1          @ q6
+    vld1.u8     {d15}, [r8,:64], r1          @ q7
+
+    bl          vpx_wide_mbfilter_neon
+
+    tst         r7, #1
+    beq         h_mbfilter
+
+     @ flat && mask were not set for any of the channels. Just store the values
+     @ from filter.
+    sub         r8, r0, r1, lsl #1
+
+    vst1.u8     {d25}, [r8,:64], r1          @ store op1
+    vst1.u8     {d24}, [r8,:64], r1          @ store op0
+    vst1.u8     {d23}, [r8,:64], r1          @ store oq0
+    vst1.u8     {d26}, [r8,:64], r1          @ store oq1
+
+    b           h_next
+
+h_mbfilter:
+    tst         r7, #2
+    beq         h_wide_mbfilter
+
+     @ flat2 was not set for any of the channels. Just store the values from
+     @ mbfilter.
+    sub         r8, r0, r1, lsl #1
+    sub         r8, r8, r1
+
+    vst1.u8     {d18}, [r8,:64], r1          @ store op2
+    vst1.u8     {d19}, [r8,:64], r1          @ store op1
+    vst1.u8     {d20}, [r8,:64], r1          @ store op0
+    vst1.u8     {d21}, [r8,:64], r1          @ store oq0
+    vst1.u8     {d22}, [r8,:64], r1          @ store oq1
+    vst1.u8     {d23}, [r8,:64], r1          @ store oq2
+
+    b           h_next
+
+h_wide_mbfilter:
+    sub         r8, r0, r1, lsl #3
+    add         r8, r8, r1
+
+    vst1.u8     {d16}, [r8,:64], r1          @ store op6
+    vst1.u8     {d24}, [r8,:64], r1          @ store op5
+    vst1.u8     {d25}, [r8,:64], r1          @ store op4
+    vst1.u8     {d26}, [r8,:64], r1          @ store op3
+    vst1.u8     {d27}, [r8,:64], r1          @ store op2
+    vst1.u8     {d18}, [r8,:64], r1          @ store op1
+    vst1.u8     {d19}, [r8,:64], r1          @ store op0
+    vst1.u8     {d20}, [r8,:64], r1          @ store oq0
+    vst1.u8     {d21}, [r8,:64], r1          @ store oq1
+    vst1.u8     {d22}, [r8,:64], r1          @ store oq2
+    vst1.u8     {d23}, [r8,:64], r1          @ store oq3
+    vst1.u8     {d1}, [r8,:64], r1           @ store oq4
+    vst1.u8     {d2}, [r8,:64], r1           @ store oq5
+    vst1.u8     {d3}, [r8,:64], r1           @ store oq6
+
+h_next:
+    add         r0, r0, #8
+    subs        r12, r12, #1
+    bne         h_count
+
+    vpop        {d8-d15}
+    pop         {r4-r8, pc}
+
+    @         @ |mb_lpf_horizontal_edge|
+
+ @ void vpx_lpf_horizontal_edge_8_neon(uint8_t *s, int pitch,
+ @                                     const uint8_t *blimit,
+ @                                     const uint8_t *limit,
+ @                                     const uint8_t *thresh)
+ @ r0    uint8_t *s,
+ @ r1    int pitch,
+ @ r2    const uint8_t *blimit,
+ @ r3    const uint8_t *limit,
+ @ sp    const uint8_t *thresh
+_vpx_lpf_horizontal_edge_8_neon:
+	vpx_lpf_horizontal_edge_8_neon: @
+    mov r12, #1
+    b mb_lpf_horizontal_edge
+    @         @ |vpx_lpf_horizontal_edge_8_neon|
+
+ @ void vpx_lpf_horizontal_edge_16_neon(uint8_t *s, int pitch,
+ @                                      const uint8_t *blimit,
+ @                                      const uint8_t *limit,
+ @                                      const uint8_t *thresh)
+ @ r0    uint8_t *s,
+ @ r1    int pitch,
+ @ r2    const uint8_t *blimit,
+ @ r3    const uint8_t *limit,
+ @ sp    const uint8_t *thresh
+_vpx_lpf_horizontal_edge_16_neon:
+	vpx_lpf_horizontal_edge_16_neon: @
+    mov r12, #2
+    b mb_lpf_horizontal_edge
+    @         @ |vpx_lpf_horizontal_edge_16_neon|
+
+ @ void vpx_lpf_vertical_16_neon(uint8_t *s, int p,
+ @                               const uint8_t *blimit,
+ @                               const uint8_t *limit,
+ @                               const uint8_t *thresh)
+ @ r0    uint8_t *s,
+ @ r1    int p, /* pitch */
+ @ r2    const uint8_t *blimit,
+ @ r3    const uint8_t *limit,
+ @ sp    const uint8_t *thresh,
+_vpx_lpf_vertical_16_neon:
+	vpx_lpf_vertical_16_neon: @
+    push        {r4-r8, lr}
+    vpush       {d8-d15}
+    ldr         r4, [sp, #88]               @ load thresh
+
+    vld1.8      {d16[]}, [r2]               @ load *blimit
+    vld1.8      {d17[]}, [r3]               @ load *limit
+    vld1.8      {d18[]}, [r4]               @ load *thresh
+
+    sub         r8, r0, #8
+
+    vld1.8      {d0}, [r8,:64], r1
+    vld1.8      {d8}, [r0,:64], r1
+    vld1.8      {d1}, [r8,:64], r1
+    vld1.8      {d9}, [r0,:64], r1
+    vld1.8      {d2}, [r8,:64], r1
+    vld1.8      {d10}, [r0,:64], r1
+    vld1.8      {d3}, [r8,:64], r1
+    vld1.8      {d11}, [r0,:64], r1
+    vld1.8      {d4}, [r8,:64], r1
+    vld1.8      {d12}, [r0,:64], r1
+    vld1.8      {d5}, [r8,:64], r1
+    vld1.8      {d13}, [r0,:64], r1
+    vld1.8      {d6}, [r8,:64], r1
+    vld1.8      {d14}, [r0,:64], r1
+    vld1.8      {d7}, [r8,:64], r1
+    vld1.8      {d15}, [r0,:64], r1
+
+    sub         r0, r0, r1, lsl #3
+
+    vtrn.32     q0, q2
+    vtrn.32     q1, q3
+    vtrn.32     q4, q6
+    vtrn.32     q5, q7
+
+    vtrn.16     q0, q1
+    vtrn.16     q2, q3
+    vtrn.16     q4, q5
+    vtrn.16     q6, q7
+
+    vtrn.8      d0, d1
+    vtrn.8      d2, d3
+    vtrn.8      d4, d5
+    vtrn.8      d6, d7
+
+    vtrn.8      d8, d9
+    vtrn.8      d10, d11
+    vtrn.8      d12, d13
+    vtrn.8      d14, d15
+
+    bl          vpx_wide_mbfilter_neon
+
+    tst         r7, #1
+    beq         v_mbfilter
+
+     @ flat && mask were not set for any of the channels. Just store the values
+     @ from filter.
+    sub         r8, r0, #2
+
+    vswp        d23, d25
+
+    vst4.8      {d23[0], d24[0], d25[0], d26[0]}, [r8], r1
+    vst4.8      {d23[1], d24[1], d25[1], d26[1]}, [r8], r1
+    vst4.8      {d23[2], d24[2], d25[2], d26[2]}, [r8], r1
+    vst4.8      {d23[3], d24[3], d25[3], d26[3]}, [r8], r1
+    vst4.8      {d23[4], d24[4], d25[4], d26[4]}, [r8], r1
+    vst4.8      {d23[5], d24[5], d25[5], d26[5]}, [r8], r1
+    vst4.8      {d23[6], d24[6], d25[6], d26[6]}, [r8], r1
+    vst4.8      {d23[7], d24[7], d25[7], d26[7]}, [r8], r1
+
+    b           v_end
+
+v_mbfilter:
+    tst         r7, #2
+    beq         v_wide_mbfilter
+
+     @ flat2 was not set for any of the channels. Just store the values from
+     @ mbfilter.
+    sub         r8, r0, #3
+
+    vst3.8      {d18[0], d19[0], d20[0]}, [r8], r1
+    vst3.8      {d21[0], d22[0], d23[0]}, [r0], r1
+    vst3.8      {d18[1], d19[1], d20[1]}, [r8], r1
+    vst3.8      {d21[1], d22[1], d23[1]}, [r0], r1
+    vst3.8      {d18[2], d19[2], d20[2]}, [r8], r1
+    vst3.8      {d21[2], d22[2], d23[2]}, [r0], r1
+    vst3.8      {d18[3], d19[3], d20[3]}, [r8], r1
+    vst3.8      {d21[3], d22[3], d23[3]}, [r0], r1
+    vst3.8      {d18[4], d19[4], d20[4]}, [r8], r1
+    vst3.8      {d21[4], d22[4], d23[4]}, [r0], r1
+    vst3.8      {d18[5], d19[5], d20[5]}, [r8], r1
+    vst3.8      {d21[5], d22[5], d23[5]}, [r0], r1
+    vst3.8      {d18[6], d19[6], d20[6]}, [r8], r1
+    vst3.8      {d21[6], d22[6], d23[6]}, [r0], r1
+    vst3.8      {d18[7], d19[7], d20[7]}, [r8], r1
+    vst3.8      {d21[7], d22[7], d23[7]}, [r0], r1
+
+    b           v_end
+
+v_wide_mbfilter:
+    sub         r8, r0, #8
+
+    vtrn.32     d0,  d26
+    vtrn.32     d16, d27
+    vtrn.32     d24, d18
+    vtrn.32     d25, d19
+
+    vtrn.16     d0,  d24
+    vtrn.16     d16, d25
+    vtrn.16     d26, d18
+    vtrn.16     d27, d19
+
+    vtrn.8      d0,  d16
+    vtrn.8      d24, d25
+    vtrn.8      d26, d27
+    vtrn.8      d18, d19
+
+    vtrn.32     d20, d1
+    vtrn.32     d21, d2
+    vtrn.32     d22, d3
+    vtrn.32     d23, d15
+
+    vtrn.16     d20, d22
+    vtrn.16     d21, d23
+    vtrn.16     d1,  d3
+    vtrn.16     d2,  d15
+
+    vtrn.8      d20, d21
+    vtrn.8      d22, d23
+    vtrn.8      d1,  d2
+    vtrn.8      d3,  d15
+
+    vst1.8      {d0}, [r8,:64], r1
+    vst1.8      {d20}, [r0,:64], r1
+    vst1.8      {d16}, [r8,:64], r1
+    vst1.8      {d21}, [r0,:64], r1
+    vst1.8      {d24}, [r8,:64], r1
+    vst1.8      {d22}, [r0,:64], r1
+    vst1.8      {d25}, [r8,:64], r1
+    vst1.8      {d23}, [r0,:64], r1
+    vst1.8      {d26}, [r8,:64], r1
+    vst1.8      {d1}, [r0,:64], r1
+    vst1.8      {d27}, [r8,:64], r1
+    vst1.8      {d2}, [r0,:64], r1
+    vst1.8      {d18}, [r8,:64], r1
+    vst1.8      {d3}, [r0,:64], r1
+    vst1.8      {d19}, [r8,:64], r1
+    vst1.8      {d15}, [r0,:64], r1
+
+v_end:
+    vpop        {d8-d15}
+    pop         {r4-r8, pc}
+
+    @         @ |vpx_lpf_vertical_16_neon|
+
+ @ void vpx_wide_mbfilter_neon() @
+ @ This is a helper function for the loopfilters. The invidual functions do the
+ @ necessary load, transpose (if necessary) and store.
+ @
+ @ r0-r3 PRESERVE
+ @ d16    blimit
+ @ d17    limit
+ @ d18    thresh
+ @ d0    p7
+ @ d1    p6
+ @ d2    p5
+ @ d3    p4
+ @ d4    p3
+ @ d5    p2
+ @ d6    p1
+ @ d7    p0
+ @ d8    q0
+ @ d9    q1
+ @ d10   q2
+ @ d11   q3
+ @ d12   q4
+ @ d13   q5
+ @ d14   q6
+ @ d15   q7
+_vpx_wide_mbfilter_neon:
+	vpx_wide_mbfilter_neon: @
+    mov         r7, #0
+
+     @ filter_mask
+    vabd.u8     d19, d4, d5                 @ abs(p3 - p2)
+    vabd.u8     d20, d5, d6                 @ abs(p2 - p1)
+    vabd.u8     d21, d6, d7                 @ abs(p1 - p0)
+    vabd.u8     d22, d9, d8                 @ abs(q1 - q0)
+    vabd.u8     d23, d10, d9                @ abs(q2 - q1)
+    vabd.u8     d24, d11, d10               @ abs(q3 - q2)
+
+     @ only compare the largest value to limit
+    vmax.u8     d19, d19, d20               @ max(abs(p3 - p2), abs(p2 - p1))
+    vmax.u8     d20, d21, d22               @ max(abs(p1 - p0), abs(q1 - q0))
+    vmax.u8     d23, d23, d24               @ max(abs(q2 - q1), abs(q3 - q2))
+    vmax.u8     d19, d19, d20
+
+    vabd.u8     d24, d7, d8                 @ abs(p0 - q0)
+
+    vmax.u8     d19, d19, d23
+
+    vabd.u8     d23, d6, d9                 @ a = abs(p1 - q1)
+    vqadd.u8    d24, d24, d24               @ b = abs(p0 - q0) * 2
+
+     @ abs () > limit
+    vcge.u8     d19, d17, d19
+
+     @ flatmask4
+    vabd.u8     d25, d7, d5                 @ abs(p0 - p2)
+    vabd.u8     d26, d8, d10                @ abs(q0 - q2)
+    vabd.u8     d27, d4, d7                 @ abs(p3 - p0)
+    vabd.u8     d28, d11, d8                @ abs(q3 - q0)
+
+     @ only compare the largest value to thresh
+    vmax.u8     d25, d25, d26               @ max(abs(p0 - p2), abs(q0 - q2))
+    vmax.u8     d26, d27, d28               @ max(abs(p3 - p0), abs(q3 - q0))
+    vmax.u8     d25, d25, d26
+    vmax.u8     d20, d20, d25
+
+    vshr.u8     d23, d23, #1                @ a = a / 2
+    vqadd.u8    d24, d24, d23               @ a = b + a
+
+    vmov.u8     d30, #1
+    vcge.u8     d24, d16, d24               @ (a > blimit * 2 + limit) * -1
+
+    vcge.u8     d20, d30, d20               @ flat
+
+    vand        d19, d19, d24               @ mask
+
+     @ hevmask
+    vcgt.u8     d21, d21, d18               @ (abs(p1 - p0) > thresh)*-1
+    vcgt.u8     d22, d22, d18               @ (abs(q1 - q0) > thresh)*-1
+    vorr        d21, d21, d22               @ hev
+
+    vand        d16, d20, d19               @ flat && mask
+    vmov        r5, r6, d16
+
+     @ flatmask5(1, p7, p6, p5, p4, p0, q0, q4, q5, q6, q7)
+    vabd.u8     d22, d3, d7                 @ abs(p4 - p0)
+    vabd.u8     d23, d12, d8                @ abs(q4 - q0)
+    vabd.u8     d24, d7, d2                 @ abs(p0 - p5)
+    vabd.u8     d25, d8, d13                @ abs(q0 - q5)
+    vabd.u8     d26, d1, d7                 @ abs(p6 - p0)
+    vabd.u8     d27, d14, d8                @ abs(q6 - q0)
+    vabd.u8     d28, d0, d7                 @ abs(p7 - p0)
+    vabd.u8     d29, d15, d8                @ abs(q7 - q0)
+
+     @ only compare the largest value to thresh
+    vmax.u8     d22, d22, d23               @ max(abs(p4 - p0), abs(q4 - q0))
+    vmax.u8     d23, d24, d25               @ max(abs(p0 - p5), abs(q0 - q5))
+    vmax.u8     d24, d26, d27               @ max(abs(p6 - p0), abs(q6 - q0))
+    vmax.u8     d25, d28, d29               @ max(abs(p7 - p0), abs(q7 - q0))
+
+    vmax.u8     d26, d22, d23
+    vmax.u8     d27, d24, d25
+    vmax.u8     d23, d26, d27
+
+    vcge.u8     d18, d30, d23               @ flat2
+
+    vmov.u8     d22, #0x80
+
+    orrs        r5, r5, r6                  @ Check for 0
+    orreq       r7, r7, #1                  @ Only do filter branch
+
+    vand        d17, d18, d16               @ flat2 && flat && mask
+    vmov        r5, r6, d17
+
+     @ mbfilter() function
+
+     @ filter() function
+     @ convert to signed
+    veor        d23, d8, d22                @ qs0
+    veor        d24, d7, d22                @ ps0
+    veor        d25, d6, d22                @ ps1
+    veor        d26, d9, d22                @ qs1
+
+    vmov.u8     d27, #3
+
+    vsub.s8     d28, d23, d24               @ ( qs0 - ps0)
+    vqsub.s8    d29, d25, d26               @ filter = clamp(ps1-qs1)
+    vmull.s8    q15, d28, d27               @ 3 * ( qs0 - ps0)
+    vand        d29, d29, d21               @ filter &= hev
+    vaddw.s8    q15, q15, d29               @ filter + 3 * (qs0 - ps0)
+    vmov.u8     d29, #4
+
+     @ filter = clamp(filter + 3 * ( qs0 - ps0))
+    vqmovn.s16  d28, q15
+
+    vand        d28, d28, d19               @ filter &= mask
+
+    vqadd.s8    d30, d28, d27               @ filter2 = clamp(filter+3)
+    vqadd.s8    d29, d28, d29               @ filter1 = clamp(filter+4)
+    vshr.s8     d30, d30, #3                @ filter2 >>= 3
+    vshr.s8     d29, d29, #3                @ filter1 >>= 3
+
+
+    vqadd.s8    d24, d24, d30               @ op0 = clamp(ps0 + filter2)
+    vqsub.s8    d23, d23, d29               @ oq0 = clamp(qs0 - filter1)
+
+     @ outer tap adjustments: ++filter1 >> 1
+    vrshr.s8    d29, d29, #1
+    vbic        d29, d29, d21               @ filter &= ~hev
+
+    vqadd.s8    d25, d25, d29               @ op1 = clamp(ps1 + filter)
+    vqsub.s8    d26, d26, d29               @ oq1 = clamp(qs1 - filter)
+
+    veor        d24, d24, d22               @ *f_op0 = u^0x80
+    veor        d23, d23, d22               @ *f_oq0 = u^0x80
+    veor        d25, d25, d22               @ *f_op1 = u^0x80
+    veor        d26, d26, d22               @ *f_oq1 = u^0x80
+
+    tst         r7, #1
+    bxne        lr
+
+    orrs        r5, r5, r6                  @ Check for 0
+    orreq       r7, r7, #2                  @ Only do mbfilter branch
+
+     @ mbfilter flat && mask branch
+     @ TODO(fgalligan): Can I decrease the cycles shifting to consective d's
+     @ and using vibt on the q's?
+    vmov.u8     d29, #2
+    vaddl.u8    q15, d7, d8                 @ op2 = p0 + q0
+    vmlal.u8    q15, d4, d27                @ op2 = p0 + q0 + p3 * 3
+    vmlal.u8    q15, d5, d29                @ op2 = p0 + q0 + p3 * 3 + p2 * 2
+    vaddl.u8    q10, d4, d5
+    vaddw.u8    q15, d6                     @ op2=p1 + p0 + q0 + p3 * 3 + p2 *2
+    vaddl.u8    q14, d6, d9
+    vqrshrn.u16 d18, q15, #3                @ r_op2
+
+    vsub.i16    q15, q10
+    vaddl.u8    q10, d4, d6
+    vadd.i16    q15, q14
+    vaddl.u8    q14, d7, d10
+    vqrshrn.u16 d19, q15, #3                @ r_op1
+
+    vsub.i16    q15, q10
+    vadd.i16    q15, q14
+    vaddl.u8    q14, d8, d11
+    vqrshrn.u16 d20, q15, #3                @ r_op0
+
+    vsubw.u8    q15, d4                     @ oq0 = op0 - p3
+    vsubw.u8    q15, d7                     @ oq0 -= p0
+    vadd.i16    q15, q14
+    vaddl.u8    q14, d9, d11
+    vqrshrn.u16 d21, q15, #3                @ r_oq0
+
+    vsubw.u8    q15, d5                     @ oq1 = oq0 - p2
+    vsubw.u8    q15, d8                     @ oq1 -= q0
+    vadd.i16    q15, q14
+    vaddl.u8    q14, d10, d11
+    vqrshrn.u16 d22, q15, #3                @ r_oq1
+
+    vsubw.u8    q15, d6                     @ oq2 = oq0 - p1
+    vsubw.u8    q15, d9                     @ oq2 -= q1
+    vadd.i16    q15, q14
+    vqrshrn.u16 d27, q15, #3                @ r_oq2
+
+     @ Filter does not set op2 or oq2, so use p2 and q2.
+    vbif        d18, d5, d16                @ t_op2 |= p2 & ~(flat & mask)
+    vbif        d19, d25, d16               @ t_op1 |= f_op1 & ~(flat & mask)
+    vbif        d20, d24, d16               @ t_op0 |= f_op0 & ~(flat & mask)
+    vbif        d21, d23, d16               @ t_oq0 |= f_oq0 & ~(flat & mask)
+    vbif        d22, d26, d16               @ t_oq1 |= f_oq1 & ~(flat & mask)
+
+    vbit        d23, d27, d16               @ t_oq2 |= r_oq2 & (flat & mask)
+    vbif        d23, d10, d16               @ t_oq2 |= q2 & ~(flat & mask)
+
+    tst         r7, #2
+    bxne        lr
+
+     @ wide_mbfilter flat2 && flat && mask branch
+    vmov.u8     d16, #7
+    vaddl.u8    q15, d7, d8                 @ op6 = p0 + q0
+    vaddl.u8    q12, d2, d3
+    vaddl.u8    q13, d4, d5
+    vaddl.u8    q14, d1, d6
+    vmlal.u8    q15, d0, d16                @ op6 += p7 * 3
+    vadd.i16    q12, q13
+    vadd.i16    q15, q14
+    vaddl.u8    q14, d2, d9
+    vadd.i16    q15, q12
+    vaddl.u8    q12, d0, d1
+    vaddw.u8    q15, d1
+    vaddl.u8    q13, d0, d2
+    vadd.i16    q14, q15, q14
+    vqrshrn.u16 d16, q15, #4                @ w_op6
+
+    vsub.i16    q15, q14, q12
+    vaddl.u8    q14, d3, d10
+    vqrshrn.u16 d24, q15, #4                @ w_op5
+
+    vsub.i16    q15, q13
+    vaddl.u8    q13, d0, d3
+    vadd.i16    q15, q14
+    vaddl.u8    q14, d4, d11
+    vqrshrn.u16 d25, q15, #4                @ w_op4
+
+    vadd.i16    q15, q14
+    vaddl.u8    q14, d0, d4
+    vsub.i16    q15, q13
+    vsub.i16    q14, q15, q14
+    vqrshrn.u16 d26, q15, #4                @ w_op3
+
+    vaddw.u8    q15, q14, d5                @ op2 += p2
+    vaddl.u8    q14, d0, d5
+    vaddw.u8    q15, d12                    @ op2 += q4
+    vbif        d26, d4, d17                @ op3 |= p3 & ~(f2 & f & m)
+    vqrshrn.u16 d27, q15, #4                @ w_op2
+
+    vsub.i16    q15, q14
+    vaddl.u8    q14, d0, d6
+    vaddw.u8    q15, d6                     @ op1 += p1
+    vaddw.u8    q15, d13                    @ op1 += q5
+    vbif        d27, d18, d17               @ op2 |= t_op2 & ~(f2 & f & m)
+    vqrshrn.u16 d18, q15, #4                @ w_op1
+
+    vsub.i16    q15, q14
+    vaddl.u8    q14, d0, d7
+    vaddw.u8    q15, d7                     @ op0 += p0
+    vaddw.u8    q15, d14                    @ op0 += q6
+    vbif        d18, d19, d17               @ op1 |= t_op1 & ~(f2 & f & m)
+    vqrshrn.u16 d19, q15, #4                @ w_op0
+
+    vsub.i16    q15, q14
+    vaddl.u8    q14, d1, d8
+    vaddw.u8    q15, d8                     @ oq0 += q0
+    vaddw.u8    q15, d15                    @ oq0 += q7
+    vbif        d19, d20, d17               @ op0 |= t_op0 & ~(f2 & f & m)
+    vqrshrn.u16 d20, q15, #4                @ w_oq0
+
+    vsub.i16    q15, q14
+    vaddl.u8    q14, d2, d9
+    vaddw.u8    q15, d9                     @ oq1 += q1
+    vaddl.u8    q4, d10, d15
+    vaddw.u8    q15, d15                    @ oq1 += q7
+    vbif        d20, d21, d17               @ oq0 |= t_oq0 & ~(f2 & f & m)
+    vqrshrn.u16 d21, q15, #4                @ w_oq1
+
+    vsub.i16    q15, q14
+    vaddl.u8    q14, d3, d10
+    vadd.i16    q15, q4
+    vaddl.u8    q4, d11, d15
+    vbif        d21, d22, d17               @ oq1 |= t_oq1 & ~(f2 & f & m)
+    vqrshrn.u16 d22, q15, #4                @ w_oq2
+
+    vsub.i16    q15, q14
+    vaddl.u8    q14, d4, d11
+    vadd.i16    q15, q4
+    vaddl.u8    q4, d12, d15
+    vbif        d22, d23, d17               @ oq2 |= t_oq2 & ~(f2 & f & m)
+    vqrshrn.u16 d23, q15, #4                @ w_oq3
+
+    vsub.i16    q15, q14
+    vaddl.u8    q14, d5, d12
+    vadd.i16    q15, q4
+    vaddl.u8    q4, d13, d15
+    vbif        d16, d1, d17                @ op6 |= p6 & ~(f2 & f & m)
+    vqrshrn.u16 d1, q15, #4                 @ w_oq4
+
+    vsub.i16    q15, q14
+    vaddl.u8    q14, d6, d13
+    vadd.i16    q15, q4
+    vaddl.u8    q4, d14, d15
+    vbif        d24, d2, d17                @ op5 |= p5 & ~(f2 & f & m)
+    vqrshrn.u16 d2, q15, #4                 @ w_oq5
+
+    vsub.i16    q15, q14
+    vbif        d25, d3, d17                @ op4 |= p4 & ~(f2 & f & m)
+    vadd.i16    q15, q4
+    vbif        d23, d11, d17               @ oq3 |= q3 & ~(f2 & f & m)
+    vqrshrn.u16 d3, q15, #4                 @ w_oq6
+    vbif        d1, d12, d17                @ oq4 |= q4 & ~(f2 & f & m)
+    vbif        d2, d13, d17                @ oq5 |= q5 & ~(f2 & f & m)
+    vbif        d3, d14, d17                @ oq6 |= q6 & ~(f2 & f & m)
+
+    bx          lr
+    @         @ |vpx_wide_mbfilter_neon|
+

thirdparty/libvpx/vpx_dsp/arm/gas_apple/save_reg_neon.s

@@ -0,0 +1,46 @@
+@ This file was created from a .asm file
+@  using the ads2gas_apple.pl script.
+
+	.set WIDE_REFERENCE, 0
+	.set ARCHITECTURE, 5
+	.set DO1STROUNDING, 0
+ @
+ @  Copyright (c) 2010 The WebM project authors. All Rights Reserved.
+ @
+ @  Use of this source code is governed by a BSD-style license
+ @  that can be found in the LICENSE file in the root of the source
+ @  tree. An additional intellectual property rights grant can be found
+ @  in the file PATENTS.  All contributing project authors may
+ @  be found in the AUTHORS file in the root of the source tree.
+ @
+
+
+    .globl _vpx_push_neon
+	.globl vpx_push_neon
+    .globl _vpx_pop_neon
+	.globl vpx_pop_neon
+
+   @ ARM
+   @ 
+   @ PRESERVE8
+
+.text
+.p2align 2
+
+_vpx_push_neon:
+	vpx_push_neon: @
+    vst1.i64            {d8, d9, d10, d11}, [r0]!
+    vst1.i64            {d12, d13, d14, d15}, [r0]!
+    bx              lr
+
+    @
+
+_vpx_pop_neon:
+	vpx_pop_neon: @
+    vld1.i64            {d8, d9, d10, d11}, [r0]!
+    vld1.i64            {d12, d13, d14, d15}, [r0]!
+    bx              lr
+
+    @
+
+

thirdparty/libvpx/vpx_dsp_rtcd.h

@@ -0,0 +1,9 @@
+#include "vpx_config.h"
+
+#if defined(WEBM_X86ASM) && (ARCH_X86 || ARCH_X86_64)
+	#include "rtcd/vpx_dsp_rtcd_x86.h"
+#elif defined(WEBM_ARMASM) && ARCH_ARM
+	#include "rtcd/vpx_dsp_rtcd_arm.h"
+#else
+	#include "rtcd/vpx_dsp_rtcd_c.h"
+#endif

thirdparty/libvpx/vpx_ports/arm_cpudetect.c

@@ -38,26 +38,7 @@ static int arm_cpu_env_mask(void) {
 }
 
 #if !CONFIG_RUNTIME_CPU_DETECT
-
+  #error "CONFIG_RUNTIME_CPU_DETECT should be enabled!"
-int arm_cpu_caps(void) {
-  /* This function should actually be a no-op. There is no way to adjust any of
-   * these because the RTCD tables do not exist: the functions are called
-   * statically */
-  int flags;
-  int mask;
-  if (!arm_cpu_env_flags(&flags)) {
-    return flags;
-  }
-  mask = arm_cpu_env_mask();
-#if HAVE_MEDIA
-  flags |= HAS_MEDIA;
-#endif /* HAVE_MEDIA */
-#if HAVE_NEON || HAVE_NEON_ASM
-  flags |= HAS_NEON;
-#endif /* HAVE_NEON  || HAVE_NEON_ASM */
-  return flags & mask;
-}
-
 #elif defined(_MSC_VER) /* end !CONFIG_RUNTIME_CPU_DETECT */
 /*For GetExceptionCode() and EXCEPTION_ILLEGAL_INSTRUCTION.*/
 #define WIN32_LEAN_AND_MEAN
@@ -76,28 +57,28 @@ int arm_cpu_caps(void) {
    * All of these instructions should be essentially nops.
    */
 #if HAVE_MEDIA
-  if (mask & HAS_MEDIA)
+  if (mask & HAS_MEDIA) {
     __try {
       /*SHADD8 r3,r3,r3*/
       __emit(0xE6333F93);
       flags |= HAS_MEDIA;
     } __except (GetExceptionCode() == EXCEPTION_ILLEGAL_INSTRUCTION) {
     /*Ignore exception.*/
+    }
   }
-}
 #endif /* HAVE_MEDIA */
 #if HAVE_NEON || HAVE_NEON_ASM
-if (mask &HAS_NEON) {
+  if (mask &HAS_NEON) {
-  __try {
+    __try {
-    /*VORR q0,q0,q0*/
+      /*VORR q0,q0,q0*/
-    __emit(0xF2200150);
+      __emit(0xF2200150);
-    flags |= HAS_NEON;
+      flags |= HAS_NEON;
-  } __except (GetExceptionCode() == EXCEPTION_ILLEGAL_INSTRUCTION) {
+    } __except (GetExceptionCode() == EXCEPTION_ILLEGAL_INSTRUCTION) {
-    /*Ignore exception.*/
+      /*Ignore exception.*/
+    }
   }
-}
 #endif /* HAVE_NEON || HAVE_NEON_ASM */
-return flags & mask;
+  return flags & mask;
 }
 
 #elif defined(__ANDROID__) /* end _MSC_VER */
@@ -170,6 +151,20 @@ int arm_cpu_caps(void) {
   return flags & mask;
 }
 #else /* end __linux__ */
-#error "--enable-runtime-cpu-detect selected, but no CPU detection method " \
+int arm_cpu_caps(void) {
-"available for your platform. Reconfigure with --disable-runtime-cpu-detect."
+  int flags;
+  int mask;
+  if (!arm_cpu_env_flags(&flags)) {
+    return flags;
+  }
+  mask = arm_cpu_env_mask();
+#if HAVE_MEDIA
+  flags |= HAS_MEDIA;
+#endif /* HAVE_MEDIA */
+#if HAVE_NEON || HAVE_NEON_ASM
+  flags |= HAS_NEON;
+#endif /* HAVE_NEON  || HAVE_NEON_ASM */
+  return flags & mask;
+}
+#warning "ARM run-time CPU detection is disabled for this platform..."
 #endif

thirdparty/libvpx/vpx_ports/system_state.h

@@ -13,10 +13,10 @@
 
 #include "./vpx_config.h"
 
-#if ARCH_X86 || ARCH_X86_64
+#if defined(WEBM_X86ASM) && (ARCH_X86 || ARCH_X86_64)
-void vpx_reset_mmx_state(void);
+	void vpx_reset_mmx_state(void);
-#define vpx_clear_system_state() vpx_reset_mmx_state()
+	#define vpx_clear_system_state() vpx_reset_mmx_state()
 #else
-#define vpx_clear_system_state()
+	#define vpx_clear_system_state()
 #endif  // ARCH_X86 || ARCH_X86_64
 #endif  // VPX_PORTS_SYSTEM_STATE_H_

thirdparty/libvpx/vpx_scale_rtcd.h

@@ -0,0 +1,44 @@
+#ifndef VPX_SCALE_RTCD_H_
+#define VPX_SCALE_RTCD_H_
+
+#ifdef RTCD_C
+#define RTCD_EXTERN
+#else
+#define RTCD_EXTERN extern
+#endif
+
+struct yv12_buffer_config;
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+void vp8_yv12_copy_frame_c(const struct yv12_buffer_config *src_ybc, struct yv12_buffer_config *dst_ybc);
+#define vp8_yv12_copy_frame vp8_yv12_copy_frame_c
+
+void vp8_yv12_extend_frame_borders_c(struct yv12_buffer_config *ybf);
+#define vp8_yv12_extend_frame_borders vp8_yv12_extend_frame_borders_c
+
+void vpx_extend_frame_borders_c(struct yv12_buffer_config *ybf);
+#define vpx_extend_frame_borders vpx_extend_frame_borders_c
+
+void vpx_extend_frame_inner_borders_c(struct yv12_buffer_config *ybf);
+#define vpx_extend_frame_inner_borders vpx_extend_frame_inner_borders_c
+
+void vpx_yv12_copy_y_c(const struct yv12_buffer_config *src_ybc, struct yv12_buffer_config *dst_ybc);
+#define vpx_yv12_copy_y vpx_yv12_copy_y_c
+
+void vpx_scale_rtcd(void);
+
+#ifdef RTCD_C
+static void setup_rtcd_internal(void)
+{
+    //Only MIPS has something here, but it is not supported
+}
+#endif
+
+#ifdef __cplusplus
+}  // extern "C"
+#endif
+
+#endif