Commit 53f64a77 authored by John Koleszar's avatar John Koleszar Committed by Code Review

Merge "arm: move unrolled loops back to generic code"

parents 9fdd90c9 19638c23
...@@ -15,36 +15,6 @@ ...@@ -15,36 +15,6 @@
extern void vp8_recon16x16mb_neon(unsigned char *pred_ptr, short *diff_ptr, unsigned char *dst_ptr, int ystride, unsigned char *udst_ptr, unsigned char *vdst_ptr); extern void vp8_recon16x16mb_neon(unsigned char *pred_ptr, short *diff_ptr, unsigned char *dst_ptr, int ystride, unsigned char *udst_ptr, unsigned char *vdst_ptr);
/*
void vp8_recon16x16mby(MACROBLOCKD *x)
{
int i;
for(i=0;i<16;i+=4)
{
//vp8_recon4b(&x->block[i]);
BLOCKD *b = &x->block[i];
vp8_recon4b (b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
}
}
*/
void vp8_recon16x16mby(const vp8_recon_rtcd_vtable_t *rtcd, MACROBLOCKD *x)
{
BLOCKD *b = &x->block[0];
RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
//b = &x->block[4];
b += 4;
RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
//b = &x->block[8];
b += 4;
RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
//b = &x->block[12];
b += 4;
RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
}
#if HAVE_ARMV7 #if HAVE_ARMV7
void vp8_recon16x16mb(const vp8_recon_rtcd_vtable_t *rtcd, MACROBLOCKD *x) void vp8_recon16x16mb(const vp8_recon_rtcd_vtable_t *rtcd, MACROBLOCKD *x)
{ {
...@@ -58,52 +28,4 @@ void vp8_recon16x16mb(const vp8_recon_rtcd_vtable_t *rtcd, MACROBLOCKD *x) ...@@ -58,52 +28,4 @@ void vp8_recon16x16mb(const vp8_recon_rtcd_vtable_t *rtcd, MACROBLOCKD *x)
vp8_recon16x16mb_neon(pred_ptr, diff_ptr, dst_ptr, ystride, udst_ptr, vdst_ptr); vp8_recon16x16mb_neon(pred_ptr, diff_ptr, dst_ptr, ystride, udst_ptr, vdst_ptr);
} }
#else
/*
void vp8_recon16x16mb(MACROBLOCKD *x)
{
int i;
for(i=0;i<16;i+=4)
{
// vp8_recon4b(&x->block[i]);
BLOCKD *b = &x->block[i];
vp8_recon4b (b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
}
for(i=16;i<24;i+=2)
{
// vp8_recon2b(&x->block[i]);
BLOCKD *b = &x->block[i];
vp8_recon2b (b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
}
}
*/
void vp8_recon16x16mb(const vp8_recon_rtcd_vtable_t *rtcd, MACROBLOCKD *x)
{
BLOCKD *b = &x->block[0];
RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b += 4;
RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b += 4;
RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b += 4;
RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b += 4;
//b = &x->block[16];
RECON_INVOKE(rtcd, recon2)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b++;
b++;
RECON_INVOKE(rtcd, recon2)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b++;
b++;
RECON_INVOKE(rtcd, recon2)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b++;
b++;
RECON_INVOKE(rtcd, recon2)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
}
#endif #endif
/*
* 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.
*/
#include "vpx_ports/config.h"
#include "recon.h"
#include "vpx_mem/vpx_mem.h"
#include "reconintra.h"
void vp8_predict_intra4x4(BLOCKD *x,
int b_mode,
unsigned char *predictor)
{
int i, r, c;
unsigned char *Above = *(x->base_dst) + x->dst - x->dst_stride;
unsigned char Left[4];
unsigned char top_left = Above[-1];
Left[0] = (*(x->base_dst))[x->dst - 1];
Left[1] = (*(x->base_dst))[x->dst - 1 + x->dst_stride];
Left[2] = (*(x->base_dst))[x->dst - 1 + 2 * x->dst_stride];
Left[3] = (*(x->base_dst))[x->dst - 1 + 3 * x->dst_stride];
switch (b_mode)
{
case B_DC_PRED:
{
int expected_dc = 0;
for (i = 0; i < 4; i++)
{
expected_dc += Above[i];
expected_dc += Left[i];
}
expected_dc = (expected_dc + 4) >> 3;
for (r = 0; r < 4; r++)
{
for (c = 0; c < 4; c++)
{
predictor[c] = expected_dc;
}
predictor += 16;
}
}
break;
case B_TM_PRED:
{
// prediction similar to true_motion prediction
for (r = 0; r < 4; r++)
{
for (c = 0; c < 4; c++)
{
int pred = Above[c] - top_left + Left[r];
if (pred < 0)
pred = 0;
if (pred > 255)
pred = 255;
predictor[c] = pred;
}
predictor += 16;
}
}
break;
case B_VE_PRED:
{
unsigned int ap[4];
ap[0] = (top_left + 2 * Above[0] + Above[1] + 2) >> 2;
ap[1] = (Above[0] + 2 * Above[1] + Above[2] + 2) >> 2;
ap[2] = (Above[1] + 2 * Above[2] + Above[3] + 2) >> 2;
ap[3] = (Above[2] + 2 * Above[3] + Above[4] + 2) >> 2;
for (r = 0; r < 4; r++)
{
for (c = 0; c < 4; c++)
{
predictor[c] = ap[c];
}
predictor += 16;
}
}
break;
case B_HE_PRED:
{
unsigned int lp[4];
lp[0] = (top_left + 2 * Left[0] + Left[1] + 2) >> 2;
lp[1] = (Left[0] + 2 * Left[1] + Left[2] + 2) >> 2;
lp[2] = (Left[1] + 2 * Left[2] + Left[3] + 2) >> 2;
lp[3] = (Left[2] + 2 * Left[3] + Left[3] + 2) >> 2;
for (r = 0; r < 4; r++)
{
for (c = 0; c < 4; c++)
{
predictor[c] = lp[r];
}
predictor += 16;
}
}
break;
case B_LD_PRED:
{
unsigned char *ptr = Above;
predictor[0 * 16 + 0] = (ptr[0] + ptr[1] * 2 + ptr[2] + 2) >> 2;
predictor[0 * 16 + 1] =
predictor[1 * 16 + 0] = (ptr[1] + ptr[2] * 2 + ptr[3] + 2) >> 2;
predictor[0 * 16 + 2] =
predictor[1 * 16 + 1] =
predictor[2 * 16 + 0] = (ptr[2] + ptr[3] * 2 + ptr[4] + 2) >> 2;
predictor[0 * 16 + 3] =
predictor[1 * 16 + 2] =
predictor[2 * 16 + 1] =
predictor[3 * 16 + 0] = (ptr[3] + ptr[4] * 2 + ptr[5] + 2) >> 2;
predictor[1 * 16 + 3] =
predictor[2 * 16 + 2] =
predictor[3 * 16 + 1] = (ptr[4] + ptr[5] * 2 + ptr[6] + 2) >> 2;
predictor[2 * 16 + 3] =
predictor[3 * 16 + 2] = (ptr[5] + ptr[6] * 2 + ptr[7] + 2) >> 2;
predictor[3 * 16 + 3] = (ptr[6] + ptr[7] * 2 + ptr[7] + 2) >> 2;
}
break;
case B_RD_PRED:
{
unsigned char pp[9];
pp[0] = Left[3];
pp[1] = Left[2];
pp[2] = Left[1];
pp[3] = Left[0];
pp[4] = top_left;
pp[5] = Above[0];
pp[6] = Above[1];
pp[7] = Above[2];
pp[8] = Above[3];
predictor[3 * 16 + 0] = (pp[0] + pp[1] * 2 + pp[2] + 2) >> 2;
predictor[3 * 16 + 1] =
predictor[2 * 16 + 0] = (pp[1] + pp[2] * 2 + pp[3] + 2) >> 2;
predictor[3 * 16 + 2] =
predictor[2 * 16 + 1] =
predictor[1 * 16 + 0] = (pp[2] + pp[3] * 2 + pp[4] + 2) >> 2;
predictor[3 * 16 + 3] =
predictor[2 * 16 + 2] =
predictor[1 * 16 + 1] =
predictor[0 * 16 + 0] = (pp[3] + pp[4] * 2 + pp[5] + 2) >> 2;
predictor[2 * 16 + 3] =
predictor[1 * 16 + 2] =
predictor[0 * 16 + 1] = (pp[4] + pp[5] * 2 + pp[6] + 2) >> 2;
predictor[1 * 16 + 3] =
predictor[0 * 16 + 2] = (pp[5] + pp[6] * 2 + pp[7] + 2) >> 2;
predictor[0 * 16 + 3] = (pp[6] + pp[7] * 2 + pp[8] + 2) >> 2;
}
break;
case B_VR_PRED:
{
unsigned char pp[9];
pp[0] = Left[3];
pp[1] = Left[2];
pp[2] = Left[1];
pp[3] = Left[0];
pp[4] = top_left;
pp[5] = Above[0];
pp[6] = Above[1];
pp[7] = Above[2];
pp[8] = Above[3];
predictor[3 * 16 + 0] = (pp[1] + pp[2] * 2 + pp[3] + 2) >> 2;
predictor[2 * 16 + 0] = (pp[2] + pp[3] * 2 + pp[4] + 2) >> 2;
predictor[3 * 16 + 1] =
predictor[1 * 16 + 0] = (pp[3] + pp[4] * 2 + pp[5] + 2) >> 2;
predictor[2 * 16 + 1] =
predictor[0 * 16 + 0] = (pp[4] + pp[5] + 1) >> 1;
predictor[3 * 16 + 2] =
predictor[1 * 16 + 1] = (pp[4] + pp[5] * 2 + pp[6] + 2) >> 2;
predictor[2 * 16 + 2] =
predictor[0 * 16 + 1] = (pp[5] + pp[6] + 1) >> 1;
predictor[3 * 16 + 3] =
predictor[1 * 16 + 2] = (pp[5] + pp[6] * 2 + pp[7] + 2) >> 2;
predictor[2 * 16 + 3] =
predictor[0 * 16 + 2] = (pp[6] + pp[7] + 1) >> 1;
predictor[1 * 16 + 3] = (pp[6] + pp[7] * 2 + pp[8] + 2) >> 2;
predictor[0 * 16 + 3] = (pp[7] + pp[8] + 1) >> 1;
}
break;
case B_VL_PRED:
{
unsigned char *pp = Above;
predictor[0 * 16 + 0] = (pp[0] + pp[1] + 1) >> 1;
predictor[1 * 16 + 0] = (pp[0] + pp[1] * 2 + pp[2] + 2) >> 2;
predictor[2 * 16 + 0] =
predictor[0 * 16 + 1] = (pp[1] + pp[2] + 1) >> 1;
predictor[1 * 16 + 1] =
predictor[3 * 16 + 0] = (pp[1] + pp[2] * 2 + pp[3] + 2) >> 2;
predictor[2 * 16 + 1] =
predictor[0 * 16 + 2] = (pp[2] + pp[3] + 1) >> 1;
predictor[3 * 16 + 1] =
predictor[1 * 16 + 2] = (pp[2] + pp[3] * 2 + pp[4] + 2) >> 2;
predictor[0 * 16 + 3] =
predictor[2 * 16 + 2] = (pp[3] + pp[4] + 1) >> 1;
predictor[1 * 16 + 3] =
predictor[3 * 16 + 2] = (pp[3] + pp[4] * 2 + pp[5] + 2) >> 2;
predictor[2 * 16 + 3] = (pp[4] + pp[5] * 2 + pp[6] + 2) >> 2;
predictor[3 * 16 + 3] = (pp[5] + pp[6] * 2 + pp[7] + 2) >> 2;
}
break;
case B_HD_PRED:
{
unsigned char pp[9];
pp[0] = Left[3];
pp[1] = Left[2];
pp[2] = Left[1];
pp[3] = Left[0];
pp[4] = top_left;
pp[5] = Above[0];
pp[6] = Above[1];
pp[7] = Above[2];
pp[8] = Above[3];
predictor[3 * 16 + 0] = (pp[0] + pp[1] + 1) >> 1;
predictor[3 * 16 + 1] = (pp[0] + pp[1] * 2 + pp[2] + 2) >> 2;
predictor[2 * 16 + 0] =
predictor[3 * 16 + 2] = (pp[1] + pp[2] + 1) >> 1;
predictor[2 * 16 + 1] =
predictor[3 * 16 + 3] = (pp[1] + pp[2] * 2 + pp[3] + 2) >> 2;
predictor[2 * 16 + 2] =
predictor[1 * 16 + 0] = (pp[2] + pp[3] + 1) >> 1;
predictor[2 * 16 + 3] =
predictor[1 * 16 + 1] = (pp[2] + pp[3] * 2 + pp[4] + 2) >> 2;
predictor[1 * 16 + 2] =
predictor[0 * 16 + 0] = (pp[3] + pp[4] + 1) >> 1;
predictor[1 * 16 + 3] =
predictor[0 * 16 + 1] = (pp[3] + pp[4] * 2 + pp[5] + 2) >> 2;
predictor[0 * 16 + 2] = (pp[4] + pp[5] * 2 + pp[6] + 2) >> 2;
predictor[0 * 16 + 3] = (pp[5] + pp[6] * 2 + pp[7] + 2) >> 2;
}
break;
case B_HU_PRED:
{
unsigned char *pp = Left;
predictor[0 * 16 + 0] = (pp[0] + pp[1] + 1) >> 1;
predictor[0 * 16 + 1] = (pp[0] + pp[1] * 2 + pp[2] + 2) >> 2;
predictor[0 * 16 + 2] =
predictor[1 * 16 + 0] = (pp[1] + pp[2] + 1) >> 1;
predictor[0 * 16 + 3] =
predictor[1 * 16 + 1] = (pp[1] + pp[2] * 2 + pp[3] + 2) >> 2;
predictor[1 * 16 + 2] =
predictor[2 * 16 + 0] = (pp[2] + pp[3] + 1) >> 1;
predictor[1 * 16 + 3] =
predictor[2 * 16 + 1] = (pp[2] + pp[3] * 2 + pp[3] + 2) >> 2;
predictor[2 * 16 + 2] =
predictor[2 * 16 + 3] =
predictor[3 * 16 + 0] =
predictor[3 * 16 + 1] =
predictor[3 * 16 + 2] =
predictor[3 * 16 + 3] = pp[3];
}
break;
}
}
extern void vp8_intra_prediction_down_copy(MACROBLOCKD *x);
/*
void vp8_recon_intra4x4mb(const vp8_recon_rtcd_vtable_t *rtcd, MACROBLOCKD *x)
{
int i;
vp8_intra_prediction_down_copy(x);
for(i=0;i<16;i++)
{
BLOCKD *b = &x->block[i];
vp8_predict_intra4x4(b, x->block[i].bmi.mode,x->block[i].predictor);
RECON_INVOKE(rtcd, recon)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
}
vp8_recon_intra_mbuv(x);
}
*/
void vp8_recon_intra4x4mb(const vp8_recon_rtcd_vtable_t *rtcd, MACROBLOCKD *x)
{
int i;
BLOCKD *b = &x->block[0];
vp8_intra_prediction_down_copy(x);
{
vp8_predict_intra4x4(b, b->bmi.mode, b->predictor);
RECON_INVOKE(rtcd, recon)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b += 1;
vp8_predict_intra4x4(b, b->bmi.mode, b->predictor);
RECON_INVOKE(rtcd, recon)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b += 1;
vp8_predict_intra4x4(b, b->bmi.mode, b->predictor);
RECON_INVOKE(rtcd, recon)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b += 1;
vp8_predict_intra4x4(b, b->bmi.mode, b->predictor);
RECON_INVOKE(rtcd, recon)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b += 1;
vp8_predict_intra4x4(b, b->bmi.mode, b->predictor);
RECON_INVOKE(rtcd, recon)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b += 1;
vp8_predict_intra4x4(b, b->bmi.mode, b->predictor);
RECON_INVOKE(rtcd, recon)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b += 1;
vp8_predict_intra4x4(b, b->bmi.mode, b->predictor);
RECON_INVOKE(rtcd, recon)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b += 1;
vp8_predict_intra4x4(b, b->bmi.mode, b->predictor);
RECON_INVOKE(rtcd, recon)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b += 1;
vp8_predict_intra4x4(b, b->bmi.mode, b->predictor);
RECON_INVOKE(rtcd, recon)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b += 1;
vp8_predict_intra4x4(b, b->bmi.mode, b->predictor);
RECON_INVOKE(rtcd, recon)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b += 1;
vp8_predict_intra4x4(b, b->bmi.mode, b->predictor);
RECON_INVOKE(rtcd, recon)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b += 1;
vp8_predict_intra4x4(b, b->bmi.mode, b->predictor);
RECON_INVOKE(rtcd, recon)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b += 1;
vp8_predict_intra4x4(b, b->bmi.mode, b->predictor);
RECON_INVOKE(rtcd, recon)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b += 1;
vp8_predict_intra4x4(b, b->bmi.mode, b->predictor);
RECON_INVOKE(rtcd, recon)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b += 1;
vp8_predict_intra4x4(b, b->bmi.mode, b->predictor);
RECON_INVOKE(rtcd, recon)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b += 1;
vp8_predict_intra4x4(b, b->bmi.mode, b->predictor);
RECON_INVOKE(rtcd, recon)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
}
vp8_recon_intra_mbuv(rtcd, x);
}
...@@ -108,6 +108,22 @@ void vp8_recon2b_c ...@@ -108,6 +108,22 @@ void vp8_recon2b_c
void vp8_recon16x16mby(const vp8_recon_rtcd_vtable_t *rtcd, MACROBLOCKD *x) void vp8_recon16x16mby(const vp8_recon_rtcd_vtable_t *rtcd, MACROBLOCKD *x)
{ {
#if ARCH_ARM
BLOCKD *b = &x->block[0];
RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
//b = &x->block[4];
b += 4;
RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
//b = &x->block[8];
b += 4;
RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
//b = &x->block[12];
b += 4;
RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
#else
int i; int i;
for (i = 0; i < 16; i += 4) for (i = 0; i < 16; i += 4)
...@@ -116,10 +132,36 @@ void vp8_recon16x16mby(const vp8_recon_rtcd_vtable_t *rtcd, MACROBLOCKD *x) ...@@ -116,10 +132,36 @@ void vp8_recon16x16mby(const vp8_recon_rtcd_vtable_t *rtcd, MACROBLOCKD *x)
RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride); RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
} }
#endif
} }
void vp8_recon16x16mb(const vp8_recon_rtcd_vtable_t *rtcd, MACROBLOCKD *x) void vp8_recon16x16mb(const vp8_recon_rtcd_vtable_t *rtcd, MACROBLOCKD *x)
{ {
#if ARCH_ARM
BLOCKD *b = &x->block[0];
RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b += 4;
RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b += 4;
RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b += 4;
RECON_INVOKE(rtcd, recon4)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b += 4;
//b = &x->block[16];
RECON_INVOKE(rtcd, recon2)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b++;
b++;
RECON_INVOKE(rtcd, recon2)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b++;
b++;
RECON_INVOKE(rtcd, recon2)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b++;
b++;
RECON_INVOKE(rtcd, recon2)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
#else
int i; int i;
for (i = 0; i < 16; i += 4) for (i = 0; i < 16; i += 4)
...@@ -135,4 +177,5 @@ void vp8_recon16x16mb(const vp8_recon_rtcd_vtable_t *rtcd, MACROBLOCKD *x) ...@@ -135,4 +177,5 @@ void vp8_recon16x16mb(const vp8_recon_rtcd_vtable_t *rtcd, MACROBLOCKD *x)
RECON_INVOKE(rtcd, recon2)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride); RECON_INVOKE(rtcd, recon2)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
} }
#endif
} }
...@@ -318,6 +318,74 @@ void vp8_recon_intra4x4mb(const vp8_recon_rtcd_vtable_t *rtcd, MACROBLOCKD *x) ...@@ -318,6 +318,74 @@ void vp8_recon_intra4x4mb(const vp8_recon_rtcd_vtable_t *rtcd, MACROBLOCKD *x)
vp8_intra_prediction_down_copy(x); vp8_intra_prediction_down_copy(x);
#if ARCH_ARM
{
BLOCKD *b = &x->block[0];
vp8_predict_intra4x4(b, b->bmi.mode, b->predictor);
RECON_INVOKE(rtcd, recon)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b += 1;
vp8_predict_intra4x4(b, b->bmi.mode, b->predictor);
RECON_INVOKE(rtcd, recon)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b += 1;
vp8_predict_intra4x4(b, b->bmi.mode, b->predictor);
RECON_INVOKE(rtcd, recon)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b += 1;
vp8_predict_intra4x4(b, b->bmi.mode, b->predictor);
RECON_INVOKE(rtcd, recon)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);
b += 1;
vp8_predict_intra4x4(b, b->bmi.mode, b->predictor);
RECON_INVOKE(rtcd, recon)(b->predictor, b->diff, *(b->base_dst) + b->dst, b->dst_stride);