arm: move unrolled loops back to generic code

Some of the ARM functions differed from their generic counterparts
only by unrolling their loops. Since this change may be useful
on other platforms, or might even supercede the looped version
in the generic case, move it back to the generic file.

This code is left under #if ARCH_ARM for now, but it may be worth
considering a different (possibly new) conditional for these. If
it turns out that this should be runtime selectable, these
functions will have to move to the RTCD infrastructure. Don't want
to take that step at this time without more profile data.

Change-Id: I4612fdbc606fbebba4971a690fb743ad184ff15f

diff --git a/vp8/common/arm/recon_arm.c b/vp8/common/arm/recon_arm.c
index 218898b441af7d16e7829ae40e6a29d3fa614fa7..a1b210130990ef9217e69202835207c5bf82ca27 100644 (file)
--- a/vp8/common/arm/recon_arm.c
+++ b/vp8/common/arm/recon_arm.c
@@ -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);
 
-/*
-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
 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);
 }
-
-#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

diff --git a/vp8/common/arm/reconintra4x4_arm.c b/vp8/common/arm/reconintra4x4_arm.c
deleted file mode 100644 (file)
index 6fd9b9c..0000000
--- a/vp8/common/arm/reconintra4x4_arm.c
+++ /dev/null
@@ -1,395 +0,0 @@
-/*
- *  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.
- */
-
-
-#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);
-
-}

diff --git a/vp8/common/recon.c b/vp8/common/recon.c
index 0b439e0540d08d6d67ee86f4fc2bda075c8e1d93..f31ffd8e4dfbcfeb22cd36ffab043f2002ff8167 100644 (file)
--- a/vp8/common/recon.c
+++ b/vp8/common/recon.c
@@ -108,6 +108,22 @@ void vp8_recon2b_c
 
 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;
 
     for (i = 0; i < 16; i += 4)
@@ -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);
     }
+#endif
 }
 
 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;
 
     for (i = 0; i < 16; i += 4)
@@ -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);
     }
+#endif
 }

diff --git a/vp8/common/reconintra4x4.c b/vp8/common/reconintra4x4.c
index c6e5fe7fdb66f556f5123ea3d4e4464166b85078..31513518454015b256a1bb308c3c13c1dd4bec13 100644 (file)
--- a/vp8/common/reconintra4x4.c
+++ b/vp8/common/reconintra4x4.c
@@ -318,6 +318,74 @@ void vp8_recon_intra4x4mb(const vp8_recon_rtcd_vtable_t *rtcd, MACROBLOCKD *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);
+        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);
+    }
+#else
     for (i = 0; i < 16; i++)
     {
         BLOCKD *b = &x->block[i];
@@ -325,6 +393,7 @@ void vp8_recon_intra4x4mb(const vp8_recon_rtcd_vtable_t *rtcd, MACROBLOCKD *x)
         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);
     }
+#endif
 
     vp8_recon_intra_mbuv(rtcd, x);
 

diff --git a/vp8/vp8_common.mk b/vp8/vp8_common.mk
index 9ab92b33431c7ea933e5de0b9b7faeb2b4809180..24cbbdd0c671ed16849bc235d4e5b8a7b20f5a77 100644 (file)
--- a/vp8/vp8_common.mk
+++ b/vp8/vp8_common.mk
@@ -119,7 +119,6 @@ VP8_COMMON_SRCS-$(HAVE_ARMV6)  += common/arm/bilinearfilter_arm.c
 VP8_COMMON_SRCS-$(HAVE_ARMV6)  += common/arm/filter_arm.c
 VP8_COMMON_SRCS-$(HAVE_ARMV6)  += common/arm/loopfilter_arm.c
 VP8_COMMON_SRCS-$(HAVE_ARMV6)  += common/arm/recon_arm.c
-VP8_COMMON_SRCS-$(HAVE_ARMV6)  += common/arm/reconintra4x4_arm.c
 VP8_COMMON_SRCS-$(HAVE_ARMV6)  += common/arm/reconintra_arm.c
 VP8_COMMON_SRCS-$(HAVE_ARMV6)  += common/arm/vpx_asm_offsets.c