summaryrefslogtreecommitdiff
path: root/vp9/encoder
diff options
context:
space:
mode:
Diffstat (limited to 'vp9/encoder')
-rw-r--r--vp9/encoder/arm/neon/vp9_subtract_neon.c81
-rw-r--r--vp9/encoder/mips/msa/vp9_subtract_msa.c264
-rw-r--r--vp9/encoder/vp9_encodemb.c60
-rw-r--r--vp9/encoder/vp9_rdopt.c11
-rw-r--r--vp9/encoder/x86/vp9_subtract_sse2.asm127
5 files changed, 17 insertions, 526 deletions
diff --git a/vp9/encoder/arm/neon/vp9_subtract_neon.c b/vp9/encoder/arm/neon/vp9_subtract_neon.c
deleted file mode 100644
index b4bf567db..000000000
--- a/vp9/encoder/arm/neon/vp9_subtract_neon.c
+++ /dev/null
@@ -1,81 +0,0 @@
-/*
- * 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.
- */
-
-#include <arm_neon.h>
-#include "./vp9_rtcd.h"
-#include "./vpx_config.h"
-
-#include "vpx/vpx_integer.h"
-
-void vp9_subtract_block_neon(int rows, int cols,
- int16_t *diff, ptrdiff_t diff_stride,
- const uint8_t *src, ptrdiff_t src_stride,
- const uint8_t *pred, ptrdiff_t pred_stride) {
- int r, c;
-
- if (cols > 16) {
- for (r = 0; r < rows; ++r) {
- for (c = 0; c < cols; c += 32) {
- const uint8x16_t v_src_00 = vld1q_u8(&src[c + 0]);
- const uint8x16_t v_src_16 = vld1q_u8(&src[c + 16]);
- const uint8x16_t v_pred_00 = vld1q_u8(&pred[c + 0]);
- const uint8x16_t v_pred_16 = vld1q_u8(&pred[c + 16]);
- const uint16x8_t v_diff_lo_00 = vsubl_u8(vget_low_u8(v_src_00),
- vget_low_u8(v_pred_00));
- const uint16x8_t v_diff_hi_00 = vsubl_u8(vget_high_u8(v_src_00),
- vget_high_u8(v_pred_00));
- const uint16x8_t v_diff_lo_16 = vsubl_u8(vget_low_u8(v_src_16),
- vget_low_u8(v_pred_16));
- const uint16x8_t v_diff_hi_16 = vsubl_u8(vget_high_u8(v_src_16),
- vget_high_u8(v_pred_16));
- vst1q_s16(&diff[c + 0], vreinterpretq_s16_u16(v_diff_lo_00));
- vst1q_s16(&diff[c + 8], vreinterpretq_s16_u16(v_diff_hi_00));
- vst1q_s16(&diff[c + 16], vreinterpretq_s16_u16(v_diff_lo_16));
- vst1q_s16(&diff[c + 24], vreinterpretq_s16_u16(v_diff_hi_16));
- }
- diff += diff_stride;
- pred += pred_stride;
- src += src_stride;
- }
- } else if (cols > 8) {
- for (r = 0; r < rows; ++r) {
- const uint8x16_t v_src = vld1q_u8(&src[0]);
- const uint8x16_t v_pred = vld1q_u8(&pred[0]);
- const uint16x8_t v_diff_lo = vsubl_u8(vget_low_u8(v_src),
- vget_low_u8(v_pred));
- const uint16x8_t v_diff_hi = vsubl_u8(vget_high_u8(v_src),
- vget_high_u8(v_pred));
- vst1q_s16(&diff[0], vreinterpretq_s16_u16(v_diff_lo));
- vst1q_s16(&diff[8], vreinterpretq_s16_u16(v_diff_hi));
- diff += diff_stride;
- pred += pred_stride;
- src += src_stride;
- }
- } else if (cols > 4) {
- for (r = 0; r < rows; ++r) {
- const uint8x8_t v_src = vld1_u8(&src[0]);
- const uint8x8_t v_pred = vld1_u8(&pred[0]);
- const uint16x8_t v_diff = vsubl_u8(v_src, v_pred);
- vst1q_s16(&diff[0], vreinterpretq_s16_u16(v_diff));
- diff += diff_stride;
- pred += pred_stride;
- src += src_stride;
- }
- } else {
- for (r = 0; r < rows; ++r) {
- for (c = 0; c < cols; ++c)
- diff[c] = src[c] - pred[c];
-
- diff += diff_stride;
- pred += pred_stride;
- src += src_stride;
- }
- }
-}
diff --git a/vp9/encoder/mips/msa/vp9_subtract_msa.c b/vp9/encoder/mips/msa/vp9_subtract_msa.c
deleted file mode 100644
index 1b8b694ce..000000000
--- a/vp9/encoder/mips/msa/vp9_subtract_msa.c
+++ /dev/null
@@ -1,264 +0,0 @@
-/*
- * Copyright (c) 2015 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 "./vp9_rtcd.h"
-#include "vp9/common/mips/msa/vp9_macros_msa.h"
-
-static void sub_blk_4x4_msa(const uint8_t *src_ptr, int32_t src_stride,
- const uint8_t *pred_ptr, int32_t pred_stride,
- int16_t *diff_ptr, int32_t diff_stride) {
- uint32_t src0, src1, src2, src3;
- uint32_t pred0, pred1, pred2, pred3;
- v16i8 src = { 0 };
- v16i8 pred = { 0 };
- v16u8 src_l0, src_l1;
- v8i16 diff0, diff1;
-
- LW4(src_ptr, src_stride, src0, src1, src2, src3);
- LW4(pred_ptr, pred_stride, pred0, pred1, pred2, pred3);
- INSERT_W4_SB(src0, src1, src2, src3, src);
- INSERT_W4_SB(pred0, pred1, pred2, pred3, pred);
- ILVRL_B2_UB(src, pred, src_l0, src_l1);
- HSUB_UB2_SH(src_l0, src_l1, diff0, diff1);
- ST8x4_UB(diff0, diff1, diff_ptr, (2 * diff_stride));
-}
-
-static void sub_blk_8x8_msa(const uint8_t *src_ptr, int32_t src_stride,
- const uint8_t *pred_ptr, int32_t pred_stride,
- int16_t *diff_ptr, int32_t diff_stride) {
- uint32_t loop_cnt;
- uint64_t src0, src1, pred0, pred1;
- v16i8 src = { 0 };
- v16i8 pred = { 0 };
- v16u8 src_l0, src_l1;
- v8i16 diff0, diff1;
-
- for (loop_cnt = 4; loop_cnt--;) {
- LD2(src_ptr, src_stride, src0, src1);
- src_ptr += (2 * src_stride);
- LD2(pred_ptr, pred_stride, pred0, pred1);
- pred_ptr += (2 * pred_stride);
-
- INSERT_D2_SB(src0, src1, src);
- INSERT_D2_SB(pred0, pred1, pred);
- ILVRL_B2_UB(src, pred, src_l0, src_l1);
- HSUB_UB2_SH(src_l0, src_l1, diff0, diff1);
- ST_SH2(diff0, diff1, diff_ptr, diff_stride);
- diff_ptr += (2 * diff_stride);
- }
-}
-
-static void sub_blk_16x16_msa(const uint8_t *src, int32_t src_stride,
- const uint8_t *pred, int32_t pred_stride,
- int16_t *diff, int32_t diff_stride) {
- int8_t count;
- v16i8 src0, src1, src2, src3, src4, src5, src6, src7;
- v16i8 pred0, pred1, pred2, pred3, pred4, pred5, pred6, pred7;
- v16u8 src_l0, src_l1;
- v8i16 diff0, diff1;
-
- for (count = 2; count--;) {
- LD_SB8(src, src_stride, src0, src1, src2, src3, src4, src5, src6, src7);
- src += (8 * src_stride);
-
- LD_SB8(pred, pred_stride,
- pred0, pred1, pred2, pred3, pred4, pred5, pred6, pred7);
- pred += (8 * pred_stride);
-
- ILVRL_B2_UB(src0, pred0, src_l0, src_l1);
- HSUB_UB2_SH(src_l0, src_l1, diff0, diff1);
- ST_SH2(diff0, diff1, diff, 8);
- diff += diff_stride;
-
- ILVRL_B2_UB(src1, pred1, src_l0, src_l1);
- HSUB_UB2_SH(src_l0, src_l1, diff0, diff1);
- ST_SH2(diff0, diff1, diff, 8);
- diff += diff_stride;
-
- ILVRL_B2_UB(src2, pred2, src_l0, src_l1);
- HSUB_UB2_SH(src_l0, src_l1, diff0, diff1);
- ST_SH2(diff0, diff1, diff, 8);
- diff += diff_stride;
-
- ILVRL_B2_UB(src3, pred3, src_l0, src_l1);
- HSUB_UB2_SH(src_l0, src_l1, diff0, diff1);
- ST_SH2(diff0, diff1, diff, 8);
- diff += diff_stride;
-
- ILVRL_B2_UB(src4, pred4, src_l0, src_l1);
- HSUB_UB2_SH(src_l0, src_l1, diff0, diff1);
- ST_SH2(diff0, diff1, diff, 8);
- diff += diff_stride;
-
- ILVRL_B2_UB(src5, pred5, src_l0, src_l1);
- HSUB_UB2_SH(src_l0, src_l1, diff0, diff1);
- ST_SH2(diff0, diff1, diff, 8);
- diff += diff_stride;
-
- ILVRL_B2_UB(src6, pred6, src_l0, src_l1);
- HSUB_UB2_SH(src_l0, src_l1, diff0, diff1);
- ST_SH2(diff0, diff1, diff, 8);
- diff += diff_stride;
-
- ILVRL_B2_UB(src7, pred7, src_l0, src_l1);
- HSUB_UB2_SH(src_l0, src_l1, diff0, diff1);
- ST_SH2(diff0, diff1, diff, 8);
- diff += diff_stride;
- }
-}
-
-static void sub_blk_32x32_msa(const uint8_t *src, int32_t src_stride,
- const uint8_t *pred, int32_t pred_stride,
- int16_t *diff, int32_t diff_stride) {
- uint32_t loop_cnt;
- v16i8 src0, src1, src2, src3, src4, src5, src6, src7;
- v16i8 pred0, pred1, pred2, pred3, pred4, pred5, pred6, pred7;
- v16u8 src_l0, src_l1;
- v8i16 diff0, diff1;
-
- for (loop_cnt = 8; loop_cnt--;) {
- LD_SB2(src, 16, src0, src1);
- src += src_stride;
- LD_SB2(src, 16, src2, src3);
- src += src_stride;
- LD_SB2(src, 16, src4, src5);
- src += src_stride;
- LD_SB2(src, 16, src6, src7);
- src += src_stride;
-
- LD_SB2(pred, 16, pred0, pred1);
- pred += pred_stride;
- LD_SB2(pred, 16, pred2, pred3);
- pred += pred_stride;
- LD_SB2(pred, 16, pred4, pred5);
- pred += pred_stride;
- LD_SB2(pred, 16, pred6, pred7);
- pred += pred_stride;
-
- ILVRL_B2_UB(src0, pred0, src_l0, src_l1);
- HSUB_UB2_SH(src_l0, src_l1, diff0, diff1);
- ST_SH2(diff0, diff1, diff, 8);
- ILVRL_B2_UB(src1, pred1, src_l0, src_l1);
- HSUB_UB2_SH(src_l0, src_l1, diff0, diff1);
- ST_SH2(diff0, diff1, diff + 16, 8);
- diff += diff_stride;
-
- ILVRL_B2_UB(src2, pred2, src_l0, src_l1);
- HSUB_UB2_SH(src_l0, src_l1, diff0, diff1);
- ST_SH2(diff0, diff1, diff, 8);
- ILVRL_B2_UB(src3, pred3, src_l0, src_l1);
- HSUB_UB2_SH(src_l0, src_l1, diff0, diff1);
- ST_SH2(diff0, diff1, diff + 16, 8);
- diff += diff_stride;
-
- ILVRL_B2_UB(src4, pred4, src_l0, src_l1);
- HSUB_UB2_SH(src_l0, src_l1, diff0, diff1);
- ST_SH2(diff0, diff1, diff, 8);
- ILVRL_B2_UB(src5, pred5, src_l0, src_l1);
- HSUB_UB2_SH(src_l0, src_l1, diff0, diff1);
- ST_SH2(diff0, diff1, diff + 16, 8);
- diff += diff_stride;
-
- ILVRL_B2_UB(src6, pred6, src_l0, src_l1);
- HSUB_UB2_SH(src_l0, src_l1, diff0, diff1);
- ST_SH2(diff0, diff1, diff, 8);
- ILVRL_B2_UB(src7, pred7, src_l0, src_l1);
- HSUB_UB2_SH(src_l0, src_l1, diff0, diff1);
- ST_SH2(diff0, diff1, diff + 16, 8);
- diff += diff_stride;
- }
-}
-
-static void sub_blk_64x64_msa(const uint8_t *src, int32_t src_stride,
- const uint8_t *pred, int32_t pred_stride,
- int16_t *diff, int32_t diff_stride) {
- uint32_t loop_cnt;
- v16i8 src0, src1, src2, src3, src4, src5, src6, src7;
- v16i8 pred0, pred1, pred2, pred3, pred4, pred5, pred6, pred7;
- v16u8 src_l0, src_l1;
- v8i16 diff0, diff1;
-
- for (loop_cnt = 32; loop_cnt--;) {
- LD_SB4(src, 16, src0, src1, src2, src3);
- src += src_stride;
- LD_SB4(src, 16, src4, src5, src6, src7);
- src += src_stride;
-
- LD_SB4(pred, 16, pred0, pred1, pred2, pred3);
- pred += pred_stride;
- LD_SB4(pred, 16, pred4, pred5, pred6, pred7);
- pred += pred_stride;
-
- ILVRL_B2_UB(src0, pred0, src_l0, src_l1);
- HSUB_UB2_SH(src_l0, src_l1, diff0, diff1);
- ST_SH2(diff0, diff1, diff, 8);
- ILVRL_B2_UB(src1, pred1, src_l0, src_l1);
- HSUB_UB2_SH(src_l0, src_l1, diff0, diff1);
- ST_SH2(diff0, diff1, diff + 16, 8);
- ILVRL_B2_UB(src2, pred2, src_l0, src_l1);
- HSUB_UB2_SH(src_l0, src_l1, diff0, diff1);
- ST_SH2(diff0, diff1, diff + 32, 8);
- ILVRL_B2_UB(src3, pred3, src_l0, src_l1);
- HSUB_UB2_SH(src_l0, src_l1, diff0, diff1);
- ST_SH2(diff0, diff1, diff + 48, 8);
- diff += diff_stride;
-
- ILVRL_B2_UB(src4, pred4, src_l0, src_l1);
- HSUB_UB2_SH(src_l0, src_l1, diff0, diff1);
- ST_SH2(diff0, diff1, diff, 8);
- ILVRL_B2_UB(src5, pred5, src_l0, src_l1);
- HSUB_UB2_SH(src_l0, src_l1, diff0, diff1);
- ST_SH2(diff0, diff1, diff + 16, 8);
- ILVRL_B2_UB(src6, pred6, src_l0, src_l1);
- HSUB_UB2_SH(src_l0, src_l1, diff0, diff1);
- ST_SH2(diff0, diff1, diff + 32, 8);
- ILVRL_B2_UB(src7, pred7, src_l0, src_l1);
- HSUB_UB2_SH(src_l0, src_l1, diff0, diff1);
- ST_SH2(diff0, diff1, diff + 48, 8);
- diff += diff_stride;
- }
-}
-
-void vp9_subtract_block_msa(int32_t rows, int32_t cols,
- int16_t *diff_ptr, ptrdiff_t diff_stride,
- const uint8_t *src_ptr, ptrdiff_t src_stride,
- const uint8_t *pred_ptr, ptrdiff_t pred_stride) {
- if (rows == cols) {
- switch (rows) {
- case 4:
- sub_blk_4x4_msa(src_ptr, src_stride, pred_ptr, pred_stride,
- diff_ptr, diff_stride);
- break;
- case 8:
- sub_blk_8x8_msa(src_ptr, src_stride, pred_ptr, pred_stride,
- diff_ptr, diff_stride);
- break;
- case 16:
- sub_blk_16x16_msa(src_ptr, src_stride, pred_ptr, pred_stride,
- diff_ptr, diff_stride);
- break;
- case 32:
- sub_blk_32x32_msa(src_ptr, src_stride, pred_ptr, pred_stride,
- diff_ptr, diff_stride);
- break;
- case 64:
- sub_blk_64x64_msa(src_ptr, src_stride, pred_ptr, pred_stride,
- diff_ptr, diff_stride);
- break;
- default:
- vp9_subtract_block_c(rows, cols, diff_ptr, diff_stride, src_ptr,
- src_stride, pred_ptr, pred_stride);
- break;
- }
- } else {
- vp9_subtract_block_c(rows, cols, diff_ptr, diff_stride, src_ptr, src_stride,
- pred_ptr, pred_stride);
- }
-}
diff --git a/vp9/encoder/vp9_encodemb.c b/vp9/encoder/vp9_encodemb.c
index 2829365e5..313094140 100644
--- a/vp9/encoder/vp9_encodemb.c
+++ b/vp9/encoder/vp9_encodemb.c
@@ -11,6 +11,7 @@
#include "./vp9_rtcd.h"
#include "./vpx_config.h"
+#include "./vpx_dsp_rtcd.h"
#include "vpx_mem/vpx_mem.h"
#include "vpx_ports/mem.h"
@@ -31,45 +32,6 @@ struct optimize_ctx {
ENTROPY_CONTEXT tl[MAX_MB_PLANE][16];
};
-void vp9_subtract_block_c(int rows, int cols,
- int16_t *diff, ptrdiff_t diff_stride,
- const uint8_t *src, ptrdiff_t src_stride,
- const uint8_t *pred, ptrdiff_t pred_stride) {
- int r, c;
-
- for (r = 0; r < rows; r++) {
- for (c = 0; c < cols; c++)
- diff[c] = src[c] - pred[c];
-
- diff += diff_stride;
- pred += pred_stride;
- src += src_stride;
- }
-}
-
-#if CONFIG_VP9_HIGHBITDEPTH
-void vp9_highbd_subtract_block_c(int rows, int cols,
- int16_t *diff, ptrdiff_t diff_stride,
- const uint8_t *src8, ptrdiff_t src_stride,
- const uint8_t *pred8, ptrdiff_t pred_stride,
- int bd) {
- int r, c;
- uint16_t *src = CONVERT_TO_SHORTPTR(src8);
- uint16_t *pred = CONVERT_TO_SHORTPTR(pred8);
- (void) bd;
-
- for (r = 0; r < rows; r++) {
- for (c = 0; c < cols; c++) {
- diff[c] = src[c] - pred[c];
- }
-
- diff += diff_stride;
- pred += pred_stride;
- src += src_stride;
- }
-}
-#endif // CONFIG_VP9_HIGHBITDEPTH
-
void vp9_subtract_plane(MACROBLOCK *x, BLOCK_SIZE bsize, int plane) {
struct macroblock_plane *const p = &x->plane[plane];
const struct macroblockd_plane *const pd = &x->e_mbd.plane[plane];
@@ -79,13 +41,13 @@ void vp9_subtract_plane(MACROBLOCK *x, BLOCK_SIZE bsize, int plane) {
#if CONFIG_VP9_HIGHBITDEPTH
if (x->e_mbd.cur_buf->flags & YV12_FLAG_HIGHBITDEPTH) {
- vp9_highbd_subtract_block(bh, bw, p->src_diff, bw, p->src.buf,
+ vpx_highbd_subtract_block(bh, bw, p->src_diff, bw, p->src.buf,
p->src.stride, pd->dst.buf, pd->dst.stride,
x->e_mbd.bd);
return;
}
#endif // CONFIG_VP9_HIGHBITDEPTH
- vp9_subtract_block(bh, bw, p->src_diff, bw, p->src.buf, p->src.stride,
+ vpx_subtract_block(bh, bw, p->src_diff, bw, p->src.buf, p->src.stride,
pd->dst.buf, pd->dst.stride);
}
@@ -838,7 +800,7 @@ void vp9_encode_block_intra(int plane, int block, BLOCK_SIZE plane_bsize,
x->skip_encode ? src_stride : dst_stride,
dst, dst_stride, i, j, plane);
if (!x->skip_recode) {
- vp9_highbd_subtract_block(32, 32, src_diff, diff_stride,
+ vpx_highbd_subtract_block(32, 32, src_diff, diff_stride,
src, src_stride, dst, dst_stride, xd->bd);
highbd_fdct32x32(x->use_lp32x32fdct, src_diff, coeff, diff_stride);
vp9_highbd_quantize_b_32x32(coeff, 1024, x->skip_block, p->zbin,
@@ -859,7 +821,7 @@ void vp9_encode_block_intra(int plane, int block, BLOCK_SIZE plane_bsize,
x->skip_encode ? src_stride : dst_stride,
dst, dst_stride, i, j, plane);
if (!x->skip_recode) {
- vp9_highbd_subtract_block(16, 16, src_diff, diff_stride,
+ vpx_highbd_subtract_block(16, 16, src_diff, diff_stride,
src, src_stride, dst, dst_stride, xd->bd);
vp9_highbd_fht16x16(src_diff, coeff, diff_stride, tx_type);
vp9_highbd_quantize_b(coeff, 256, x->skip_block, p->zbin, p->round,
@@ -881,7 +843,7 @@ void vp9_encode_block_intra(int plane, int block, BLOCK_SIZE plane_bsize,
x->skip_encode ? src_stride : dst_stride,
dst, dst_stride, i, j, plane);
if (!x->skip_recode) {
- vp9_highbd_subtract_block(8, 8, src_diff, diff_stride,
+ vpx_highbd_subtract_block(8, 8, src_diff, diff_stride,
src, src_stride, dst, dst_stride, xd->bd);
vp9_highbd_fht8x8(src_diff, coeff, diff_stride, tx_type);
vp9_highbd_quantize_b(coeff, 64, x->skip_block, p->zbin, p->round,
@@ -904,7 +866,7 @@ void vp9_encode_block_intra(int plane, int block, BLOCK_SIZE plane_bsize,
dst, dst_stride, i, j, plane);
if (!x->skip_recode) {
- vp9_highbd_subtract_block(4, 4, src_diff, diff_stride,
+ vpx_highbd_subtract_block(4, 4, src_diff, diff_stride,
src, src_stride, dst, dst_stride, xd->bd);
if (tx_type != DCT_DCT)
vp9_highbd_fht4x4(src_diff, coeff, diff_stride, tx_type);
@@ -946,7 +908,7 @@ void vp9_encode_block_intra(int plane, int block, BLOCK_SIZE plane_bsize,
x->skip_encode ? src_stride : dst_stride,
dst, dst_stride, i, j, plane);
if (!x->skip_recode) {
- vp9_subtract_block(32, 32, src_diff, diff_stride,
+ vpx_subtract_block(32, 32, src_diff, diff_stride,
src, src_stride, dst, dst_stride);
fdct32x32(x->use_lp32x32fdct, src_diff, coeff, diff_stride);
vp9_quantize_b_32x32(coeff, 1024, x->skip_block, p->zbin, p->round,
@@ -966,7 +928,7 @@ void vp9_encode_block_intra(int plane, int block, BLOCK_SIZE plane_bsize,
x->skip_encode ? src_stride : dst_stride,
dst, dst_stride, i, j, plane);
if (!x->skip_recode) {
- vp9_subtract_block(16, 16, src_diff, diff_stride,
+ vpx_subtract_block(16, 16, src_diff, diff_stride,
src, src_stride, dst, dst_stride);
vp9_fht16x16(src_diff, coeff, diff_stride, tx_type);
vp9_quantize_b(coeff, 256, x->skip_block, p->zbin, p->round,
@@ -986,7 +948,7 @@ void vp9_encode_block_intra(int plane, int block, BLOCK_SIZE plane_bsize,
x->skip_encode ? src_stride : dst_stride,
dst, dst_stride, i, j, plane);
if (!x->skip_recode) {
- vp9_subtract_block(8, 8, src_diff, diff_stride,
+ vpx_subtract_block(8, 8, src_diff, diff_stride,
src, src_stride, dst, dst_stride);
vp9_fht8x8(src_diff, coeff, diff_stride, tx_type);
vp9_quantize_b(coeff, 64, x->skip_block, p->zbin, p->round, p->quant,
@@ -1007,7 +969,7 @@ void vp9_encode_block_intra(int plane, int block, BLOCK_SIZE plane_bsize,
dst, dst_stride, i, j, plane);
if (!x->skip_recode) {
- vp9_subtract_block(4, 4, src_diff, diff_stride,
+ vpx_subtract_block(4, 4, src_diff, diff_stride,
src, src_stride, dst, dst_stride);
if (tx_type != DCT_DCT)
vp9_fht4x4(src_diff, coeff, diff_stride, tx_type);
diff --git a/vp9/encoder/vp9_rdopt.c b/vp9/encoder/vp9_rdopt.c
index e6b7f193a..dc054f0c1 100644
--- a/vp9/encoder/vp9_rdopt.c
+++ b/vp9/encoder/vp9_rdopt.c
@@ -12,6 +12,7 @@
#include <math.h>
#include "./vp9_rtcd.h"
+#include "./vpx_dsp_rtcd.h"
#include "vpx_mem/vpx_mem.h"
#include "vpx_ports/mem.h"
@@ -832,7 +833,7 @@ static int64_t rd_pick_intra4x4block(VP9_COMP *cpi, MACROBLOCK *x, int ib,
x->skip_encode ? src : dst,
x->skip_encode ? src_stride : dst_stride,
dst, dst_stride, idx, idy, 0);
- vp9_highbd_subtract_block(4, 4, src_diff, 8, src, src_stride,
+ vpx_highbd_subtract_block(4, 4, src_diff, 8, src, src_stride,
dst, dst_stride, xd->bd);
if (xd->lossless) {
const scan_order *so = &vp9_default_scan_orders[TX_4X4];
@@ -932,7 +933,7 @@ static int64_t rd_pick_intra4x4block(VP9_COMP *cpi, MACROBLOCK *x, int ib,
x->skip_encode ? src : dst,
x->skip_encode ? src_stride : dst_stride,
dst, dst_stride, idx, idy, 0);
- vp9_subtract_block(4, 4, src_diff, 8, src, src_stride, dst, dst_stride);
+ vpx_subtract_block(4, 4, src_diff, 8, src, src_stride, dst, dst_stride);
if (xd->lossless) {
const scan_order *so = &vp9_default_scan_orders[TX_4X4];
@@ -1394,16 +1395,16 @@ static int64_t encode_inter_mb_segment(VP9_COMP *cpi,
#if CONFIG_VP9_HIGHBITDEPTH
if (xd->cur_buf->flags & YV12_FLAG_HIGHBITDEPTH) {
- vp9_highbd_subtract_block(
+ vpx_highbd_subtract_block(
height, width, vp9_raster_block_offset_int16(BLOCK_8X8, i, p->src_diff),
8, src, p->src.stride, dst, pd->dst.stride, xd->bd);
} else {
- vp9_subtract_block(
+ vpx_subtract_block(
height, width, vp9_raster_block_offset_int16(BLOCK_8X8, i, p->src_diff),
8, src, p->src.stride, dst, pd->dst.stride);
}
#else
- vp9_subtract_block(height, width,
+ vpx_subtract_block(height, width,
vp9_raster_block_offset_int16(BLOCK_8X8, i, p->src_diff),
8, src, p->src.stride, dst, pd->dst.stride);
#endif // CONFIG_VP9_HIGHBITDEPTH
diff --git a/vp9/encoder/x86/vp9_subtract_sse2.asm b/vp9/encoder/x86/vp9_subtract_sse2.asm
deleted file mode 100644
index 982408083..000000000
--- a/vp9/encoder/x86/vp9_subtract_sse2.asm
+++ /dev/null
@@ -1,127 +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 "third_party/x86inc/x86inc.asm"
-
-SECTION .text
-
-; void vp9_subtract_block(int rows, int cols,
-; int16_t *diff, ptrdiff_t diff_stride,
-; const uint8_t *src, ptrdiff_t src_stride,
-; const uint8_t *pred, ptrdiff_t pred_stride)
-
-INIT_XMM sse2
-cglobal subtract_block, 7, 7, 8, \
- rows, cols, diff, diff_stride, src, src_stride, \
- pred, pred_stride
-%define pred_str colsq
- pxor m7, m7 ; dedicated zero register
- cmp colsd, 4
- je .case_4
- cmp colsd, 8
- je .case_8
- cmp colsd, 16
- je .case_16
- cmp colsd, 32
- je .case_32
-
-%macro loop16 6
- mova m0, [srcq+%1]
- mova m4, [srcq+%2]
- mova m1, [predq+%3]
- mova m5, [predq+%4]
- punpckhbw m2, m0, m7
- punpckhbw m3, m1, m7
- punpcklbw m0, m7
- punpcklbw m1, m7
- psubw m2, m3
- psubw m0, m1
- punpckhbw m1, m4, m7
- punpckhbw m3, m5, m7
- punpcklbw m4, m7
- punpcklbw m5, m7
- psubw m1, m3
- psubw m4, m5
- mova [diffq+mmsize*0+%5], m0
- mova [diffq+mmsize*1+%5], m2
- mova [diffq+mmsize*0+%6], m4
- mova [diffq+mmsize*1+%6], m1
-%endmacro
-
- mov pred_str, pred_stridemp
-.loop_64:
- loop16 0*mmsize, 1*mmsize, 0*mmsize, 1*mmsize, 0*mmsize, 2*mmsize
- loop16 2*mmsize, 3*mmsize, 2*mmsize, 3*mmsize, 4*mmsize, 6*mmsize
- lea diffq, [diffq+diff_strideq*2]
- add predq, pred_str
- add srcq, src_strideq
- dec rowsd
- jg .loop_64
- RET
-
-.case_32:
- mov pred_str, pred_stridemp
-.loop_32:
- loop16 0, mmsize, 0, mmsize, 0, 2*mmsize
- lea diffq, [diffq+diff_strideq*2]
- add predq, pred_str
- add srcq, src_strideq
- dec rowsd
- jg .loop_32
- RET
-
-.case_16:
- mov pred_str, pred_stridemp
-.loop_16:
- loop16 0, src_strideq, 0, pred_str, 0, diff_strideq*2
- lea diffq, [diffq+diff_strideq*4]
- lea predq, [predq+pred_str*2]
- lea srcq, [srcq+src_strideq*2]
- sub rowsd, 2
- jg .loop_16
- RET
-
-%macro loop_h 0
- movh m0, [srcq]
- movh m2, [srcq+src_strideq]
- movh m1, [predq]
- movh m3, [predq+pred_str]
- punpcklbw m0, m7
- punpcklbw m1, m7
- punpcklbw m2, m7
- punpcklbw m3, m7
- psubw m0, m1
- psubw m2, m3
- mova [diffq], m0
- mova [diffq+diff_strideq*2], m2
-%endmacro
-
-.case_8:
- mov pred_str, pred_stridemp
-.loop_8:
- loop_h
- lea diffq, [diffq+diff_strideq*4]
- lea srcq, [srcq+src_strideq*2]
- lea predq, [predq+pred_str*2]
- sub rowsd, 2
- jg .loop_8
- RET
-
-INIT_MMX
-.case_4:
- mov pred_str, pred_stridemp
-.loop_4:
- loop_h
- lea diffq, [diffq+diff_strideq*4]
- lea srcq, [srcq+src_strideq*2]
- lea predq, [predq+pred_str*2]
- sub rowsd, 2
- jg .loop_4
- RET
generated by cgit v1.2.3 (git 2.25.1) at 2024-10-10 07:24:01 +0000