source/libvpx/vpx_dsp/mips/vpx_convolve8_avg_horiz_msa.c - Issue 1302353004: libvpx: Pull from upstream

Unified Diff: source/libvpx/vpx_dsp/mips/vpx_convolve8_avg_horiz_msa.c

Issue 1302353004: libvpx: Pull from upstream (Closed) Base URL: https://chromium.googlesource.com/chromium/deps/libvpx.git@master

Patch Set: Created 5 years, 4 months ago

Use n/p to move between diff chunks; N/P to move between comments. Draft comments are only viewable by you.

Jump to:

View side-by-side diff with in-line comments

Download patch

« no previous file with comments | « source/libvpx/vpx_dsp/mips/vpx_convolve8_avg_horiz_dspr2.c ('k') | source/libvpx/vpx_dsp/mips/vpx_convolve8_avg_msa.c » ('j') | no next file with comments »
Expand Comments ('e') | Collapse Comments ('c') | Hide Comments ('s')

Index: source/libvpx/vpx_dsp/mips/vpx_convolve8_avg_horiz_msa.c

diff --git a/source/libvpx/vpx_dsp/mips/vpx_convolve8_avg_horiz_msa.c b/source/libvpx/vpx_dsp/mips/vpx_convolve8_avg_horiz_msa.c

index c5ad08cc9ff506998d87675e7ea2e9b8ec239391..f6244d834b7d4f24acd22c1b961071c62e71e798 100644

--- a/source/libvpx/vpx_dsp/mips/vpx_convolve8_avg_horiz_msa.c

+++ b/source/libvpx/vpx_dsp/mips/vpx_convolve8_avg_horiz_msa.c

@@ -8,6 +8,7 @@

* be found in the AUTHORS file in the root of the source tree.

+#include <assert.h>

#include "./vpx_dsp_rtcd.h"

#include "vpx_dsp/mips/vpx_convolve_msa.h"

@@ -323,7 +324,7 @@ static void common_hz_2t_and_aver_dst_4x4_msa(const uint8_t *src,

int8_t *filter) {

v16i8 src0, src1, src2, src3, mask;

v16u8 filt0, dst0, dst1, dst2, dst3, vec0, vec1, res0, res1;

- v8u16 vec2, vec3, const255, filt;

+ v8u16 vec2, vec3, filt;

mask = LD_SB(&mc_filt_mask_arr[16]);

@@ -331,14 +332,11 @@ static void common_hz_2t_and_aver_dst_4x4_msa(const uint8_t *src,

filt = LD_UH(filter);

filt0 = (v16u8)__msa_splati_h((v8i16)filt, 0);

- const255 = (v8u16)__msa_ldi_h(255);

LD_SB4(src, src_stride, src0, src1, src2, src3);

LD_UB4(dst, dst_stride, dst0, dst1, dst2, dst3);

VSHF_B2_UB(src0, src1, src2, src3, mask, mask, vec0, vec1);

DOTP_UB2_UH(vec0, vec1, filt0, filt0, vec2, vec3);

SRARI_H2_UH(vec2, vec3, FILTER_BITS);

- MIN_UH2_UH(vec2, vec3, const255);

PCKEV_B2_UB(vec2, vec2, vec3, vec3, res0, res1);

ILVR_W2_UB(dst1, dst0, dst3, dst2, dst0, dst2);

AVER_UB2_UB(res0, dst0, res1, dst2, res0, res1);

@@ -353,7 +351,7 @@ static void common_hz_2t_and_aver_dst_4x8_msa(const uint8_t *src,

v16i8 src0, src1, src2, src3, src4, src5, src6, src7, mask;

v16u8 filt0, vec0, vec1, vec2, vec3, res0, res1, res2, res3;

v16u8 dst0, dst1, dst2, dst3, dst4, dst5, dst6, dst7;

- v8u16 vec4, vec5, vec6, vec7, const255, filt;

+ v8u16 vec4, vec5, vec6, vec7, filt;

mask = LD_SB(&mc_filt_mask_arr[16]);

@@ -361,8 +359,6 @@ static void common_hz_2t_and_aver_dst_4x8_msa(const uint8_t *src,

filt = LD_UH(filter);

filt0 = (v16u8)__msa_splati_h((v8i16)filt, 0);

- const255 = (v8u16)__msa_ldi_h(255);

LD_SB8(src, src_stride, src0, src1, src2, src3, src4, src5, src6, src7);

LD_UB8(dst, dst_stride, dst0, dst1, dst2, dst3, dst4, dst5, dst6, dst7);

VSHF_B2_UB(src0, src1, src2, src3, mask, mask, vec0, vec1);

@@ -370,7 +366,6 @@ static void common_hz_2t_and_aver_dst_4x8_msa(const uint8_t *src,

DOTP_UB4_UH(vec0, vec1, vec2, vec3, filt0, filt0, filt0, filt0, vec4, vec5,

vec6, vec7);

SRARI_H4_UH(vec4, vec5, vec6, vec7, FILTER_BITS);

- MIN_UH4_UH(vec4, vec5, vec6, vec7, const255);

PCKEV_B4_UB(vec4, vec4, vec5, vec5, vec6, vec6, vec7, vec7, res0, res1, res2,

res3);

ILVR_W4_UB(dst1, dst0, dst3, dst2, dst5, dst4, dst7, dst6, dst0, dst2, dst4,

@@ -402,7 +397,7 @@ static void common_hz_2t_and_aver_dst_8x4_msa(const uint8_t *src,

int8_t *filter) {

v16i8 src0, src1, src2, src3, mask;

v16u8 filt0, dst0, dst1, dst2, dst3;

- v8u16 vec0, vec1, vec2, vec3, const255, filt;

+ v8u16 vec0, vec1, vec2, vec3, filt;

mask = LD_SB(&mc_filt_mask_arr[0]);

@@ -410,8 +405,6 @@ static void common_hz_2t_and_aver_dst_8x4_msa(const uint8_t *src,

filt = LD_UH(filter);

filt0 = (v16u8)__msa_splati_h((v8i16)filt, 0);

- const255 = (v8u16)__msa_ldi_h(255);

LD_SB4(src, src_stride, src0, src1, src2, src3);

VSHF_B2_UH(src0, src0, src1, src1, mask, mask, vec0, vec1);

VSHF_B2_UH(src2, src2, src3, src3, mask, mask, vec2, vec3);

@@ -419,7 +412,6 @@ static void common_hz_2t_and_aver_dst_8x4_msa(const uint8_t *src,

vec2, vec3);

SRARI_H4_UH(vec0, vec1, vec2, vec3, FILTER_BITS);

LD_UB4(dst, dst_stride, dst0, dst1, dst2, dst3);

- MIN_UH4_UH(vec0, vec1, vec2, vec3, const255);

PCKEV_AVG_ST8x4_UB(vec0, dst0, vec1, dst1, vec2, dst2, vec3, dst3,

dst, dst_stride);

}

@@ -432,7 +424,7 @@ static void common_hz_2t_and_aver_dst_8x8mult_msa(const uint8_t *src,

int32_t height) {

v16i8 src0, src1, src2, src3, mask;

v16u8 filt0, dst0, dst1, dst2, dst3;

- v8u16 vec0, vec1, vec2, vec3, const255, filt;

+ v8u16 vec0, vec1, vec2, vec3, filt;

mask = LD_SB(&mc_filt_mask_arr[0]);

@@ -440,8 +432,6 @@ static void common_hz_2t_and_aver_dst_8x8mult_msa(const uint8_t *src,

filt = LD_UH(filter);

filt0 = (v16u8)__msa_splati_h((v8i16)filt, 0);

- const255 = (v8u16)__msa_ldi_h(255);

LD_SB4(src, src_stride, src0, src1, src2, src3);

src += (4 * src_stride);

VSHF_B2_UH(src0, src0, src1, src1, mask, mask, vec0, vec1);

@@ -450,7 +440,6 @@ static void common_hz_2t_and_aver_dst_8x8mult_msa(const uint8_t *src,

vec2, vec3);

SRARI_H4_UH(vec0, vec1, vec2, vec3, FILTER_BITS);

LD_UB4(dst, dst_stride, dst0, dst1, dst2, dst3);

- MIN_UH4_UH(vec0, vec1, vec2, vec3, const255);

LD_SB4(src, src_stride, src0, src1, src2, src3);

src += (4 * src_stride);

PCKEV_AVG_ST8x4_UB(vec0, dst0, vec1, dst1, vec2, dst2, vec3, dst3,

@@ -463,7 +452,6 @@ static void common_hz_2t_and_aver_dst_8x8mult_msa(const uint8_t *src,

vec2, vec3);

SRARI_H4_UH(vec0, vec1, vec2, vec3, FILTER_BITS);

LD_UB4(dst, dst_stride, dst0, dst1, dst2, dst3);

- MIN_UH4_UH(vec0, vec1, vec2, vec3, const255);

PCKEV_AVG_ST8x4_UB(vec0, dst0, vec1, dst1, vec2, dst2, vec3, dst3,

dst, dst_stride);

dst += (4 * dst_stride);

@@ -478,7 +466,6 @@ static void common_hz_2t_and_aver_dst_8x8mult_msa(const uint8_t *src,

vec2, vec3);

SRARI_H4_UH(vec0, vec1, vec2, vec3, FILTER_BITS);

LD_UB4(dst, dst_stride, dst0, dst1, dst2, dst3);

- MIN_UH4_UH(vec0, vec1, vec2, vec3, const255);

LD_SB4(src, src_stride, src0, src1, src2, src3);

PCKEV_AVG_ST8x4_UB(vec0, dst0, vec1, dst1, vec2, dst2, vec3, dst3,

dst, dst_stride);

@@ -490,7 +477,6 @@ static void common_hz_2t_and_aver_dst_8x8mult_msa(const uint8_t *src,

vec2, vec3);

SRARI_H4_UH(vec0, vec1, vec2, vec3, FILTER_BITS);

LD_UB4(dst, dst_stride, dst0, dst1, dst2, dst3);

- MIN_UH4_UH(vec0, vec1, vec2, vec3, const255);

PCKEV_AVG_ST8x4_UB(vec0, dst0, vec1, dst1, vec2, dst2, vec3, dst3,

dst, dst_stride);

}

@@ -520,7 +506,7 @@ static void common_hz_2t_and_aver_dst_16w_msa(const uint8_t *src,

v16i8 src0, src1, src2, src3, src4, src5, src6, src7, mask;

v16u8 filt0, dst0, dst1, dst2, dst3;

v16u8 vec0, vec1, vec2, vec3, vec4, vec5, vec6, vec7;

- v8u16 res0, res1, res2, res3, res4, res5, res6, res7, const255, filt;

+ v8u16 res0, res1, res2, res3, res4, res5, res6, res7, filt;

mask = LD_SB(&mc_filt_mask_arr[0]);

@@ -528,8 +514,6 @@ static void common_hz_2t_and_aver_dst_16w_msa(const uint8_t *src,

filt = LD_UH(filter);

filt0 = (v16u8)__msa_splati_h((v8i16)filt, 0);

- const255 = (v8u16)__msa_ldi_h(255);

LD_SB4(src, src_stride, src0, src2, src4, src6);

LD_SB4(src + 8, src_stride, src1, src3, src5, src7);

src += (4 * src_stride);

@@ -545,8 +529,6 @@ static void common_hz_2t_and_aver_dst_16w_msa(const uint8_t *src,

SRARI_H4_UH(res0, res1, res2, res3, FILTER_BITS);

SRARI_H4_UH(res4, res5, res6, res7, FILTER_BITS);

LD_UB4(dst, dst_stride, dst0, dst1, dst2, dst3);

- MIN_UH4_UH(res0, res1, res2, res3, const255);

- MIN_UH4_UH(res4, res5, res6, res7, const255);

PCKEV_AVG_ST_UB(res1, res0, dst0, dst);

dst += dst_stride;

PCKEV_AVG_ST_UB(res3, res2, dst1, dst);

@@ -572,8 +554,6 @@ static void common_hz_2t_and_aver_dst_16w_msa(const uint8_t *src,

SRARI_H4_UH(res0, res1, res2, res3, FILTER_BITS);

SRARI_H4_UH(res4, res5, res6, res7, FILTER_BITS);

LD_UB4(dst, dst_stride, dst0, dst1, dst2, dst3);

- MIN_UH4_UH(res0, res1, res2, res3, const255);

- MIN_UH4_UH(res4, res5, res6, res7, const255);

PCKEV_AVG_ST_UB(res1, res0, dst0, dst);

dst += dst_stride;

PCKEV_AVG_ST_UB(res3, res2, dst1, dst);

@@ -595,7 +575,7 @@ static void common_hz_2t_and_aver_dst_32w_msa(const uint8_t *src,

v16i8 src0, src1, src2, src3, src4, src5, src6, src7, mask;

v16u8 filt0, dst0, dst1, dst2, dst3;

v16u8 vec0, vec1, vec2, vec3, vec4, vec5, vec6, vec7;

- v8u16 res0, res1, res2, res3, res4, res5, res6, res7, const255, filt;

+ v8u16 res0, res1, res2, res3, res4, res5, res6, res7, filt;

mask = LD_SB(&mc_filt_mask_arr[0]);

@@ -603,8 +583,6 @@ static void common_hz_2t_and_aver_dst_32w_msa(const uint8_t *src,

filt = LD_UH(filter);

filt0 = (v16u8)__msa_splati_h((v8i16)filt, 0);

- const255 = (v8u16)__msa_ldi_h(255);

for (loop_cnt = (height >> 1); loop_cnt--;) {

src0 = LD_SB(src);

src2 = LD_SB(src + 16);

@@ -627,8 +605,6 @@ static void common_hz_2t_and_aver_dst_32w_msa(const uint8_t *src,

res6, res7);

SRARI_H4_UH(res0, res1, res2, res3, FILTER_BITS);

SRARI_H4_UH(res4, res5, res6, res7, FILTER_BITS);

- MIN_UH4_UH(res0, res1, res2, res3, const255);

- MIN_UH4_UH(res4, res5, res6, res7, const255);

LD_UB2(dst, 16, dst0, dst1);

PCKEV_AVG_ST_UB(res1, res0, dst0, dst);

PCKEV_AVG_ST_UB(res3, res2, dst1, (dst + 16));

@@ -650,7 +626,7 @@ static void common_hz_2t_and_aver_dst_64w_msa(const uint8_t *src,

v16i8 src0, src1, src2, src3, src4, src5, src6, src7, mask;

v16u8 filt0, dst0, dst1, dst2, dst3;

v16u8 vec0, vec1, vec2, vec3, vec4, vec5, vec6, vec7;

- v8u16 out0, out1, out2, out3, out4, out5, out6, out7, const255, filt;

+ v8u16 out0, out1, out2, out3, out4, out5, out6, out7, filt;

mask = LD_SB(&mc_filt_mask_arr[0]);

@@ -658,8 +634,6 @@ static void common_hz_2t_and_aver_dst_64w_msa(const uint8_t *src,

filt = LD_UH(filter);

filt0 = (v16u8)__msa_splati_h((v8i16)filt, 0);

- const255 = (v8u16)__msa_ldi_h(255);

for (loop_cnt = height; loop_cnt--;) {

LD_SB4(src, 16, src0, src2, src4, src6);

src7 = LD_SB(src + 56);

@@ -677,8 +651,6 @@ static void common_hz_2t_and_aver_dst_64w_msa(const uint8_t *src,

SRARI_H4_UH(out0, out1, out2, out3, FILTER_BITS);

SRARI_H4_UH(out4, out5, out6, out7, FILTER_BITS);

LD_UB4(dst, 16, dst0, dst1, dst2, dst3);

- MIN_UH4_UH(out0, out1, out2, out3, const255);

- MIN_UH4_UH(out4, out5, out6, out7, const255);

PCKEV_AVG_ST_UB(out1, out0, dst0, dst);

PCKEV_AVG_ST_UB(out3, out2, dst1, dst + 16);

PCKEV_AVG_ST_UB(out5, out4, dst2, dst + 32);

@@ -694,19 +666,8 @@ void vpx_convolve8_avg_horiz_msa(const uint8_t *src, ptrdiff_t src_stride,

int w, int h) {

int8_t cnt, filt_hor[8];

- if (16 != x_step_q4) {

- vpx_convolve8_avg_horiz_c(src, src_stride, dst, dst_stride,

- filter_x, x_step_q4, filter_y, y_step_q4,

- w, h);

- return;

- }

- if (((const int32_t *)filter_x)[1] == 0x800000) {

- vpx_convolve_avg(src, src_stride, dst, dst_stride,

- filter_x, x_step_q4, filter_y, y_step_q4,

- w, h);

- return;

- }

+ assert(x_step_q4 == 16);

+ assert(((const int32_t *)filter_x)[1] != 0x800000);

for (cnt = 0; cnt < 8; ++cnt) {

filt_hor[cnt] = filter_x[cnt];

« no previous file with comments | « source/libvpx/vpx_dsp/mips/vpx_convolve8_avg_horiz_dspr2.c ('k') | source/libvpx/vpx_dsp/mips/vpx_convolve8_avg_msa.c » ('j') | no next file with comments »