diff options
| author |  commit-bot@chromium.org <commit-bot@chromium.org@2bbb7eff-a529-9590-31e7-b0007b416f81> | 2014-04-01 14:01:32 +0000 |
|---|---|---|
| committer | commit-bot@chromium.org <commit-bot@chromium.org@2bbb7eff-a529-9590-31e7-b0007b416f81> | 2014-04-01 14:01:32 +0000 |
| commit | 25f7455f3a7cf2c440509bead85486079f1e4b31 (patch) | |
| tree | d9f77e2f326b21f340a42b2dd2072e7dc97d29e4 /src/opts/SkBlitRow_opts_SSE2.cpp | |
| parent | 96050eb5a16059238d060f0db7d2a2e48d190c29 (diff) | |
Xfermode: SSE2 implementation of multiply_modeproc
This patch implements basics for Xfermode SSE optimization. Based on
these basics, SSE2 implementation of multiply_modeproc is provided. SSE2
implementation for other modes will come in future. With this patch
performance of Xfermode_Multiply will improve about 45%. Here are the
data on desktop i7-3770.
before:
Xfermode_Multiply 8888: cmsecs = 33.30 565: cmsecs = 45.65
after:
Xfermode_Multiply 8888: cmsecs = 17.18 565: cmsecs = 24.87
BUG=
R=mtklein@google.com
Author: qiankun.miao@intel.com
Review URL: https://codereview.chromium.org/202903004
git-svn-id: http://skia.googlecode.com/svn/trunk@14006 2bbb7eff-a529-9590-31e7-b0007b416f81
Diffstat (limited to 'src/opts/SkBlitRow_opts_SSE2.cpp')
| -rw-r--r-- | src/opts/SkBlitRow_opts_SSE2.cpp | 40 |
1 files changed, 20 insertions, 20 deletions
diff --git a/src/opts/SkBlitRow_opts_SSE2.cpp b/src/opts/SkBlitRow_opts_SSE2.cpp index 4aa08e8165..d1474f4a67 100644 --- a/src/opts/SkBlitRow_opts_SSE2.cpp +++ b/src/opts/SkBlitRow_opts_SSE2.cpp @@ -914,7 +914,7 @@ void S32_D565_Opaque_SSE2(uint16_t* SK_RESTRICT dst, __m128i b = _mm_packs_epi32(b1, b2); // Store 8 16-bit colors in dst. - __m128i d_pixel = SkPackRGB16_SSE(r, g, b); + __m128i d_pixel = SkPackRGB16_SSE2(r, g, b); _mm_store_si128(d++, d_pixel); count -= 8; } @@ -983,54 +983,54 @@ void S32A_D565_Opaque_SSE2(uint16_t* SK_RESTRICT dst, __m128i dst_pixel = _mm_load_si128(d); // Extract A from src. - __m128i sa1 = _mm_slli_epi32(src_pixel1,(24 - SK_A32_SHIFT)); + __m128i sa1 = _mm_slli_epi32(src_pixel1, (24 - SK_A32_SHIFT)); sa1 = _mm_srli_epi32(sa1, 24); - __m128i sa2 = _mm_slli_epi32(src_pixel2,(24 - SK_A32_SHIFT)); + __m128i sa2 = _mm_slli_epi32(src_pixel2, (24 - SK_A32_SHIFT)); sa2 = _mm_srli_epi32(sa2, 24); __m128i sa = _mm_packs_epi32(sa1, sa2); // Extract R from src. - __m128i sr1 = _mm_slli_epi32(src_pixel1,(24 - SK_R32_SHIFT)); + __m128i sr1 = _mm_slli_epi32(src_pixel1, (24 - SK_R32_SHIFT)); sr1 = _mm_srli_epi32(sr1, 24); - __m128i sr2 = _mm_slli_epi32(src_pixel2,(24 - SK_R32_SHIFT)); + __m128i sr2 = _mm_slli_epi32(src_pixel2, (24 - SK_R32_SHIFT)); sr2 = _mm_srli_epi32(sr2, 24); __m128i sr = _mm_packs_epi32(sr1, sr2); // Extract G from src. - __m128i sg1 = _mm_slli_epi32(src_pixel1,(24 - SK_G32_SHIFT)); + __m128i sg1 = _mm_slli_epi32(src_pixel1, (24 - SK_G32_SHIFT)); sg1 = _mm_srli_epi32(sg1, 24); - __m128i sg2 = _mm_slli_epi32(src_pixel2,(24 - SK_G32_SHIFT)); + __m128i sg2 = _mm_slli_epi32(src_pixel2, (24 - SK_G32_SHIFT)); sg2 = _mm_srli_epi32(sg2, 24); __m128i sg = _mm_packs_epi32(sg1, sg2); // Extract B from src. - __m128i sb1 = _mm_slli_epi32(src_pixel1,(24 - SK_B32_SHIFT)); + __m128i sb1 = _mm_slli_epi32(src_pixel1, (24 - SK_B32_SHIFT)); sb1 = _mm_srli_epi32(sb1, 24); - __m128i sb2 = _mm_slli_epi32(src_pixel2,(24 - SK_B32_SHIFT)); + __m128i sb2 = _mm_slli_epi32(src_pixel2, (24 - SK_B32_SHIFT)); sb2 = _mm_srli_epi32(sb2, 24); __m128i sb = _mm_packs_epi32(sb1, sb2); // Extract R G B from dst. - __m128i dr = _mm_srli_epi16(dst_pixel,SK_R16_SHIFT); + __m128i dr = _mm_srli_epi16(dst_pixel, SK_R16_SHIFT); dr = _mm_and_si128(dr, r16_mask); - __m128i dg = _mm_srli_epi16(dst_pixel,SK_G16_SHIFT); + __m128i dg = _mm_srli_epi16(dst_pixel, SK_G16_SHIFT); dg = _mm_and_si128(dg, g16_mask); - __m128i db = _mm_srli_epi16(dst_pixel,SK_B16_SHIFT); + __m128i db = _mm_srli_epi16(dst_pixel, SK_B16_SHIFT); db = _mm_and_si128(db, b16_mask); __m128i isa = _mm_sub_epi16(var255, sa); // 255 -sa // Calculate R G B of result. // Original algorithm is in SkSrcOver32To16(). - dr = _mm_add_epi16(sr, SkMul16ShiftRound_SSE(dr, isa, SK_R16_BITS)); + dr = _mm_add_epi16(sr, SkMul16ShiftRound_SSE2(dr, isa, SK_R16_BITS)); dr = _mm_srli_epi16(dr, 8 - SK_R16_BITS); - dg = _mm_add_epi16(sg, SkMul16ShiftRound_SSE(dg, isa, SK_G16_BITS)); + dg = _mm_add_epi16(sg, SkMul16ShiftRound_SSE2(dg, isa, SK_G16_BITS)); dg = _mm_srli_epi16(dg, 8 - SK_G16_BITS); - db = _mm_add_epi16(sb, SkMul16ShiftRound_SSE(db, isa, SK_B16_BITS)); + db = _mm_add_epi16(sb, SkMul16ShiftRound_SSE2(db, isa, SK_B16_BITS)); db = _mm_srli_epi16(db, 8 - SK_B16_BITS); // Pack R G B into 16-bit color. - __m128i d_pixel = SkPackRGB16_SSE(dr, dg, db); + __m128i d_pixel = SkPackRGB16_SSE2(dr, dg, db); // Store 8 16-bit colors in dst. _mm_store_si128(d++, d_pixel); @@ -1143,7 +1143,7 @@ void S32_D565_Opaque_Dither_SSE2(uint16_t* SK_RESTRICT dst, sb = _mm_srli_epi16(sb, SK_B32_BITS - SK_B16_BITS); // Pack and store 16-bit dst pixel. - __m128i d_pixel = SkPackRGB16_SSE(sr, sg, sb); + __m128i d_pixel = SkPackRGB16_SSE2(sr, sg, sb); _mm_store_si128(d++, d_pixel); count -= 8; @@ -1242,9 +1242,9 @@ void S32A_D565_Opaque_Dither_SSE2(uint16_t* SK_RESTRICT dst, __m128i dst_pixel = _mm_load_si128(d); // Extract A from src. - __m128i sa1 = _mm_slli_epi32(src_pixel1,(24 - SK_A32_SHIFT)); + __m128i sa1 = _mm_slli_epi32(src_pixel1, (24 - SK_A32_SHIFT)); sa1 = _mm_srli_epi32(sa1, 24); - __m128i sa2 = _mm_slli_epi32(src_pixel2,(24 - SK_A32_SHIFT)); + __m128i sa2 = _mm_slli_epi32(src_pixel2, (24 - SK_A32_SHIFT)); sa2 = _mm_srli_epi32(sa2, 24); __m128i sa = _mm_packs_epi32(sa1, sa2); @@ -1323,7 +1323,7 @@ void S32A_D565_Opaque_Dither_SSE2(uint16_t* SK_RESTRICT dst, db = _mm_srli_epi16(db, 5); // Package and store dst pixel. - __m128i d_pixel = SkPackRGB16_SSE(dr, dg, db); + __m128i d_pixel = SkPackRGB16_SSE2(dr, dg, db); _mm_store_si128(d++, d_pixel); count -= 8; |