Chromium Code Reviews
chromiumcodereview-hr@appspot.gserviceaccount.com (chromiumcodereview-hr) | Please choose your nickname with Settings | Help | Chromium Project | Gerrit Changes | Sign out
(136)

Issues Search
    My Issues | Starred     Open | Closed | All

Unified Diff: third_party/WebKit/Source/platform/graphics/cpu/mips/WebGLImageConversionMSA.h

Issue 2392443003: Add MSA (MIPS SIMD Arch) optimized color conversion functions (Closed)
Patch Set: Created 4 years, 2 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 | « third_party/WebKit/Source/platform/cpu/mips/CommonMacrosMSA.h ('k') | third_party/WebKit/Source/platform/graphics/gpu/WebGLImageConversion.cpp » ('j') | no next file with comments »
Expand Comments ('e') | Collapse Comments ('c') | Show Comments Hide Comments ('s')
Index: third_party/WebKit/Source/platform/graphics/cpu/mips/WebGLImageConversionMSA.h
diff --git a/third_party/WebKit/Source/platform/graphics/cpu/mips/WebGLImageConversionMSA.h b/third_party/WebKit/Source/platform/graphics/cpu/mips/WebGLImageConversionMSA.h
index 1f2dd7f2bc233f91a0c64b4d3f3aff8d33828342..5c24f6995cc4af071e272afab1bb156172c9a678 100644
--- a/third_party/WebKit/Source/platform/graphics/cpu/mips/WebGLImageConversionMSA.h
+++ b/third_party/WebKit/Source/platform/graphics/cpu/mips/WebGLImageConversionMSA.h
@@ -715,6 +715,436 @@ ALWAYS_INLINE void packOneRowOfRGBA8ToUnsignedShort565MSA(
pixelsPerRow &= 7;
}
+
+ALWAYS_INLINE void packOneRowOfRGBA8ToUnsignedShort4444MSA(
+ const uint8_t*& source,
+ uint16_t*& destination,
+ unsigned& pixelsPerRow) {
+ unsigned i;
+ v16u8 src0, src1, src2, src3, src4, src5, src6, src7;
+ v16u8 vec0, vec1, vec2, vec3, vec4, vec5, vec6, vec7;
+ v8u16 dst0, dst1, dst2, dst3;
+
+ for (i = (pixelsPerRow >> 5); i--;) {
+ LD_UB8(source, 16, src0, src1, src2, src3, src4, src5, src6, src7);
+ SRLI_H4_UB(src0, src1, src2, src3, vec0, vec1, vec2, vec3, 12);
+ SRLI_H4_UB(src4, src5, src6, src7, vec4, vec5, vec6, vec7, 12);
+ BINSLI_B2_UB(vec0, src0, vec1, src1, vec0, vec1, 3);
+ BINSLI_B2_UB(vec2, src2, vec3, src3, vec2, vec3, 3);
+ BINSLI_B2_UB(vec4, src4, vec5, src5, vec4, vec5, 3);
+ BINSLI_B2_UB(vec6, src6, vec7, src7, vec6, vec7, 3);
+ PCKEV_B4_UH(vec1, vec0, vec3, vec2, vec5, vec4, vec7, vec6, dst0, dst1,
+ dst2, dst3);
+ SHF_B4_UH(dst0, dst1, dst2, dst3, 177);
+ ST_UH4(dst0, dst1, dst2, dst3, destination, 8);
+ }
+
+ if (pixelsPerRow & 31) {
+ if (pixelsPerRow & 16) {
+ if ((pixelsPerRow & 8) && (pixelsPerRow & 4)) {
+ LD_UB7(source, 16, src0, src1, src2, src3, src4, src5, src6);
+ SRLI_H4_UB(src0, src1, src2, src3, vec0, vec1, vec2, vec3, 12);
+ SRLI_H2_UB(src4, src5, vec4, vec5, 12);
+ vec6 = (v16u8)SRLI_H(src6, 12);
+ BINSLI_B2_UB(vec0, src0, vec1, src1, vec0, vec1, 3);
+ BINSLI_B2_UB(vec2, src2, vec3, src3, vec2, vec3, 3);
+ BINSLI_B2_UB(vec4, src4, vec5, src5, vec4, vec5, 3);
+ vec6 = (v16u8)__msa_binsli_b((v16u8)vec6, (v16u8)src6, 3);
+ PCKEV_B2_UH(vec1, vec0, vec3, vec2, dst0, dst1);
+ PCKEV_B2_UH(vec5, vec4, vec6, vec6, dst2, dst3);
+ SHF_B4_UH(dst0, dst1, dst2, dst3, 177);
+ ST_UH3(dst0, dst1, dst2, destination, 8);
+ ST8x1_UB(dst3, destination);
+ destination += 4;
+ } else if (pixelsPerRow & 8) {
+ LD_UB6(source, 16, src0, src1, src2, src3, src4, src5);
+ SRLI_H4_UB(src0, src1, src2, src3, vec0, vec1, vec2, vec3, 12);
+ SRLI_H2_UB(src4, src5, vec4, vec5, 12);
+ BINSLI_B2_UB(vec0, src0, vec1, src1, vec0, vec1, 3);
+ BINSLI_B2_UB(vec2, src2, vec3, src3, vec2, vec3, 3);
+ BINSLI_B2_UB(vec4, src4, vec5, src5, vec4, vec5, 3);
+ PCKEV_B3_UH(vec1, vec0, vec3, vec2, vec5, vec4, dst0, dst1, dst2);
+ SHF_B3_UH(dst0, dst1, dst2, 177);
+ ST_UH3(dst0, dst1, dst2, destination, 8);
+ } else if (pixelsPerRow & 4) {
+ LD_UB5(source, 16, src0, src1, src2, src3, src4);
+ SRLI_H4_UB(src0, src1, src2, src3, vec0, vec1, vec2, vec3, 12);
+ vec4 = (v16u8)SRLI_H(src4, 12);
+ BINSLI_B2_UB(vec0, src0, vec1, src1, vec0, vec1, 3);
+ BINSLI_B2_UB(vec2, src2, vec3, src3, vec2, vec3, 3);
+ vec4 = (v16u8)__msa_binsli_b((v16u8)vec4, (v16u8)src4, 3);
+ PCKEV_B3_UH(vec1, vec0, vec3, vec2, vec4, vec4, dst0, dst1, dst2);
+ SHF_B3_UH(dst0, dst1, dst2, 177);
+ ST_UH2(dst0, dst1, destination, 8);
+ ST8x1_UB(dst2, destination);
+ destination += 4;
+ } else {
+ LD_UB4(source, 16, src0, src1, src2, src3);
+ SRLI_H4_UB(src0, src1, src2, src3, vec0, vec1, vec2, vec3, 12);
+ BINSLI_B2_UB(vec0, src0, vec1, src1, vec0, vec1, 3);
+ BINSLI_B2_UB(vec2, src2, vec3, src3, vec2, vec3, 3);
+ PCKEV_B2_UH(vec1, vec0, vec3, vec2, dst0, dst1);
+ SHF_B2_UH(dst0, dst1, 177);
+ ST_UH2(dst0, dst1, destination, 8);
+ }
+ } else if ((pixelsPerRow & 8) && (pixelsPerRow & 4)) {
+ LD_UB3(source, 16, src0, src1, src2);
+ SRLI_H2_UB(src0, src1, vec0, vec1, 12);
+ vec2 = (v16u8)SRLI_H(src2, 12);
+ BINSLI_B2_UB(vec0, src0, vec1, src1, vec0, vec1, 3);
+ vec2 = (v16u8)__msa_binsli_b((v16u8)vec2, (v16u8)src2, 3);
+ PCKEV_B2_UH(vec1, vec0, vec2, vec2, dst0, dst1);
+ SHF_B2_UH(dst0, dst1, 177);
+ ST_UH(dst0, destination);
+ destination += 8;
+ ST8x1_UB(dst1, destination);
+ destination += 4;
+ } else if (pixelsPerRow & 16) {
+ LD_UB4(source, 16, src0, src1, src2, src3);
+ SRLI_H4_UB(src0, src1, src2, src3, vec0, vec1, vec2, vec3, 12);
+ BINSLI_B2_UB(vec0, src0, vec1, src1, vec0, vec1, 3);
+ BINSLI_B2_UB(vec2, src2, vec3, src3, vec2, vec3, 3);
+ PCKEV_B2_UH(vec1, vec0, vec3, vec2, dst0, dst1);
+ SHF_B2_UH(dst0, dst1, 177);
+ ST_UH2(dst0, dst1, destination, 8);
+ } else if (pixelsPerRow & 8) {
+ LD_UB2(source, 16, src0, src1);
+ SRLI_H2_UB(src0, src1, vec0, vec1, 12);
+ BINSLI_B2_UB(vec0, src0, vec1, src1, vec0, vec1, 3);
+ dst0 = (v8u16)__msa_pckev_b((v16i8)vec1, (v16i8)vec0);
+ dst0 = (v8u16)__msa_shf_b((v16i8)dst0, 177);
+ ST_UH(dst0, destination);
+ destination += 8;
+ } else if (pixelsPerRow & 4) {
+ src0 = LD_UB(source);
+ source += 16;
+ vec0 = (v16u8)SRLI_H(src0, 12);
+ vec0 = (v16u8)__msa_binsli_b((v16u8)vec0, (v16u8)src0, 3);
+ dst0 = (v8u16)__msa_pckev_b((v16i8)vec0, (v16i8)vec0);
+ dst0 = (v8u16)__msa_shf_b((v16i8)dst0, 177);
+ ST8x1_UB(dst0, destination);
+ destination += 4;
+ }
+ }
+
+ pixelsPerRow &= 3;
+}
+
+ALWAYS_INLINE void packOneRowOfRGBA8LittleToR8MSA(const uint8_t*& source,
+ uint8_t*& destination,
+ unsigned& pixelsPerRow) {
+ unsigned i;
+ v16u8 src0, src1, src2, src3, src4, src5, src6, src7;
+ v16u8 src0A, src1A, src2A, src3A, src4A, src5A, src6A, src7A;
+ v16u8 src0R, src1R, src2R, src3R, src4R, src5R, src6R, src7R;
+ v16u8 dst0, dst1, dst2, dst3, dst4, dst5, dst6, dst7;
+ v4f32 fsrc0A, fsrc1A, fsrc2A, fsrc3A, fsrc4A, fsrc5A, fsrc6A, fsrc7A;
+ v4f32 fsrc0R, fsrc1R, fsrc2R, fsrc3R, fsrc4R, fsrc5R, fsrc6R, fsrc7R;
+ v4f32 fdst0R, fdst1R, fdst2R, fdst3R, fdst4R, fdst5R, fdst6R, fdst7R;
+ const v16u8 alphaMask = {0, 0, 0, 255, 0, 0, 0, 255,
+ 0, 0, 0, 255, 0, 0, 0, 255};
+ const v4u32 vCnst255 = (v4u32)__msa_ldi_w(255);
+ const v4f32 vfCnst255 = __msa_ffint_u_w(vCnst255);
+
+ for (i = (pixelsPerRow >> 5); i--;) {
+ LD_UB8(source, 16, src0, src1, src2, src3, src4, src5, src6, src7);
+ CEQI_B4_UB(src0, src1, src2, src3, 0, src0A, src1A, src2A, src3A);
+ CEQI_B4_UB(src4, src5, src6, src7, 0, src4A, src5A, src6A, src7A);
+ src0A = __msa_bmnz_v(src0, alphaMask, src0A);
+ src1A = __msa_bmnz_v(src1, alphaMask, src1A);
+ src2A = __msa_bmnz_v(src2, alphaMask, src2A);
+ src3A = __msa_bmnz_v(src3, alphaMask, src3A);
+ src4A = __msa_bmnz_v(src4, alphaMask, src4A);
+ src5A = __msa_bmnz_v(src5, alphaMask, src5A);
+ src6A = __msa_bmnz_v(src6, alphaMask, src6A);
+ src7A = __msa_bmnz_v(src7, alphaMask, src7A);
+ AND_V4_UB(src0A, src1A, src2A, src3A, alphaMask, src0A, src1A, src2A,
+ src3A);
+ AND_V4_UB(src4A, src5A, src6A, src7A, alphaMask, src4A, src5A, src6A,
+ src7A);
+ src0A = SLDI_UB(src0A, src0A, 3);
+ src1A = SLDI_UB(src1A, src1A, 3);
+ src2A = SLDI_UB(src2A, src2A, 3);
+ src3A = SLDI_UB(src3A, src3A, 3);
+ src4A = SLDI_UB(src4A, src4A, 3);
+ src5A = SLDI_UB(src5A, src5A, 3);
+ src6A = SLDI_UB(src6A, src6A, 3);
+ src7A = SLDI_UB(src7A, src7A, 3);
+ AND_V4_UB(src0, src1, src2, src3, vCnst255, src0R, src1R, src2R, src3R);
+ AND_V4_UB(src4, src5, src6, src7, vCnst255, src4R, src5R, src6R, src7R);
+ FFINTU_W4_SP(src0A, src1A, src2A, src3A, fsrc0A, fsrc1A, fsrc2A, fsrc3A);
+ FFINTU_W4_SP(src4A, src5A, src6A, src7A, fsrc4A, fsrc5A, fsrc6A, fsrc7A);
+ FFINTU_W4_SP(src0R, src1R, src2R, src3R, fsrc0R, fsrc1R, fsrc2R, fsrc3R);
+ FFINTU_W4_SP(src4R, src5R, src6R, src7R, fsrc4R, fsrc5R, fsrc6R, fsrc7R);
+ DIV4(vfCnst255, fsrc0A, vfCnst255, fsrc1A, vfCnst255, fsrc2A, vfCnst255,
+ fsrc3A, fsrc0A, fsrc1A, fsrc2A, fsrc3A);
+ DIV4(vfCnst255, fsrc4A, vfCnst255, fsrc5A, vfCnst255, fsrc6A, vfCnst255,
+ fsrc7A, fsrc4A, fsrc5A, fsrc6A, fsrc7A);
+ MUL4(fsrc0R, fsrc0A, fsrc1R, fsrc1A, fsrc2R, fsrc2A, fsrc3R, fsrc3A, fdst0R,
+ fdst1R, fdst2R, fdst3R);
+ MUL4(fsrc4R, fsrc4A, fsrc5R, fsrc5A, fsrc6R, fsrc6A, fsrc7R, fsrc7A, fdst4R,
+ fdst5R, fdst6R, fdst7R);
+ FTRUNCU_W4_UB(fdst0R, fdst1R, fdst2R, fdst3R, dst0, dst1, dst2, dst3);
+ FTRUNCU_W4_UB(fdst4R, fdst5R, fdst6R, fdst7R, dst4, dst5, dst6, dst7);
+ PCKEV_H4_UB(dst1, dst0, dst3, dst2, dst5, dst4, dst7, dst6, dst0, dst2,
+ dst4, dst6);
+ PCKEV_B2_UB(dst2, dst0, dst6, dst4, dst0, dst1);
+ ST_UB2(dst0, dst1, destination, 16);
+ }
+
+ if (pixelsPerRow & 31) {
+ if ((pixelsPerRow & 16) && (pixelsPerRow & 8)) {
+ LD_UB6(source, 16, src0, src1, src2, src3, src4, src5);
+ CEQI_B4_UB(src0, src1, src2, src3, 0, src0A, src1A, src2A, src3A);
+ CEQI_B2_UB(src4, src5, 0, src4A, src5A);
+ src0A = __msa_bmnz_v(src0, alphaMask, src0A);
+ src1A = __msa_bmnz_v(src1, alphaMask, src1A);
+ src2A = __msa_bmnz_v(src2, alphaMask, src2A);
+ src3A = __msa_bmnz_v(src3, alphaMask, src3A);
+ src4A = __msa_bmnz_v(src4, alphaMask, src4A);
+ src5A = __msa_bmnz_v(src5, alphaMask, src5A);
+ AND_V4_UB(src0A, src1A, src2A, src3A, alphaMask, src0A, src1A, src2A,
+ src3A);
+ AND_V2_UB(src4A, src5A, alphaMask, src4A, src5A);
+ src0A = SLDI_UB(src0A, src0A, 3);
+ src1A = SLDI_UB(src1A, src1A, 3);
+ src2A = SLDI_UB(src2A, src2A, 3);
+ src3A = SLDI_UB(src3A, src3A, 3);
+ src4A = SLDI_UB(src4A, src4A, 3);
+ src5A = SLDI_UB(src5A, src5A, 3);
+ AND_V4_UB(src0, src1, src2, src3, vCnst255, src0R, src1R, src2R, src3R);
+ AND_V2_UB(src4, src5, vCnst255, src4R, src5R);
+ FFINTU_W4_SP(src0A, src1A, src2A, src3A, fsrc0A, fsrc1A, fsrc2A, fsrc3A);
+ FFINTU_W2_SP(src4A, src5A, fsrc4A, fsrc5A);
+ FFINTU_W4_SP(src0R, src1R, src2R, src3R, fsrc0R, fsrc1R, fsrc2R, fsrc3R);
+ FFINTU_W2_SP(src4R, src5R, fsrc4R, fsrc5R);
+ DIV4(vfCnst255, fsrc0A, vfCnst255, fsrc1A, vfCnst255, fsrc2A, vfCnst255,
+ fsrc3A, fsrc0A, fsrc1A, fsrc2A, fsrc3A);
+ DIV2(vfCnst255, fsrc4A, vfCnst255, fsrc5A, fsrc4A, fsrc5A);
+ MUL4(fsrc0R, fsrc0A, fsrc1R, fsrc1A, fsrc2R, fsrc2A, fsrc3R, fsrc3A,
+ fdst0R, fdst1R, fdst2R, fdst3R);
+ MUL2(fsrc4R, fsrc4A, fsrc5R, fsrc5A, fdst4R, fdst5R);
+ FTRUNCU_W4_UB(fdst0R, fdst1R, fdst2R, fdst3R, dst0, dst1, dst2, dst3);
+ FTRUNCU_W2_UB(fdst4R, fdst5R, dst4, dst5);
+ PCKEV_H3_UB(dst1, dst0, dst3, dst2, dst5, dst4, dst0, dst2, dst4);
+ PCKEV_B2_UB(dst2, dst0, dst4, dst4, dst0, dst1);
+ ST_UB(dst0, destination);
+ destination += 16;
+ ST8x1_UB(dst1, destination);
+ destination += 8;
+ } else if (pixelsPerRow & 16) {
+ LD_UB4(source, 16, src0, src1, src2, src3);
+ CEQI_B4_UB(src0, src1, src2, src3, 0, src0A, src1A, src2A, src3A);
+ src0A = __msa_bmnz_v(src0, alphaMask, src0A);
+ src1A = __msa_bmnz_v(src1, alphaMask, src1A);
+ src2A = __msa_bmnz_v(src2, alphaMask, src2A);
+ src3A = __msa_bmnz_v(src3, alphaMask, src3A);
+ AND_V4_UB(src0A, src1A, src2A, src3A, alphaMask, src0A, src1A, src2A,
+ src3A);
+ src0A = SLDI_UB(src0A, src0A, 3);
+ src1A = SLDI_UB(src1A, src1A, 3);
+ src2A = SLDI_UB(src2A, src2A, 3);
+ src3A = SLDI_UB(src3A, src3A, 3);
+ AND_V4_UB(src0, src1, src2, src3, vCnst255, src0R, src1R, src2R, src3R);
+ FFINTU_W4_SP(src0A, src1A, src2A, src3A, fsrc0A, fsrc1A, fsrc2A, fsrc3A);
+ FFINTU_W4_SP(src0R, src1R, src2R, src3R, fsrc0R, fsrc1R, fsrc2R, fsrc3R);
+ DIV4(vfCnst255, fsrc0A, vfCnst255, fsrc1A, vfCnst255, fsrc2A, vfCnst255,
+ fsrc3A, fsrc0A, fsrc1A, fsrc2A, fsrc3A);
+ MUL4(fsrc0R, fsrc0A, fsrc1R, fsrc1A, fsrc2R, fsrc2A, fsrc3R, fsrc3A,
+ fdst0R, fdst1R, fdst2R, fdst3R);
+ FTRUNCU_W4_UB(fdst0R, fdst1R, fdst2R, fdst3R, dst0, dst1, dst2, dst3);
+ PCKEV_H2_UB(dst1, dst0, dst3, dst2, dst0, dst2);
+ dst0 = (v16u8)__msa_pckev_b((v16i8)dst2, (v16i8)dst0);
+ ST_UB(dst0, destination);
+ destination += 16;
+ } else if (pixelsPerRow & 8) {
+ LD_UB2(source, 16, src0, src1);
+ CEQI_B2_UB(src0, src1, 0, src0A, src1A);
+ src0A = __msa_bmnz_v(src0, alphaMask, src0A);
+ src1A = __msa_bmnz_v(src1, alphaMask, src1A);
+ AND_V2_UB(src0A, src1A, alphaMask, src0A, src1A);
+ src0A = SLDI_UB(src0A, src0A, 3);
+ src1A = SLDI_UB(src1A, src1A, 3);
+ AND_V2_UB(src0, src1, vCnst255, src0R, src1R);
+ FFINTU_W2_SP(src0A, src1A, fsrc0A, fsrc1A);
+ FFINTU_W2_SP(src0R, src1R, fsrc0R, fsrc1R);
+ DIV2(vfCnst255, fsrc0A, vfCnst255, fsrc1A, fsrc0A, fsrc1A);
+ MUL2(fsrc0R, fsrc0A, fsrc1R, fsrc1A, fdst0R, fdst1R);
+ FTRUNCU_W2_UB(fdst0R, fdst1R, dst0, dst1);
+ dst0 = (v16u8)__msa_pckev_h((v8i16)dst1, (v8i16)dst0);
+ dst0 = (v16u8)__msa_pckev_b((v16i8)dst0, (v16i8)dst0);
+ ST8x1_UB(dst0, destination);
+ destination += 8;
+ }
+ }
+
+ pixelsPerRow &= 7;
+}
+
+ALWAYS_INLINE void packOneRowOfRGBA8LittleToRA8MSA(const uint8_t*& source,
+ uint8_t*& destination,
+ unsigned& pixelsPerRow) {
+ unsigned i;
+ v16u8 src0, src1, src2, src3, src4, src5, src6, src7;
+ v16u8 src0A, src1A, src2A, src3A, src4A, src5A, src6A, src7A;
+ v16u8 src0R, src1R, src2R, src3R, src4R, src5R, src6R, src7R;
+ v16u8 dst0, dst1, dst2, dst3, dst4, dst5, dst6, dst7;
+ v4f32 fsrc0A, fsrc1A, fsrc2A, fsrc3A, fsrc4A, fsrc5A, fsrc6A, fsrc7A;
+ v4f32 fsrc0R, fsrc1R, fsrc2R, fsrc3R, fsrc4R, fsrc5R, fsrc6R, fsrc7R;
+ v4f32 fdst0R, fdst1R, fdst2R, fdst3R, fdst4R, fdst5R, fdst6R, fdst7R;
+ const v16u8 alphaMask = {0, 0, 0, 255, 0, 0, 0, 255,
+ 0, 0, 0, 255, 0, 0, 0, 255};
+ const v16i8 vshfm = {0, 19, 4, 23, 8, 27, 12, 31, 0, 0, 0, 0, 0, 0, 0, 0};
+ const v4u32 vCnst255 = (v4u32)__msa_ldi_w(255);
+ const v4f32 vfCnst255 = __msa_ffint_u_w(vCnst255);
+
+ for (i = (pixelsPerRow >> 5); i--;) {
+ LD_UB8(source, 16, src0, src1, src2, src3, src4, src5, src6, src7);
+ CEQI_B4_UB(src0, src1, src2, src3, 0, src0A, src1A, src2A, src3A);
+ CEQI_B4_UB(src4, src5, src6, src7, 0, src4A, src5A, src6A, src7A);
+ src0A = __msa_bmnz_v(src0, alphaMask, src0A);
+ src1A = __msa_bmnz_v(src1, alphaMask, src1A);
+ src2A = __msa_bmnz_v(src2, alphaMask, src2A);
+ src3A = __msa_bmnz_v(src3, alphaMask, src3A);
+ src4A = __msa_bmnz_v(src4, alphaMask, src4A);
+ src5A = __msa_bmnz_v(src5, alphaMask, src5A);
+ src6A = __msa_bmnz_v(src6, alphaMask, src6A);
+ src7A = __msa_bmnz_v(src7, alphaMask, src7A);
+ AND_V4_UB(src0A, src1A, src2A, src3A, alphaMask, src0A, src1A, src2A,
+ src3A);
+ AND_V4_UB(src4A, src5A, src6A, src7A, alphaMask, src4A, src5A, src6A,
+ src7A);
+ src0A = SLDI_UB(src0A, src0A, 3);
+ src1A = SLDI_UB(src1A, src1A, 3);
+ src2A = SLDI_UB(src2A, src2A, 3);
+ src3A = SLDI_UB(src3A, src3A, 3);
+ src4A = SLDI_UB(src4A, src4A, 3);
+ src5A = SLDI_UB(src5A, src5A, 3);
+ src6A = SLDI_UB(src6A, src6A, 3);
+ src7A = SLDI_UB(src7A, src7A, 3);
+ AND_V4_UB(src0, src1, src2, src3, vCnst255, src0R, src1R, src2R, src3R);
+ AND_V4_UB(src4, src5, src6, src7, vCnst255, src4R, src5R, src6R, src7R);
+ FFINTU_W4_SP(src0A, src1A, src2A, src3A, fsrc0A, fsrc1A, fsrc2A, fsrc3A);
+ FFINTU_W4_SP(src4A, src5A, src6A, src7A, fsrc4A, fsrc5A, fsrc6A, fsrc7A);
+ FFINTU_W4_SP(src0R, src1R, src2R, src3R, fsrc0R, fsrc1R, fsrc2R, fsrc3R);
+ FFINTU_W4_SP(src4R, src5R, src6R, src7R, fsrc4R, fsrc5R, fsrc6R, fsrc7R);
+ DIV4(vfCnst255, fsrc0A, vfCnst255, fsrc1A, vfCnst255, fsrc2A, vfCnst255,
+ fsrc3A, fsrc0A, fsrc1A, fsrc2A, fsrc3A);
+ DIV4(vfCnst255, fsrc4A, vfCnst255, fsrc5A, vfCnst255, fsrc6A, vfCnst255,
+ fsrc7A, fsrc4A, fsrc5A, fsrc6A, fsrc7A);
+ MUL4(fsrc0R, fsrc0A, fsrc1R, fsrc1A, fsrc2R, fsrc2A, fsrc3R, fsrc3A, fdst0R,
+ fdst1R, fdst2R, fdst3R);
+ MUL4(fsrc4R, fsrc4A, fsrc5R, fsrc5A, fsrc6R, fsrc6A, fsrc7R, fsrc7A, fdst4R,
+ fdst5R, fdst6R, fdst7R);
+ FTRUNCU_W4_UB(fdst0R, fdst1R, fdst2R, fdst3R, dst0, dst1, dst2, dst3);
+ FTRUNCU_W4_UB(fdst4R, fdst5R, fdst6R, fdst7R, dst4, dst5, dst6, dst7);
+ dst0 = VSHF_UB(dst0, src0, vshfm);
+ dst1 = VSHF_UB(dst1, src1, vshfm);
+ dst2 = VSHF_UB(dst2, src2, vshfm);
+ dst3 = VSHF_UB(dst3, src3, vshfm);
+ dst4 = VSHF_UB(dst4, src4, vshfm);
+ dst5 = VSHF_UB(dst5, src5, vshfm);
+ dst6 = VSHF_UB(dst6, src6, vshfm);
+ dst7 = VSHF_UB(dst7, src7, vshfm);
+ ILVR_D4_UB(dst1, dst0, dst3, dst2, dst5, dst4, dst7, dst6, dst0, dst1, dst2,
+ dst3);
+ ST_UB4(dst0, dst1, dst2, dst3, destination, 16);
+ }
+
+ if (pixelsPerRow & 31) {
+ if ((pixelsPerRow & 16) && (pixelsPerRow & 8)) {
+ LD_UB6(source, 16, src0, src1, src2, src3, src4, src5);
+ CEQI_B4_UB(src0, src1, src2, src3, 0, src0A, src1A, src2A, src3A);
+ CEQI_B2_UB(src4, src5, 0, src4A, src5A);
+ src0A = __msa_bmnz_v(src0, alphaMask, src0A);
+ src1A = __msa_bmnz_v(src1, alphaMask, src1A);
+ src2A = __msa_bmnz_v(src2, alphaMask, src2A);
+ src3A = __msa_bmnz_v(src3, alphaMask, src3A);
+ src4A = __msa_bmnz_v(src4, alphaMask, src4A);
+ src5A = __msa_bmnz_v(src5, alphaMask, src5A);
+ AND_V4_UB(src0A, src1A, src2A, src3A, alphaMask, src0A, src1A, src2A,
+ src3A);
+ AND_V2_UB(src4A, src5A, alphaMask, src4A, src5A);
+ src0A = SLDI_UB(src0A, src0A, 3);
+ src1A = SLDI_UB(src1A, src1A, 3);
+ src2A = SLDI_UB(src2A, src2A, 3);
+ src3A = SLDI_UB(src3A, src3A, 3);
+ src4A = SLDI_UB(src4A, src4A, 3);
+ src5A = SLDI_UB(src5A, src5A, 3);
+ AND_V4_UB(src0, src1, src2, src3, vCnst255, src0R, src1R, src2R, src3R);
+ AND_V2_UB(src4, src5, vCnst255, src4R, src5R);
+ FFINTU_W4_SP(src0A, src1A, src2A, src3A, fsrc0A, fsrc1A, fsrc2A, fsrc3A);
+ FFINTU_W2_SP(src4A, src5A, fsrc4A, fsrc5A);
+ FFINTU_W4_SP(src0R, src1R, src2R, src3R, fsrc0R, fsrc1R, fsrc2R, fsrc3R);
+ FFINTU_W2_SP(src4R, src5R, fsrc4R, fsrc5R);
+ DIV4(vfCnst255, fsrc0A, vfCnst255, fsrc1A, vfCnst255, fsrc2A, vfCnst255,
+ fsrc3A, fsrc0A, fsrc1A, fsrc2A, fsrc3A);
+ DIV2(vfCnst255, fsrc4A, vfCnst255, fsrc5A, fsrc4A, fsrc5A);
+ MUL4(fsrc0R, fsrc0A, fsrc1R, fsrc1A, fsrc2R, fsrc2A, fsrc3R, fsrc3A,
+ fdst0R, fdst1R, fdst2R, fdst3R);
+ MUL2(fsrc4R, fsrc4A, fsrc5R, fsrc5A, fdst4R, fdst5R);
+ FTRUNCU_W4_UB(fdst0R, fdst1R, fdst2R, fdst3R, dst0, dst1, dst2, dst3);
+ FTRUNCU_W2_UB(fdst4R, fdst5R, dst4, dst5);
+ dst0 = VSHF_UB(dst0, src0, vshfm);
+ dst1 = VSHF_UB(dst1, src1, vshfm);
+ dst2 = VSHF_UB(dst2, src2, vshfm);
+ dst3 = VSHF_UB(dst3, src3, vshfm);
+ dst4 = VSHF_UB(dst4, src4, vshfm);
+ dst5 = VSHF_UB(dst5, src5, vshfm);
+ ILVR_D3_UB(dst1, dst0, dst3, dst2, dst5, dst4, dst0, dst1, dst2);
+ ST_UB3(dst0, dst1, dst2, destination, 16);
+ } else if (pixelsPerRow & 16) {
+ LD_UB4(source, 16, src0, src1, src2, src3);
+ CEQI_B4_UB(src0, src1, src2, src3, 0, src0A, src1A, src2A, src3A);
+ src0A = __msa_bmnz_v(src0, alphaMask, src0A);
+ src1A = __msa_bmnz_v(src1, alphaMask, src1A);
+ src2A = __msa_bmnz_v(src2, alphaMask, src2A);
+ src3A = __msa_bmnz_v(src3, alphaMask, src3A);
+ AND_V4_UB(src0A, src1A, src2A, src3A, alphaMask, src0A, src1A, src2A,
+ src3A);
+ src0A = SLDI_UB(src0A, src0A, 3);
+ src1A = SLDI_UB(src1A, src1A, 3);
+ src2A = SLDI_UB(src2A, src2A, 3);
+ src3A = SLDI_UB(src3A, src3A, 3);
+ AND_V4_UB(src0, src1, src2, src3, vCnst255, src0R, src1R, src2R, src3R);
+ FFINTU_W4_SP(src0A, src1A, src2A, src3A, fsrc0A, fsrc1A, fsrc2A, fsrc3A);
+ FFINTU_W4_SP(src0R, src1R, src2R, src3R, fsrc0R, fsrc1R, fsrc2R, fsrc3R);
+ DIV4(vfCnst255, fsrc0A, vfCnst255, fsrc1A, vfCnst255, fsrc2A, vfCnst255,
+ fsrc3A, fsrc0A, fsrc1A, fsrc2A, fsrc3A);
+ MUL4(fsrc0R, fsrc0A, fsrc1R, fsrc1A, fsrc2R, fsrc2A, fsrc3R, fsrc3A,
+ fdst0R, fdst1R, fdst2R, fdst3R);
+ FTRUNCU_W4_UB(fdst0R, fdst1R, fdst2R, fdst3R, dst0, dst1, dst2, dst3);
+ dst0 = VSHF_UB(dst0, src0, vshfm);
+ dst1 = VSHF_UB(dst1, src1, vshfm);
+ dst2 = VSHF_UB(dst2, src2, vshfm);
+ dst3 = VSHF_UB(dst3, src3, vshfm);
+ ILVR_D2_UB(dst1, dst0, dst3, dst2, dst0, dst1);
+ ST_UB2(dst0, dst1, destination, 16);
+ } else if (pixelsPerRow & 8) {
+ LD_UB2(source, 16, src0, src1);
+ CEQI_B2_UB(src0, src1, 0, src0A, src1A);
+ src0A = __msa_bmnz_v(src0, alphaMask, src0A);
+ src1A = __msa_bmnz_v(src1, alphaMask, src1A);
+ AND_V2_UB(src0A, src1A, alphaMask, src0A, src1A);
+ src0A = SLDI_UB(src0A, src0A, 3);
+ src1A = SLDI_UB(src1A, src1A, 3);
+ AND_V2_UB(src0, src1, vCnst255, src0R, src1R);
+ FFINTU_W2_SP(src0A, src1A, fsrc0A, fsrc1A);
+ FFINTU_W2_SP(src0R, src1R, fsrc0R, fsrc1R);
+ DIV2(vfCnst255, fsrc0A, vfCnst255, fsrc1A, fsrc0A, fsrc1A);
+ MUL2(fsrc0R, fsrc0A, fsrc1R, fsrc1A, fdst0R, fdst1R);
+ FTRUNCU_W2_UB(fdst0R, fdst1R, dst0, dst1);
+ dst0 = VSHF_UB(dst0, src0, vshfm);
+ dst1 = VSHF_UB(dst1, src1, vshfm);
+ dst0 = (v16u8)__msa_ilvr_d((v2i64)dst1, (v2i64)dst0);
+ ST_UB(dst0, destination);
+ destination += 16;
+ }
+ }
+
+ pixelsPerRow &= 7;
+}
+
} // namespace SIMD
} // namespace blink
« no previous file with comments | « third_party/WebKit/Source/platform/cpu/mips/CommonMacrosMSA.h ('k') | third_party/WebKit/Source/platform/graphics/gpu/WebGLImageConversion.cpp » ('j') | no next file with comments »

Powered by Google App Engine
This is Rietveld 408576698