Chromium Code Reviews
chromiumcodereview-hr@appspot.gserviceaccount.com (chromiumcodereview-hr) | Please choose your nickname with Settings | Help | Chromium Project | Gerrit Changes | Sign out
(777)

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

Issue 2289693002: Add MIPS SIMD Arch (MSA) optimized WebGL image conversion function (Closed)
Patch Set: Fixed merge conflicts Created 4 years, 3 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
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
new file mode 100644
index 0000000000000000000000000000000000000000..0019526a7e93184c3a3523716b27a349918b4f56
--- /dev/null
+++ b/third_party/WebKit/Source/platform/graphics/cpu/mips/WebGLImageConversionMSA.h
@@ -0,0 +1,108 @@
+// Copyright 2016 The Chromium Authors. All rights reserved.
+// Use of this source code is governed by a BSD-style license that can be
+// found in the LICENSE file.
+
+#ifndef WebGLImageConversionMSA_h
+#define WebGLImageConversionMSA_h
+
+#if HAVE(MIPS_MSA_INTRINSICS)
+
+#include "platform/cpu/mips/CommonMacrosMSA.h"
+
+namespace blink {
+
+namespace SIMD {
+
+#define SEPERATE_RGBA_FRM_16BIT_5551INPUT(in, out_r, out_g, out_b, out_a) \
+ cnst31 = (v8u16)__msa_ldi_h(0x1F); \
+ cnst7 = (v8u16)__msa_ldi_h(0x7); \
+ cnst1 = (v8u16)__msa_ldi_h(0x1); \
+ out_r = (v8u16)SRLI_H(in, 11); \
+ out_g = ((v8u16)SRLI_H(in, 6)) & cnst31; \
+ out_b = ((v8u16)SRLI_H(in, 1)) & cnst31; \
+ out_a = in & cnst1; \
+ out_r = ((v8u16)SLLI_H(out_r, 3)) | (out_r & cnst7); \
+ out_g = ((v8u16)SLLI_H(out_g, 3)) | (out_g & cnst7); \
+ out_b = ((v8u16)SLLI_H(out_b, 3)) | (out_b & cnst7); \
+ out_a = (v8u16)CEQI_H((v8i16)out_a, 1); \
+
+ALWAYS_INLINE void unpackOneRowOfRGBA5551ToRGBA8MSA(const uint16_t*& source, uint8_t*& destination, unsigned& pixelsPerRow)
+{
+ unsigned i;
+ v8u16 src0, src1, src2, src3;
+ v8u16 src0r, src0g, src0b, src0a, src1r, src1g, src1b, src1a;
+ v8u16 src2r, src2g, src2b, src2a, src3r, src3g, src3b, src3a;
+ v8u16 cnst31, cnst7, cnst1;
+ v16u8 dst0, dst1, dst2, dst3, dst4, dst5, dst6, dst7;
+ v16u8 dst8, dst9, dst10, dst11, dst12, dst13, dst14, dst15;
+ v16u8 out0, out1, out2, out3, out4, out5, out6, out7;
+
+ for (i = (pixelsPerRow >> 5); i--;) {
+ LD_UH4(source, 8, src0, src1, src2, src3);
+ SEPERATE_RGBA_FRM_16BIT_5551INPUT(src0, src0r, src0g, src0b, src0a);
+ SEPERATE_RGBA_FRM_16BIT_5551INPUT(src1, src1r, src1g, src1b, src1a);
+ SEPERATE_RGBA_FRM_16BIT_5551INPUT(src2, src2r, src2g, src2b, src2a);
+ SEPERATE_RGBA_FRM_16BIT_5551INPUT(src3, src3r, src3g, src3b, src3a);
+ ILVRL_B2_UB(src0g, src0r, dst0, dst1);
+ ILVRL_B2_UB(src0a, src0b, dst2, dst3);
+ ILVRL_B2_UB(src1g, src1r, dst4, dst5);
+ ILVRL_B2_UB(src1a, src1b, dst6, dst7);
+ ILVRL_B2_UB(src2g, src2r, dst8, dst9);
+ ILVRL_B2_UB(src2a, src2b, dst10, dst11);
+ ILVRL_B2_UB(src3g, src3r, dst12, dst13);
+ ILVRL_B2_UB(src3a, src3b, dst14, dst15);
+ ILVEV_H2_UB(dst0, dst2, dst1, dst3, out0, out1);
+ ILVEV_H2_UB(dst4, dst6, dst5, dst7, out2, out3);
+ ILVEV_H2_UB(dst8, dst10, dst9, dst11, out4, out5);
+ ILVEV_H2_UB(dst12, dst14, dst13, dst15, out6, out7);
+ ST_UB8(out0, out1, out2, out3, out4, out5, out6, out7, destination, 16);
+ }
+
+ if (pixelsPerRow & 31) {
+ if ((pixelsPerRow & 16) && (pixelsPerRow & 8)) {
+ LD_UH3(source, 8, src0, src1, src2);
+ SEPERATE_RGBA_FRM_16BIT_5551INPUT(src0, src0r, src0g, src0b, src0a);
+ SEPERATE_RGBA_FRM_16BIT_5551INPUT(src1, src1r, src1g, src1b, src1a);
+ SEPERATE_RGBA_FRM_16BIT_5551INPUT(src2, src2r, src2g, src2b, src2a);
+ ILVRL_B2_UB(src0g, src0r, dst0, dst1);
+ ILVRL_B2_UB(src0a, src0b, dst2, dst3);
+ ILVRL_B2_UB(src1g, src1r, dst4, dst5);
+ ILVRL_B2_UB(src1a, src1b, dst6, dst7);
+ ILVRL_B2_UB(src2g, src2r, dst8, dst9);
+ ILVRL_B2_UB(src2a, src2b, dst10, dst11);
+ ILVEV_H2_UB(dst0, dst2, dst1, dst3, out0, out1);
+ ILVEV_H2_UB(dst4, dst6, dst5, dst7, out2, out3);
+ ILVEV_H2_UB(dst8, dst10, dst9, dst11, out4, out5);
+ ST_UB6(out0, out1, out2, out3, out4, out5, destination, 16);
+ } else if (pixelsPerRow & 16) {
+ LD_UH2(source, 8, src0, src1);
+ SEPERATE_RGBA_FRM_16BIT_5551INPUT(src0, src0r, src0g, src0b, src0a);
+ SEPERATE_RGBA_FRM_16BIT_5551INPUT(src1, src1r, src1g, src1b, src1a);
+ ILVRL_B2_UB(src0g, src0r, dst0, dst1);
+ ILVRL_B2_UB(src0a, src0b, dst2, dst3);
+ ILVRL_B2_UB(src1g, src1r, dst4, dst5);
+ ILVRL_B2_UB(src1a, src1b, dst6, dst7);
+ ILVEV_H2_UB(dst0, dst2, dst1, dst3, out0, out1);
+ ILVEV_H2_UB(dst4, dst6, dst5, dst7, out2, out3);
+ ST_UB4(out0, out1, out2, out3, destination, 16);
+ } else if (pixelsPerRow & 8) {
+ src0 = LD_UH(source);
+ source += 8;
+ SEPERATE_RGBA_FRM_16BIT_5551INPUT(src0, src0r, src0g, src0b, src0a);
+ ILVRL_B2_UB(src0g, src0r, dst0, dst1);
+ ILVRL_B2_UB(src0a, src0b, dst2, dst3);
+ ILVEV_H2_UB(dst0, dst2, dst1, dst3, out0, out1);
+ ST_UB2(out0, out1, destination, 16);
+ }
+ }
+
+ pixelsPerRow &= 7;
+}
+
+} // namespace SIMD
+
+} // namespace blink
+
+#endif // HAVE(MIPS_MSA_INTRINSICS)
+
+#endif // WebGLImageConversionMSA_h

Powered by Google App Engine
This is Rietveld 408576698