On Wed, Feb 25, 2015 at 7:34 AM, chen <[email protected]> wrote: > it's right, just move pw_257 into const.asm in future
I pushed this one, but then realized that the -m32 build is broken on LInux. It looks to be caused by the usage of r4b http://pastie.org/9982589 > At 2015-02-25 04:53:04,[email protected] wrote: >># HG changeset patch >># User David T Yuen <[email protected]> >># Date 1424810386 28800 >># Node ID 8528fea56a4202c871544621d5806ee686e671fc >># Parent 7252c10278a1e1aeddbe489180aad35196262919 >>asm: intrapred dc4 sse2 >> >>This replaces c code for systems using ssse3 to sse2 processors >>The code is backported from intrapred dc4 sse4 >> >>./test/TestBench --testbench intrapred >>Testing only harnesses that match name <intrapred> >>Using random seed 54ECE101 8bpp >>Testing primitives: SSE2 >>Testing primitives: SSE3 >> >>Test performance improvement with full optimizations >>== intrapred primitives == >>intra_dc_4x4[f=0] 1.26x 192.96 242.35 >>intra_dc_4x4[f=1] 1.25x 409.99 512.69 >> >>diff -r 7252c10278a1 -r 8528fea56a42 source/common/x86/asm-primitives.cpp >>--- a/source/common/x86/asm-primitives.cpp Mon Feb 23 23:11:49 2015 -0600 >>+++ b/source/common/x86/asm-primitives.cpp Tue Feb 24 12:39:46 2015 -0800 >>@@ -1071,6 +1071,8 @@ >> ALL_LUMA_TU_S(cpy1Dto2D_shr, cpy1Dto2D_shr_, sse2); >> ALL_LUMA_TU_S(ssd_s, pixel_ssd_s_, sse2); >> >>+ p.cu[BLOCK_4x4].intra_pred[DC_IDX] = x265_intra_pred_dc4_sse2; >>+ >> p.cu[BLOCK_4x4].calcresidual = x265_getResidual4_sse2; >> p.cu[BLOCK_8x8].calcresidual = x265_getResidual8_sse2; >> >>diff -r 7252c10278a1 -r 8528fea56a42 source/common/x86/intrapred.h >>--- a/source/common/x86/intrapred.h Mon Feb 23 23:11:49 2015 -0600 >>+++ b/source/common/x86/intrapred.h Tue Feb 24 12:39:46 2015 -0800 >>@@ -26,6 +26,7 @@ >> #ifndef X265_INTRAPRED_H >> #define X265_INTRAPRED_H >> >>+void x265_intra_pred_dc4_sse2 (pixel* dst, intptr_t dstStride, const >> pixel*srcPix, int, int filter); >> void x265_intra_pred_dc4_sse4 (pixel* dst, intptr_t dstStride, const >> pixel*srcPix, int, int filter); >> void x265_intra_pred_dc8_sse4(pixel* dst, intptr_t dstStride, const pixel* >> srcPix, int, int filter); >> void x265_intra_pred_dc16_sse4(pixel* dst, intptr_t dstStride, const >> pixel* srcPix, int, int filter); >>diff -r 7252c10278a1 -r 8528fea56a42 source/common/x86/intrapred8.asm >>--- a/source/common/x86/intrapred8.asm Mon Feb 23 23:11:49 2015 -0600 >>+++ b/source/common/x86/intrapred8.asm Tue Feb 24 12:39:46 2015 -0800 >>@@ -65,6 +65,8 @@ >> pw_planar32_L: dw 31, 30, 29, 28, 27, 26, 25, 24 >> pw_planar32_H: dw 23, 22, 21, 20, 19, 18, 17, 16 >> >>+pw_257: times 8 dw 257 >>+ >> const ang_table >> %assign x 0 >> %rep 32 >>@@ -89,6 +91,71 @@ >> >> ;--------------------------------------------------------------------------------------------- >> ; void intra_pred_dc(pixel* dst, intptr_t dstStride, pixel *srcPix, int >> dirMode, int bFilter) >> >> ;--------------------------------------------------------------------------------------------- >>+INIT_XMM sse2 >>+cglobal intra_pred_dc4, 5,5,3 >>+ inc r2 >>+ pxor m0, m0 >>+ movu m1, [r2] >>+ pshufd m1, m1, 0xF8 >>+ psadbw m1, m0 ; m1 = sum >>+ >>+ test r4d, r4d >>+ >>+ paddw m1, [pw_4] >>+ psraw m1, 3 >>+ movd r4d, m1 ; r4d = dc_val >>+ pmullw m1, [pw_257] >>+ pshuflw m1, m1, 0x00 >>+ >>+ ; store DC 4x4 >>+ lea r3, [r1 * 3] >>+ movd [r0], m1 >>+ movd [r0 + r1], m1 >>+ movd [r0 + r1 * 2], m1 >>+ movd [r0 + r3], m1 >>+ >>+ ; do DC filter >>+ jz .end >>+ lea r3d, [r4d * 2 + 2] ; r3d = DC * 2 + 2 >>+ add r4d, r3d ; r4d = DC * 3 + 2 >>+ movd m1, r4d >>+ pshuflw m1, m1, 0 ; m1 = pixDCx3 >>+ >>+ ; filter top >>+ movd m2, [r2] >>+ punpcklbw m2, m0 >>+ paddw m2, m1 >>+ psraw m2, 2 >>+ packuswb m2, m2 >>+ movd [r0], m2 ; overwrite top-left pixel, we will >> update it later >>+ >>+ ; filter top-left >>+ movzx r4d, byte [r2 + 8] >>+ add r3d, r4d >>+ movzx r4d, byte [r2] >>+ add r3d, r4d >>+ shr r3d, 2 >>+ mov [r0], r3b >>+ >>+ ; filter left >>+ add r0, r1 >>+ movq m2, [r2 + 9] >>+ punpcklbw m2, m0 >>+ paddw m2, m1 >>+ psraw m2, 2 >>+ packuswb m2, m2 >>+ movq r4, m2 >>+ mov [r0], r4b >>+ shr r4, 8 >>+ mov [r0 + r1], r4b >>+ shr r4, 8 >>+ mov [r0 + r1 * 2], r4b >>+.end: >>+ RET >>+ >>+;--------------------------------------------------------------------------------------------- >>+; void intra_pred_dc(pixel* dst, intptr_t dstStride, pixel *srcPix, int >> dirMode, int bFilter) >>+;--------------------------------------------------------------------------------------------- >> INIT_XMM sse4 >> cglobal intra_pred_dc4, 5,5,3 >> inc r2 >>_______________________________________________ >>x265-devel mailing list >>[email protected] >>https://mailman.videolan.org/listinfo/x265-devel > > > _______________________________________________ > x265-devel mailing list > [email protected] > https://mailman.videolan.org/listinfo/x265-devel > -- Steve Borho _______________________________________________ x265-devel mailing list [email protected] https://mailman.videolan.org/listinfo/x265-devel
