[Processing/x86] Add an AVX2 implementation of GeneralBilinearFastDownsample

Keep track of relative pixel offsets and utilize pshufb to efficiently extract relevant pixels for horizontal scaling ratios <= 8. Because pshufb does not cross 128-bit lanes, the overhead of address calculations and loads is relatively greater as compared with an SSSE3 implementation. Fall back to a generic approach for ratios > 8. The implementation assumes that data beyond the end of each line, before the next line begins, can be dirtied; which AFAICT is safe with the current usage of these routines. Speedup is ~10.42x/~5.23x (32-bit/64-bit) for horizontal ratios <= 2, ~9.49x/~4.64x for ratios within (2, 4], ~6.43x/~3.18x for ratios within (4, 8], and ~5.42x/~2.50x for ratios > 8 when not memory-bound on Haswell as compared with the current SSE2 implementation.

Sindre Aamås committed 10 years ago

Commit b43e58a1 parent b1013095 files · +709 −8Browse files

Showing the changes to codec/processing/src/downsample/downsample.h (1 of 5 files). Show all

Unified · Split

codec/processing/src/downsample/downsample.h+4−0

@@ -103,6 +103,7 @@ GeneralDownsampleFunc GeneralBilinearFastDownsamplerWrap_sse2; · 102 unmodified lines
103	103	GeneralDownsampleFunc GeneralBilinearAccurateDownsamplerWrap_sse2;
104	104	GeneralDownsampleFunc GeneralBilinearFastDownsamplerWrap_ssse3;
105	105	GeneralDownsampleFunc GeneralBilinearAccurateDownsamplerWrap_sse41;
	106	+GeneralDownsampleFunc GeneralBilinearFastDownsamplerWrap_avx2;
106	107
107	108	SpecificDownsampleFunc DyadicBilinearOneThirdDownsampler_ssse3;
108	109	SpecificDownsampleFunc DyadicBilinearOneThirdDownsampler_sse4;
@@ -122,6 +123,9 @@ void GeneralBilinearFastDownsampler_ssse3 (uint8_t* pDst, int32_t iDstStride, in · 13 unmodified lines
122	123	void GeneralBilinearAccurateDownsampler_sse41 (uint8_t* pDst, int32_t iDstStride, int32_t iDstWidth,
123	124	int32_t iDstHeight, uint8_t* pSrc, int32_t iSrcStride, uint32_t uiScaleX,
124	125	uint32_t uiScaleY);
	126	+void GeneralBilinearFastDownsampler_avx2 (uint8_t* pDst, int32_t iDstStride, int32_t iDstWidth,
	127	+ int32_t iDstHeight, uint8_t* pSrc, int32_t iSrcStride, uint32_t uiScaleX,
	128	+ uint32_t uiScaleY);
125	129
126	130	WELSVP_EXTERN_C_END
127	131	#endif

[Processing/x86] Add an AVX2 implementation of GeneralBilinearFastDownsample

Keyboard shortcuts

General

Issues and pull requests

Issues, pull requests and Inbox

Repository code