Keep track of relative pixel offsets and utilize pshufb to efficiently extract relevant pixels for horizontal scaling ratios <= 4. Fall back to a generic approach for ratios > 4. Note that the generic approach can be backported to SSE2. 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 ~6.67x/~3.26x (32-bit/64-bit) for horizontal ratios <= 2, ~6.24x/~3.00x for ratios within (2, 4], and ~4.89x/~2.17x for ratios > 4 when not memory-bound on Haswell as compared with the current SSE2 implementation.
Showing the changes to codec/processing/src/downsample/downsample.cpp (1 of 6 files). Show all
Unified · Split
| @@ -100,6 +100,7 @@ void CDownsampling::InitDownsampleFuncs (SDownsampleFuncs& sDownsampleFunc, int · 99 unmodified lines | |||
| 100 | sDownsampleFunc.pfHalfAverage[1] = DyadicBilinearDownsamplerWidthx16_ssse3; | 100 | sDownsampleFunc.pfHalfAverage[1] = DyadicBilinearDownsamplerWidthx16_ssse3; |
| 101 | sDownsampleFunc.pfOneThirdDownsampler = DyadicBilinearOneThirdDownsampler_ssse3; | 101 | sDownsampleFunc.pfOneThirdDownsampler = DyadicBilinearOneThirdDownsampler_ssse3; |
| 102 | sDownsampleFunc.pfQuarterDownsampler = DyadicBilinearQuarterDownsampler_ssse3; | 102 | sDownsampleFunc.pfQuarterDownsampler = DyadicBilinearQuarterDownsampler_ssse3; |
| 103 | sDownsampleFunc.pfGeneralRatioLuma = GeneralBilinearFastDownsamplerWrap_ssse3; | ||
| 103 | } | 104 | } |
| 104 | if (iCpuFlag & WELS_CPU_SSE41) { | 105 | if (iCpuFlag & WELS_CPU_SSE41) { |
| 105 | sDownsampleFunc.pfHalfAverage[0] = DyadicBilinearDownsamplerWidthx32_sse4; | 106 | sDownsampleFunc.pfHalfAverage[0] = DyadicBilinearDownsamplerWidthx32_sse4; |