[Mesa-dev] [PATCH] swr: [rasterizer core/memory] Finish R24_UNORM_X8_TYPELESS for AVX512

[Cherniak, Bruce]

Reviewed-by: Bruce Cherniak <bruce.cherniak at intel.com> 

> On Dec 7, 2016, at 5:55 PM, Rowley, Timothy O <timothy.o.rowley at intel.com> wrote:
> 
> This one-off specialization was missed.
> ---
> src/gallium/drivers/swr/rasterizer/core/backend.h  |  4 +-
> .../drivers/swr/rasterizer/memory/StoreTile.h      | 46 +++++++++++-----------
> 2 files changed, 24 insertions(+), 26 deletions(-)
> 
> diff --git a/src/gallium/drivers/swr/rasterizer/core/backend.h b/src/gallium/drivers/swr/rasterizer/core/backend.h
> index 706b372..48cfce2 100644
> --- a/src/gallium/drivers/swr/rasterizer/core/backend.h
> +++ b/src/gallium/drivers/swr/rasterizer/core/backend.h
> @@ -463,7 +463,7 @@ inline void SetupBarycentricCoeffs(BarycentricCoeffs *coeffs, const SWR_TRIANGLE
> 
> inline void SetupRenderBuffers(uint8_t *pColorBuffer[SWR_NUM_RENDERTARGETS], uint8_t **pDepthBuffer, uint8_t **pStencilBuffer, uint32_t colorBufferCount, RenderOutputBuffers &renderBuffers)
> {
> -    SWR_ASSERT(colorBufferCount <= SWR_NUM_RENDERTARGETS);
> +    assert(colorBufferCount <= SWR_NUM_RENDERTARGETS);
> 
>     if (pColorBuffer)
>     {
> @@ -754,8 +754,6 @@ INLINE void OutputMerger(SWR_PS_CONTEXT &psContext, uint8_t* (&pColorBase)[SWR_N
> INLINE void OutputMerger(SWR_PS_CONTEXT &psContext, uint8_t* (&pColorBase)[SWR_NUM_RENDERTARGETS], uint32_t sample, const SWR_BLEND_STATE *pBlendState,
>     const PFN_BLEND_JIT_FUNC(&pfnBlendFunc)[SWR_NUM_RENDERTARGETS], simdscalar &coverageMask, simdscalar depthPassMask, const uint32_t NumRT, bool useAlternateOffset)
> {
> -    assert(sample == 0); // will need up upate Raster Tile Color Offsets to support more than single sample here..
> -
>     // type safety guaranteed from template instantiation in BEChooser<>::GetFunc
>     uint32_t rasterTileColorOffset = RasterTileColorOffset(sample);
> 
> diff --git a/src/gallium/drivers/swr/rasterizer/memory/StoreTile.h b/src/gallium/drivers/swr/rasterizer/memory/StoreTile.h
> index 558f048..4fa6683 100644
> --- a/src/gallium/drivers/swr/rasterizer/memory/StoreTile.h
> +++ b/src/gallium/drivers/swr/rasterizer/memory/StoreTile.h
> @@ -545,38 +545,38 @@ struct ConvertPixelsSOAtoAOS<R32_FLOAT, R24_UNORM_X8_TYPELESS>
>     INLINE static void Convert(const uint8_t* pSrc, uint8_t* (&ppDsts)[NumDests])
>     {
> #if USE_8x2_TILE_BACKEND
> -        static const uint32_t MAX_RASTER_TILE_BYTES = 16 * 4; // 16 pixels * 4 bytes per pixel
> +        simd16scalar comp = _simd16_load_ps(reinterpret_cast<const float *>(pSrc));
> 
> -        OSALIGNSIMD16(uint8_t) soaTile[MAX_RASTER_TILE_BYTES];
> -        OSALIGNSIMD16(uint8_t) aosTile[MAX_RASTER_TILE_BYTES];
> +        // clamp
> +        const simd16scalar zero = _simd16_setzero_ps();
> +        const simd16scalar ones = _simd16_set1_ps(1.0f);
> 
> -        // Convert from SrcFormat --> DstFormat
> -        simd16vector src;
> -        LoadSOA<SrcFormat>(pSrc, src);
> -        StoreSOA<DstFormat>(src, soaTile);
> +        comp = _simd16_max_ps(comp, zero);
> +        comp = _simd16_min_ps(comp, ones);
> 
> -        // Convert from SOA --> AOS
> -        FormatTraits<DstFormat>::TransposeT::Transpose_16(soaTile, aosTile);
> -
> -        // Store data into destination but don't overwrite the X8 bits
> -        // Each 4-pixel row is 16-bytes
> +        // normalize
> +        comp = _simd16_mul_ps(comp, _simd16_set1_ps(FormatTraits<DstFormat>::fromFloat(0)));
> 
> -        simdscalari loadlo = _simd_load_si(reinterpret_cast<simdscalari *>(aosTile));
> -        simdscalari loadhi = _simd_load_si(reinterpret_cast<simdscalari *>(aosTile + sizeof(simdscalari)));
> +        simd16scalari temp = _simd16_cvtps_epi32(comp);
> 
> -        simdscalari templo = _simd_unpacklo_epi64(loadlo, loadhi);
> -        simdscalari temphi = _simd_unpackhi_epi64(loadlo, loadhi);
> +        // swizzle
> +        temp = _simd16_permute_epi32(temp, _simd16_set_epi32(15, 14, 11, 10, 13, 12, 9, 8, 7, 6, 3, 2, 5, 4, 1, 0));
> 
> +        // merge/store data into destination but don't overwrite the X8 bits
>         simdscalari destlo = _simd_loadu2_si(reinterpret_cast<__m128i *>(ppDsts[1]), reinterpret_cast<__m128i *>(ppDsts[0]));
>         simdscalari desthi = _simd_loadu2_si(reinterpret_cast<__m128i *>(ppDsts[3]), reinterpret_cast<__m128i *>(ppDsts[2]));
> 
> -        simdscalari mask = _simd_set1_epi32(0x00FFFFFF);
> +        simd16scalari dest = _simd16_setzero_si();
> +
> +        dest = _simd16_insert_si(dest, destlo, 0);
> +        dest = _simd16_insert_si(dest, desthi, 1);
> +
> +        simd16scalari mask = _simd16_set1_epi32(0x00FFFFFF);
> 
> -        destlo = _simd_or_si(_simd_andnot_si(mask, destlo), _simd_and_si(mask, templo));
> -        desthi = _simd_or_si(_simd_andnot_si(mask, desthi), _simd_and_si(mask, templo));
> +        dest = _simd16_or_si(_simd16_andnot_si(mask, dest), _simd16_and_si(mask, temp));
> 
> -        _simd_storeu2_si(reinterpret_cast<__m128i *>(ppDsts[1]), reinterpret_cast<__m128i *>(ppDsts[0]), destlo);
> -        _simd_storeu2_si(reinterpret_cast<__m128i *>(ppDsts[3]), reinterpret_cast<__m128i *>(ppDsts[2]), desthi);
> +        _simd_storeu2_si(reinterpret_cast<__m128i *>(ppDsts[1]), reinterpret_cast<__m128i *>(ppDsts[0]), _simd16_extract_si(dest, 0));
> +        _simd_storeu2_si(reinterpret_cast<__m128i *>(ppDsts[3]), reinterpret_cast<__m128i *>(ppDsts[2]), _simd16_extract_si(dest, 1));
> #else
>         static const uint32_t MAX_RASTER_TILE_BYTES = 128; // 8 pixels * 16 bytes per pixel
> 
> @@ -663,7 +663,7 @@ INLINE static void FlatConvert(const uint8_t* pSrc, uint8_t* pDst0, uint8_t* pDs
>     simd16scalari src3 = _simd16_cvtps_epi32(comp3); // padded byte aaaaaaaaaaaaaaaa
> 
>     // SOA to AOS conversion
> -    src1 = _simd16_slli_epi32(src1, 8);
> +    src1 = _simd16_slli_epi32(src1,  8);
>     src2 = _simd16_slli_epi32(src2, 16);
>     src3 = _simd16_slli_epi32(src3, 24);
> 
> @@ -836,7 +836,7 @@ INLINE static void FlatConvertNoAlpha(const uint8_t* pSrc, uint8_t* pDst0, uint8
>     simd16scalari src2 = _simd16_cvtps_epi32(comp2); // padded byte bbbbbbbbbbbbbbbb
> 
>     // SOA to AOS conversion
> -    src1 = _simd16_slli_epi32(src1, 8);
> +    src1 = _simd16_slli_epi32(src1,  8);
>     src2 = _simd16_slli_epi32(src2, 16);
> 
>     simd16scalari final = _simd16_or_si(_simd16_or_si(src0, src1), src2);                       // 0 1 2 3 4 5 6 7 8 9 A B C D E F
> -- 
> 2.7.4
>