<div dir="ltr"><b><span style="font-size:12.8000001907349px">This looks highly suspect to me. Why would the destination stride be</span><br style="font-size:12.8000001907349px"><span style="font-size:12.8000001907349px">different depending on the subpel offsets? You're using MAX_CU_SIZE here</span><br style="font-size:12.8000001907349px"></b><span style="font-size:12.8000001907349px"><b>but dstStride is used everywhere below.</b><br></span><br> luma and chroma are using the same C function filterPixelToShort_C. The dstStride is MAX_CU_SIZE for luma and MAX_CU_SIZE/2 for chroma i420,i422, MAX_CU_SIZE for i444. For earlier C function we passed dstStride(MAX_CU_SIZE) directly to C function.<br><br><div><span style="color:rgb(80,0,80);font-size:12.8000001907349px">> - p.chroma[X265_CSP_I444].pu[</span><span style="color:rgb(80,0,80);font-size:12.8000001907349px">LUMA_ ## W ## x ## H].chroma_p2s = pixelToShort_c<MAX_CU_SIZE, W, H>;</span></div><div>now only height and width are passed from here. </div><div><br>This all to make use of same c/asm primitive for luma and chroma<font color="#500050"><span style="font-size:12.8000001907349px"><br></span></font><div><br></div><div><span style="font-size:12.8000001907349px"> <b> the 'luma_' prefix is redundant, everything within pu[] is implied to</b></span><b><br style="font-size:12.8000001907349px"><span style="font-size:12.8000001907349px"> be luma since it is not in chroma[].pu[]</span></b><br><br>I will change luma_p2s to convert_p2s everywhere<br><br></div></div></div><div class="gmail_extra"><br><div class="gmail_quote">On Mon, Mar 30, 2015 at 10:59 PM, Steve Borho <span dir="ltr"><<a href="mailto:steve@borho.org" target="_blank">steve@borho.org</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div class="HOEnZb"><div class="h5">On 03/30, <a href="mailto:rajesh@multicorewareinc.com">rajesh@multicorewareinc.com</a> wrote:<br>
> # HG changeset patch<br>
> # User Rajesh Paulraj<<a href="mailto:rajesh@multicorewareinc.com">rajesh@multicorewareinc.com</a>><br>
> # Date 1427292575 -19800<br>
> # Wed Mar 25 19:39:35 2015 +0530<br>
> # Node ID 8797b42373debaab0ef1a2a1f4f8776a69916cfb<br>
> # Parent 22a312799bb033d40a66fc83a1ac7af192ce2420<br>
> primivites: remove filter_p2s and move luma_p2s into PU<br>
><br>
> diff -r 22a312799bb0 -r 8797b42373de source/common/ipfilter.cpp<br>
> --- a/source/common/ipfilter.cpp Fri Mar 27 22:59:30 2015 -0500<br>
> +++ b/source/common/ipfilter.cpp Wed Mar 25 19:39:35 2015 +0530<br>
> @@ -34,27 +34,8 @@<br>
> #endif<br>
><br>
> namespace {<br>
> -template<int dstStride, int width, int height><br>
> -void pixelToShort_c(const pixel* src, intptr_t srcStride, int16_t* dst)<br>
> -{<br>
> - int shift = IF_INTERNAL_PREC - X265_DEPTH;<br>
> - int row, col;<br>
> -<br>
> - for (row = 0; row < height; row++)<br>
> - {<br>
> - for (col = 0; col < width; col++)<br>
> - {<br>
> - int16_t val = src[col] << shift;<br>
> - dst[col] = val - (int16_t)IF_INTERNAL_OFFS;<br>
> - }<br>
> -<br>
> - src += srcStride;<br>
> - dst += dstStride;<br>
> - }<br>
> -}<br>
> -<br>
> -template<int dstStride><br>
> -void filterPixelToShort_c(const pixel* src, intptr_t srcStride, int16_t* dst, int width, int height)<br>
> +template<int width, int height><br>
> +void filterPixelToShort_c(const pixel* src, intptr_t srcStride, int16_t* dst, int16_t dstStride)<br>
> {<br>
> int shift = IF_INTERNAL_PREC - X265_DEPTH;<br>
> int row, col;<br>
> @@ -398,7 +379,7 @@<br>
> p.chroma[X265_CSP_I420].pu[CHROMA_420_ ## W ## x ## H].filter_vps = interp_vert_ps_c<4, W, H>; \<br>
> p.chroma[X265_CSP_I420].pu[CHROMA_420_ ## W ## x ## H].filter_vsp = interp_vert_sp_c<4, W, H>; \<br>
> p.chroma[X265_CSP_I420].pu[CHROMA_420_ ## W ## x ## H].filter_vss = interp_vert_ss_c<4, W, H>; \<br>
> - p.chroma[X265_CSP_I420].pu[CHROMA_420_ ## W ## x ## H].chroma_p2s = pixelToShort_c<MAX_CU_SIZE / 2, W, H>;<br>
> + p.chroma[X265_CSP_I420].pu[CHROMA_420_ ## W ## x ## H].chroma_p2s = filterPixelToShort_c<W, H>;<br>
><br>
> #define CHROMA_422(W, H) \<br>
> p.chroma[X265_CSP_I422].pu[CHROMA_422_ ## W ## x ## H].filter_hpp = interp_horiz_pp_c<4, W, H>; \<br>
> @@ -407,7 +388,7 @@<br>
> p.chroma[X265_CSP_I422].pu[CHROMA_422_ ## W ## x ## H].filter_vps = interp_vert_ps_c<4, W, H>; \<br>
> p.chroma[X265_CSP_I422].pu[CHROMA_422_ ## W ## x ## H].filter_vsp = interp_vert_sp_c<4, W, H>; \<br>
> p.chroma[X265_CSP_I422].pu[CHROMA_422_ ## W ## x ## H].filter_vss = interp_vert_ss_c<4, W, H>; \<br>
> - p.chroma[X265_CSP_I422].pu[CHROMA_422_ ## W ## x ## H].chroma_p2s = pixelToShort_c<MAX_CU_SIZE / 2, W, H>;<br>
> + p.chroma[X265_CSP_I422].pu[CHROMA_422_ ## W ## x ## H].chroma_p2s = filterPixelToShort_c<W, H>;<br>
><br>
> #define CHROMA_444(W, H) \<br>
> p.chroma[X265_CSP_I444].pu[LUMA_ ## W ## x ## H].filter_hpp = interp_horiz_pp_c<4, W, H>; \<br>
> @@ -416,7 +397,7 @@<br>
> p.chroma[X265_CSP_I444].pu[LUMA_ ## W ## x ## H].filter_vps = interp_vert_ps_c<4, W, H>; \<br>
> p.chroma[X265_CSP_I444].pu[LUMA_ ## W ## x ## H].filter_vsp = interp_vert_sp_c<4, W, H>; \<br>
> p.chroma[X265_CSP_I444].pu[LUMA_ ## W ## x ## H].filter_vss = interp_vert_ss_c<4, W, H>; \<br>
> - p.chroma[X265_CSP_I444].pu[LUMA_ ## W ## x ## H].chroma_p2s = pixelToShort_c<MAX_CU_SIZE, W, H>;<br>
> + p.chroma[X265_CSP_I444].pu[LUMA_ ## W ## x ## H].chroma_p2s = filterPixelToShort_c<W, H>;<br>
><br>
> #define LUMA(W, H) \<br>
> p.pu[LUMA_ ## W ## x ## H].luma_hpp = interp_horiz_pp_c<8, W, H>; \<br>
> @@ -426,7 +407,7 @@<br>
> p.pu[LUMA_ ## W ## x ## H].luma_vsp = interp_vert_sp_c<8, W, H>; \<br>
> p.pu[LUMA_ ## W ## x ## H].luma_vss = interp_vert_ss_c<8, W, H>; \<br>
> p.pu[LUMA_ ## W ## x ## H].luma_hvpp = interp_hv_pp_c<8, W, H>; \<br>
> - p.pu[LUMA_ ## W ## x ## H].filter_p2s = pixelToShort_c<MAX_CU_SIZE, W, H><br>
> + p.pu[LUMA_ ## W ## x ## H].luma_p2s = filterPixelToShort_c<W, H>;<br>
><br>
> void setupFilterPrimitives_c(EncoderPrimitives& p)<br>
> {<br>
> @@ -530,11 +511,116 @@<br>
> CHROMA_444(48, 64);<br>
> CHROMA_444(64, 16);<br>
> CHROMA_444(16, 64);<br>
> - p.luma_p2s = filterPixelToShort_c<MAX_CU_SIZE>;<br>
><br>
> - p.chroma[X265_CSP_I444].p2s = filterPixelToShort_c<MAX_CU_SIZE>;<br>
> - p.chroma[X265_CSP_I420].p2s = filterPixelToShort_c<MAX_CU_SIZE / 2>;<br>
> - p.chroma[X265_CSP_I422].p2s = filterPixelToShort_c<MAX_CU_SIZE / 2>;<br>
> + p.pu[LUMA_4x4].luma_p2s = filterPixelToShort_c<4, 4>;<br>
> + p.pu[LUMA_4x8].luma_p2s = filterPixelToShort_c<4, 8>;<br>
> + p.pu[LUMA_4x16].luma_p2s = filterPixelToShort_c<4, 16>;<br>
> + p.pu[LUMA_8x4].luma_p2s = filterPixelToShort_c<8, 4>;<br>
> + p.pu[LUMA_8x8].luma_p2s = filterPixelToShort_c<8, 8>;<br>
> + p.pu[LUMA_8x16].luma_p2s = filterPixelToShort_c<8, 16>;<br>
> + p.pu[LUMA_8x32].luma_p2s = filterPixelToShort_c<8, 32>;<br>
> + p.pu[LUMA_16x4].luma_p2s = filterPixelToShort_c<16, 4>;<br>
> + p.pu[LUMA_16x8].luma_p2s = filterPixelToShort_c<16, 8>;<br>
> + p.pu[LUMA_16x12].luma_p2s = filterPixelToShort_c<16, 12>;<br>
> + p.pu[LUMA_16x16].luma_p2s = filterPixelToShort_c<16, 16>;<br>
> + p.pu[LUMA_16x32].luma_p2s = filterPixelToShort_c<16, 32>;<br>
> + p.pu[LUMA_16x64].luma_p2s = filterPixelToShort_c<16, 64>;<br>
> + p.pu[LUMA_32x8].luma_p2s = filterPixelToShort_c<32, 8>;<br>
> + p.pu[LUMA_32x16].luma_p2s = filterPixelToShort_c<32, 16>;<br>
> + p.pu[LUMA_32x24].luma_p2s = filterPixelToShort_c<32, 24>;<br>
> + p.pu[LUMA_32x32].luma_p2s = filterPixelToShort_c<32, 32>;<br>
> + p.pu[LUMA_32x64].luma_p2s = filterPixelToShort_c<32, 64>;<br>
> + p.pu[LUMA_64x16].luma_p2s = filterPixelToShort_c<64, 16>;<br>
> + p.pu[LUMA_64x32].luma_p2s = filterPixelToShort_c<64, 32>;<br>
> + p.pu[LUMA_64x48].luma_p2s = filterPixelToShort_c<64, 48>;<br>
> + p.pu[LUMA_64x64].luma_p2s = filterPixelToShort_c<64, 64>;<br>
> + p.pu[LUMA_12x16].luma_p2s = filterPixelToShort_c<12, 16>;<br>
> + p.pu[LUMA_24x32].luma_p2s = filterPixelToShort_c<24, 32>;<br>
> + p.pu[LUMA_48x64].luma_p2s = filterPixelToShort_c<48, 64>;<br>
> +<br>
> + p.chroma[X265_CSP_I420].pu[CHROMA_420_2x2].chroma_p2s = filterPixelToShort_c<2, 2>;<br>
> + p.chroma[X265_CSP_I420].pu[CHROMA_420_4x4].chroma_p2s = filterPixelToShort_c<4, 4>;<br>
> + p.chroma[X265_CSP_I420].pu[CHROMA_420_8x8].chroma_p2s = filterPixelToShort_c<8, 8>;<br>
> + p.chroma[X265_CSP_I420].pu[CHROMA_420_16x16].chroma_p2s = filterPixelToShort_c<16, 16>;<br>
> + p.chroma[X265_CSP_I420].pu[CHROMA_420_32x32].chroma_p2s = filterPixelToShort_c<32, 32>;<br>
> +<br>
> + p.chroma[X265_CSP_I420].pu[CHROMA_420_4x2].chroma_p2s = filterPixelToShort_c<4, 2>;<br>
> + p.chroma[X265_CSP_I420].pu[CHROMA_420_2x4].chroma_p2s = filterPixelToShort_c<2, 4>;<br>
> + p.chroma[X265_CSP_I420].pu[CHROMA_420_8x4].chroma_p2s = filterPixelToShort_c<8, 4>;<br>
> + p.chroma[X265_CSP_I420].pu[CHROMA_420_4x8].chroma_p2s = filterPixelToShort_c<4, 8>;<br>
> + p.chroma[X265_CSP_I420].pu[CHROMA_420_16x8].chroma_p2s = filterPixelToShort_c<16, 8>;<br>
> + p.chroma[X265_CSP_I420].pu[CHROMA_420_8x16].chroma_p2s = filterPixelToShort_c<8, 16>;<br>
> + p.chroma[X265_CSP_I420].pu[CHROMA_420_32x16].chroma_p2s = filterPixelToShort_c<32, 16>;<br>
> + p.chroma[X265_CSP_I420].pu[CHROMA_420_16x32].chroma_p2s = filterPixelToShort_c<16, 32>;<br>
> +<br>
> + p.chroma[X265_CSP_I420].pu[CHROMA_420_8x6].chroma_p2s = filterPixelToShort_c<8, 6>;<br>
> + p.chroma[X265_CSP_I420].pu[CHROMA_420_6x8].chroma_p2s = filterPixelToShort_c<6, 8>;<br>
> + p.chroma[X265_CSP_I420].pu[CHROMA_420_8x2].chroma_p2s = filterPixelToShort_c<8, 2>;<br>
> + p.chroma[X265_CSP_I420].pu[CHROMA_420_2x8].chroma_p2s = filterPixelToShort_c<2, 8>;<br>
> + p.chroma[X265_CSP_I420].pu[CHROMA_420_16x12].chroma_p2s = filterPixelToShort_c<16, 12>;<br>
> + p.chroma[X265_CSP_I420].pu[CHROMA_420_12x16].chroma_p2s = filterPixelToShort_c<12, 16>;<br>
> + p.chroma[X265_CSP_I420].pu[CHROMA_420_16x4].chroma_p2s = filterPixelToShort_c<16, 4>;<br>
> + p.chroma[X265_CSP_I420].pu[CHROMA_420_4x16].chroma_p2s = filterPixelToShort_c<4, 16>;<br>
> + p.chroma[X265_CSP_I420].pu[CHROMA_420_32x24].chroma_p2s = filterPixelToShort_c<32, 24>;<br>
> + p.chroma[X265_CSP_I420].pu[CHROMA_420_24x32].chroma_p2s = filterPixelToShort_c<24, 32>;<br>
> + p.chroma[X265_CSP_I420].pu[CHROMA_420_32x8].chroma_p2s = filterPixelToShort_c<32, 8>;<br>
> + p.chroma[X265_CSP_I420].pu[CHROMA_420_8x32].chroma_p2s = filterPixelToShort_c<8, 32>;<br>
> +<br>
> + p.chroma[X265_CSP_I422].pu[CHROMA_422_2x4].chroma_p2s = filterPixelToShort_c<2, 4>;<br>
> + p.chroma[X265_CSP_I422].pu[CHROMA_422_4x8].chroma_p2s = filterPixelToShort_c<4, 8>;<br>
> + p.chroma[X265_CSP_I422].pu[CHROMA_422_8x16].chroma_p2s = filterPixelToShort_c<8, 16>;<br>
> + p.chroma[X265_CSP_I422].pu[CHROMA_422_16x32].chroma_p2s = filterPixelToShort_c<16, 32>;<br>
> + p.chroma[X265_CSP_I422].pu[CHROMA_422_32x64].chroma_p2s = filterPixelToShort_c<32, 64>;<br>
> +<br>
> + p.chroma[X265_CSP_I422].pu[CHROMA_422_4x4].chroma_p2s = filterPixelToShort_c<4, 4>;<br>
> + p.chroma[X265_CSP_I422].pu[CHROMA_422_2x8].chroma_p2s = filterPixelToShort_c<2, 8>;<br>
> + p.chroma[X265_CSP_I422].pu[CHROMA_422_8x8].chroma_p2s = filterPixelToShort_c<8, 8>;<br>
> + p.chroma[X265_CSP_I422].pu[CHROMA_422_4x16].chroma_p2s = filterPixelToShort_c<4, 16>;<br>
> + p.chroma[X265_CSP_I422].pu[CHROMA_422_16x16].chroma_p2s = filterPixelToShort_c<16, 16>;<br>
> + p.chroma[X265_CSP_I422].pu[CHROMA_422_8x32].chroma_p2s = filterPixelToShort_c<8, 32>;<br>
> + p.chroma[X265_CSP_I422].pu[CHROMA_422_32x32].chroma_p2s = filterPixelToShort_c<32, 32>;<br>
> + p.chroma[X265_CSP_I422].pu[CHROMA_422_16x64].chroma_p2s = filterPixelToShort_c<16, 64>;<br>
> +<br>
> + p.chroma[X265_CSP_I422].pu[CHROMA_422_8x12].chroma_p2s = filterPixelToShort_c<8, 12>;<br>
> + p.chroma[X265_CSP_I422].pu[CHROMA_422_6x16].chroma_p2s = filterPixelToShort_c<6, 16>;<br>
> + p.chroma[X265_CSP_I422].pu[CHROMA_422_8x4].chroma_p2s = filterPixelToShort_c<8, 4>;<br>
> + p.chroma[X265_CSP_I422].pu[CHROMA_422_2x16].chroma_p2s = filterPixelToShort_c<2, 16>;<br>
> + p.chroma[X265_CSP_I422].pu[CHROMA_422_16x24].chroma_p2s = filterPixelToShort_c<16, 24>;<br>
> + p.chroma[X265_CSP_I422].pu[CHROMA_422_12x32].chroma_p2s = filterPixelToShort_c<12, 32>;<br>
> + p.chroma[X265_CSP_I422].pu[CHROMA_422_16x8].chroma_p2s = filterPixelToShort_c<16, 8>;<br>
> + p.chroma[X265_CSP_I422].pu[CHROMA_422_4x32].chroma_p2s = filterPixelToShort_c<4, 32>;<br>
> + p.chroma[X265_CSP_I422].pu[CHROMA_422_32x48].chroma_p2s = filterPixelToShort_c<32, 48>;<br>
> + p.chroma[X265_CSP_I422].pu[CHROMA_422_24x64].chroma_p2s = filterPixelToShort_c<24, 64>;<br>
> + p.chroma[X265_CSP_I422].pu[CHROMA_422_32x16].chroma_p2s = filterPixelToShort_c<32, 16>;<br>
> + p.chroma[X265_CSP_I422].pu[CHROMA_422_8x64].chroma_p2s = filterPixelToShort_c<8, 64>;<br>
> +<br>
> + p.chroma[X265_CSP_I444].pu[CHROMA_420_2x2].chroma_p2s = filterPixelToShort_c<2, 2>;<br>
> + p.chroma[X265_CSP_I444].pu[CHROMA_420_4x4].chroma_p2s = filterPixelToShort_c<4, 4>;<br>
> + p.chroma[X265_CSP_I444].pu[CHROMA_420_8x8].chroma_p2s = filterPixelToShort_c<8, 8>;<br>
> + p.chroma[X265_CSP_I444].pu[CHROMA_420_16x16].chroma_p2s = filterPixelToShort_c<16, 16>;<br>
> + p.chroma[X265_CSP_I444].pu[CHROMA_420_32x32].chroma_p2s = filterPixelToShort_c<32, 32>;<br>
> +<br>
> + p.chroma[X265_CSP_I444].pu[CHROMA_420_4x2].chroma_p2s = filterPixelToShort_c<4, 2>;<br>
> + p.chroma[X265_CSP_I444].pu[CHROMA_420_2x4].chroma_p2s = filterPixelToShort_c<2, 4>;<br>
> + p.chroma[X265_CSP_I444].pu[CHROMA_420_8x4].chroma_p2s = filterPixelToShort_c<8, 4>;<br>
> + p.chroma[X265_CSP_I444].pu[CHROMA_420_4x8].chroma_p2s = filterPixelToShort_c<4, 8>;<br>
> + p.chroma[X265_CSP_I444].pu[CHROMA_420_16x8].chroma_p2s = filterPixelToShort_c<16, 8>;<br>
> + p.chroma[X265_CSP_I444].pu[CHROMA_420_8x16].chroma_p2s = filterPixelToShort_c<8, 16>;<br>
> + p.chroma[X265_CSP_I444].pu[CHROMA_420_32x16].chroma_p2s = filterPixelToShort_c<32, 16>;<br>
> + p.chroma[X265_CSP_I444].pu[CHROMA_420_16x32].chroma_p2s = filterPixelToShort_c<16, 32>;<br>
> +<br>
> + p.chroma[X265_CSP_I444].pu[CHROMA_420_8x6].chroma_p2s = filterPixelToShort_c<8, 6>;<br>
> + p.chroma[X265_CSP_I444].pu[CHROMA_420_6x8].chroma_p2s = filterPixelToShort_c<6, 8>;<br>
> + p.chroma[X265_CSP_I444].pu[CHROMA_420_8x2].chroma_p2s = filterPixelToShort_c<8, 2>;<br>
> + p.chroma[X265_CSP_I444].pu[CHROMA_420_2x8].chroma_p2s = filterPixelToShort_c<2, 8>;<br>
> + p.chroma[X265_CSP_I444].pu[CHROMA_420_16x12].chroma_p2s = filterPixelToShort_c<16, 12>;<br>
> + p.chroma[X265_CSP_I444].pu[CHROMA_420_12x16].chroma_p2s = filterPixelToShort_c<12, 16>;<br>
> + p.chroma[X265_CSP_I444].pu[CHROMA_420_16x4].chroma_p2s = filterPixelToShort_c<16, 4>;<br>
> + p.chroma[X265_CSP_I444].pu[CHROMA_420_4x16].chroma_p2s = filterPixelToShort_c<4, 16>;<br>
> + p.chroma[X265_CSP_I444].pu[CHROMA_420_32x24].chroma_p2s = filterPixelToShort_c<32, 24>;<br>
> + p.chroma[X265_CSP_I444].pu[CHROMA_420_24x32].chroma_p2s = filterPixelToShort_c<24, 32>;<br>
> + p.chroma[X265_CSP_I444].pu[CHROMA_420_32x8].chroma_p2s = filterPixelToShort_c<32, 8>;<br>
> + p.chroma[X265_CSP_I444].pu[CHROMA_420_8x32].chroma_p2s = filterPixelToShort_c<8, 32>;<br>
><br>
> p.extendRowBorder = extendCURowColBorder;<br>
> }<br>
> diff -r 22a312799bb0 -r 8797b42373de source/common/predict.cpp<br>
> --- a/source/common/predict.cpp Fri Mar 27 22:59:30 2015 -0500<br>
> +++ b/source/common/predict.cpp Wed Mar 25 19:39:35 2015 +0530<br>
> @@ -288,7 +288,7 @@<br>
> X265_CHECK(dstStride == MAX_CU_SIZE, "stride expected to be max cu size\n");<br>
><br>
> if (!(yFrac | xFrac))<br>
> - primitives.luma_p2s(src, srcStride, dst, pu.width, pu.height);<br>
> + primitives.pu[partEnum].luma_p2s(src, srcStride, dst, MAX_CU_SIZE);<br>
<br>
</div></div>This looks highly suspect to me. Why would the destination stride be<br>
different depending on the subpel offsets? You're using MAX_CU_SIZE here<br>
but dstStride is used everywhere below.<br>
<div><div class="h5"><br>
> else if (!yFrac)<br>
> primitives.pu[partEnum].luma_hps(src, srcStride, dst, dstStride, xFrac, 0);<br>
> else if (!xFrac)<br>
> @@ -373,16 +373,23 @@<br>
> int yFrac = mv.y & ((1 << shiftVer) - 1);<br>
><br>
> int partEnum = partitionFromSizes(pu.width, pu.height);<br>
> -<br>
> +<br>
> uint32_t cxWidth = pu.width >> m_hChromaShift;<br>
> - uint32_t cxHeight = pu.height >> m_vChromaShift;<br>
><br>
> - X265_CHECK(((cxWidth | cxHeight) % 2) == 0, "chroma block size expected to be multiple of 2\n");<br>
> + X265_CHECK(((cxWidth | (pu.height >> m_vChromaShift)) % 2) == 0, "chroma block size expected to be multiple of 2\n");<br>
><br>
> if (!(yFrac | xFrac))<br>
> {<br>
> - primitives.chroma[m_csp].p2s(refCb, refStride, dstCb, cxWidth, cxHeight);<br>
> - primitives.chroma[m_csp].p2s(refCr, refStride, dstCr, cxWidth, cxHeight);<br>
> + if (m_csp != X265_CSP_I444)<br>
> + {<br>
> + primitives.chroma[m_csp].pu[partEnum].chroma_p2s(refCb, refStride, dstCb, MAX_CU_SIZE / 2);<br>
> + primitives.chroma[m_csp].pu[partEnum].chroma_p2s(refCr, refStride, dstCr, MAX_CU_SIZE / 2);<br>
> + }<br>
> + else<br>
> + {<br>
> + primitives.chroma[m_csp].pu[partEnum].chroma_p2s(refCb, refStride, dstCb, MAX_CU_SIZE);<br>
> + primitives.chroma[m_csp].pu[partEnum].chroma_p2s(refCr, refStride, dstCr, MAX_CU_SIZE);<br>
> + }<br>
<br>
</div></div>Ditto for everything here<br>
<div><div class="h5"><br>
> }<br>
> else if (!yFrac)<br>
> {<br>
> diff -r 22a312799bb0 -r 8797b42373de source/common/primitives.cpp<br>
> --- a/source/common/primitives.cpp Fri Mar 27 22:59:30 2015 -0500<br>
> +++ b/source/common/primitives.cpp Wed Mar 25 19:39:35 2015 +0530<br>
> @@ -90,7 +90,6 @@<br>
><br>
> /* alias chroma 4:4:4 from luma primitives (all but chroma filters) */<br>
><br>
> - p.chroma[X265_CSP_I444].p2s = p.luma_p2s;<br>
> p.chroma[X265_CSP_I444].cu[BLOCK_4x4].sa8d = NULL;<br>
><br>
> for (int i = 0; i < NUM_PU_SIZES; i++)<br>
> @@ -98,7 +97,7 @@<br>
> p.chroma[X265_CSP_I444].pu[i].copy_pp = p.pu[i].copy_pp;<br>
> p.chroma[X265_CSP_I444].pu[i].addAvg = p.pu[i].addAvg;<br>
> p.chroma[X265_CSP_I444].pu[i].satd = p.pu[i].satd;<br>
> - p.chroma[X265_CSP_I444].pu[i].chroma_p2s = p.pu[i].filter_p2s;<br>
> + p.chroma[X265_CSP_I444].pu[i].chroma_p2s = p.pu[i].luma_p2s;<br>
> }<br>
><br>
> for (int i = 0; i < NUM_CU_SIZES; i++)<br>
> diff -r 22a312799bb0 -r 8797b42373de source/common/primitives.h<br>
> --- a/source/common/primitives.h Fri Mar 27 22:59:30 2015 -0500<br>
> +++ b/source/common/primitives.h Wed Mar 25 19:39:35 2015 +0530<br>
> @@ -156,8 +156,7 @@<br>
> typedef void (*filter_sp_t) (const int16_t* src, intptr_t srcStride, pixel* dst, intptr_t dstStride, int coeffIdx);<br>
> typedef void (*filter_ss_t) (const int16_t* src, intptr_t srcStride, int16_t* dst, intptr_t dstStride, int coeffIdx);<br>
> typedef void (*filter_hv_pp_t) (const pixel* src, intptr_t srcStride, pixel* dst, intptr_t dstStride, int idxX, int idxY);<br>
> -typedef void (*filter_p2s_wxh_t)(const pixel* src, intptr_t srcStride, int16_t* dst, int width, int height);<br>
> -typedef void (*filter_p2s_t)(const pixel* src, intptr_t srcStride, int16_t* dst);<br>
> +typedef void (*filter_p2s_t)(const pixel* src, intptr_t srcStride, int16_t* dst, int16_t dstStride);<br>
><br>
> typedef void (*copy_pp_t)(pixel* dst, intptr_t dstStride, const pixel* src, intptr_t srcStride); // dst is aligned<br>
> typedef void (*copy_sp_t)(pixel* dst, intptr_t dstStride, const int16_t* src, intptr_t srcStride);<br>
> @@ -211,7 +210,7 @@<br>
> addAvg_t addAvg; // bidir motion compensation, uses 16bit values<br>
><br>
> copy_pp_t copy_pp;<br>
> - filter_p2s_t filter_p2s;<br>
> + filter_p2s_t luma_p2s;<br>
> }<br>
> pu[NUM_PU_SIZES];<br>
<br>
</div></div> the 'luma_' prefix is redundant, everything within pu[] is implied to<br>
be luma since it is not in chroma[].pu[]<br>
<div><div class="h5"><br>
><br>
> @@ -290,7 +289,6 @@<br>
> weightp_sp_t weight_sp;<br>
> weightp_pp_t weight_pp;<br>
><br>
> - filter_p2s_wxh_t luma_p2s;<br>
><br>
> findPosLast_t findPosLast;<br>
><br>
> @@ -337,7 +335,6 @@<br>
> }<br>
> cu[NUM_CU_SIZES];<br>
><br>
> - filter_p2s_wxh_t p2s; // takes width/height as arguments<br>
> }<br>
> chroma[X265_CSP_COUNT];<br>
> };<br>
> diff -r 22a312799bb0 -r 8797b42373de source/common/x86/asm-primitives.cpp<br>
> --- a/source/common/x86/asm-primitives.cpp Fri Mar 27 22:59:30 2015 -0500<br>
> +++ b/source/common/x86/asm-primitives.cpp Wed Mar 25 19:39:35 2015 +0530<br>
> @@ -859,9 +859,6 @@<br>
> PIXEL_AVG_W4(mmx2);<br>
> LUMA_VAR(sse2);<br>
><br>
> - p.luma_p2s = x265_luma_p2s_sse2;<br>
> - p.chroma[X265_CSP_I420].p2s = x265_chroma_p2s_sse2;<br>
> - p.chroma[X265_CSP_I422].p2s = x265_chroma_p2s_sse2;<br>
><br>
> ALL_LUMA_TU(blockfill_s, blockfill_s, sse2);<br>
> ALL_LUMA_TU_S(cpy1Dto2D_shr, cpy1Dto2D_shr_, sse2);<br>
> @@ -1249,31 +1246,7 @@<br>
> ASSIGN_SSE_PP(ssse3);<br>
> <a href="http://p.cu" target="_blank">p.cu</a>[BLOCK_4x4].sse_pp = x265_pixel_ssd_4x4_ssse3;<br>
> p.chroma[X265_CSP_I422].cu[BLOCK_422_4x8].sse_pp = x265_pixel_ssd_4x8_ssse3;<br>
> - p.pu[LUMA_4x4].filter_p2s = x265_pixelToShort_4x4_ssse3;<br>
> - p.pu[LUMA_4x8].filter_p2s = x265_pixelToShort_4x8_ssse3;<br>
> - p.pu[LUMA_4x16].filter_p2s = x265_pixelToShort_4x16_ssse3;<br>
> - p.pu[LUMA_8x4].filter_p2s = x265_pixelToShort_8x4_ssse3;<br>
> - p.pu[LUMA_8x8].filter_p2s = x265_pixelToShort_8x8_ssse3;<br>
> - p.pu[LUMA_8x16].filter_p2s = x265_pixelToShort_8x16_ssse3;<br>
> - p.pu[LUMA_8x32].filter_p2s = x265_pixelToShort_8x32_ssse3;<br>
> - p.pu[LUMA_16x4].filter_p2s = x265_pixelToShort_16x4_ssse3;<br>
> - p.pu[LUMA_16x8].filter_p2s = x265_pixelToShort_16x8_ssse3;<br>
> - p.pu[LUMA_16x12].filter_p2s = x265_pixelToShort_16x12_ssse3;<br>
> - p.pu[LUMA_16x16].filter_p2s = x265_pixelToShort_16x16_ssse3;<br>
> - p.pu[LUMA_16x32].filter_p2s = x265_pixelToShort_16x32_ssse3;<br>
> - p.pu[LUMA_16x64].filter_p2s = x265_pixelToShort_16x64_ssse3;<br>
> - p.pu[LUMA_32x8].filter_p2s = x265_pixelToShort_32x8_ssse3;<br>
> - p.pu[LUMA_32x16].filter_p2s = x265_pixelToShort_32x16_ssse3;<br>
> - p.pu[LUMA_32x24].filter_p2s = x265_pixelToShort_32x24_ssse3;<br>
> - p.pu[LUMA_32x32].filter_p2s = x265_pixelToShort_32x32_ssse3;<br>
> - p.pu[LUMA_32x64].filter_p2s = x265_pixelToShort_32x64_ssse3;<br>
> - p.pu[LUMA_64x16].filter_p2s = x265_pixelToShort_64x16_ssse3;<br>
> - p.pu[LUMA_64x32].filter_p2s = x265_pixelToShort_64x32_ssse3;<br>
> - p.pu[LUMA_64x48].filter_p2s = x265_pixelToShort_64x48_ssse3;<br>
> - p.pu[LUMA_64x64].filter_p2s = x265_pixelToShort_64x64_ssse3;<br>
><br>
> - p.chroma[X265_CSP_I420].p2s = x265_chroma_p2s_ssse3;<br>
> - p.chroma[X265_CSP_I422].p2s = x265_chroma_p2s_ssse3;<br>
><br>
> p.dst4x4 = x265_dst4_ssse3;<br>
> <a href="http://p.cu" target="_blank">p.cu</a>[BLOCK_8x8].idct = x265_idct8_ssse3;<br>
> diff -r 22a312799bb0 -r 8797b42373de source/common/x86/ipfilter8.asm<br>
> --- a/source/common/x86/ipfilter8.asm Fri Mar 27 22:59:30 2015 -0500<br>
> +++ b/source/common/x86/ipfilter8.asm Wed Mar 25 19:39:35 2015 +0530<br>
> @@ -7740,320 +7740,6 @@<br>
> FILTER_V4_W16n_H2 64, 48<br>
> FILTER_V4_W16n_H2 48, 64<br>
> FILTER_V4_W16n_H2 64, 16<br>
> -;-----------------------------------------------------------------------------<br>
> -; void pixelToShort(pixel *src, intptr_t srcStride, int16_t *dst, int width, int height)<br>
> -;-----------------------------------------------------------------------------<br>
> -%macro PIXEL_WH_4xN 2<br>
> -INIT_XMM ssse3<br>
> -cglobal pixelToShort_%1x%2, 3, 7, 6<br>
> -<br>
> - ; load width and height<br>
> - mov r3d, %1<br>
> - mov r4d, %2<br>
> - ; load constant<br>
> - mova m4, [pb_128]<br>
> - mova m5, [tab_c_64_n64]<br>
> -.loopH:<br>
> - xor r5d, r5d<br>
> -<br>
> -.loopW:<br>
> - mov r6, r0<br>
> - movh m0, [r6]<br>
> - punpcklbw m0, m4<br>
> - pmaddubsw m0, m5<br>
> -<br>
> - movh m1, [r6 + r1]<br>
> - punpcklbw m1, m4<br>
> - pmaddubsw m1, m5<br>
> -<br>
> - movh m2, [r6 + r1 * 2]<br>
> - punpcklbw m2, m4<br>
> - pmaddubsw m2, m5<br>
> -<br>
> - lea r6, [r6 + r1 * 2]<br>
> - movh m3, [r6 + r1]<br>
> - punpcklbw m3, m4<br>
> - pmaddubsw m3, m5<br>
> -<br>
> - add r5, 8<br>
> - cmp r5, r3<br>
> - jg .width4<br>
> - movu [r2 + r5 * 2 + FENC_STRIDE * 0 - 16], m0<br>
> - movu [r2 + r5 * 2 + FENC_STRIDE * 2 - 16], m1<br>
> - movu [r2 + r5 * 2 + FENC_STRIDE * 4 - 16], m2<br>
> - movu [r2 + r5 * 2 + FENC_STRIDE * 6 - 16], m3<br>
> - je .nextH<br>
> - jmp .loopW<br>
> -<br>
> -.width4:<br>
> - movh [r2 + r5 * 2 + FENC_STRIDE * 0 - 16], m0<br>
> - movh [r2 + r5 * 2 + FENC_STRIDE * 2 - 16], m1<br>
> - movh [r2 + r5 * 2 + FENC_STRIDE * 4 - 16], m2<br>
> - movh [r2 + r5 * 2 + FENC_STRIDE * 6 - 16], m3<br>
> -<br>
> -.nextH:<br>
> - lea r0, [r0 + r1 * 4]<br>
> - add r2, FENC_STRIDE * 8<br>
> -<br>
> - sub r4d, 4<br>
> - jnz .loopH<br>
> - RET<br>
> -%endmacro<br>
> -PIXEL_WH_4xN 4, 4<br>
> -PIXEL_WH_4xN 4, 8<br>
> -PIXEL_WH_4xN 4, 16<br>
> -<br>
> -;-----------------------------------------------------------------------------<br>
> -; void pixelToShort(pixel *src, intptr_t srcStride, int16_t *dst, int width, int height)<br>
> -;-----------------------------------------------------------------------------<br>
> -%macro PIXEL_WH_8xN 2<br>
> -INIT_XMM ssse3<br>
> -cglobal pixelToShort_%1x%2, 3, 7, 6<br>
> -<br>
> - ; load width and height<br>
> - mov r3d, %1<br>
> - mov r4d, %2<br>
> -<br>
> - ; load constant<br>
> - mova m4, [pb_128]<br>
> - mova m5, [tab_c_64_n64]<br>
> -<br>
> -.loopH<br>
> - xor r5d, r5d<br>
> -.loopW<br>
> - lea r6, [r0 + r5]<br>
> -<br>
> - movh m0, [r6]<br>
> - punpcklbw m0, m4<br>
> - pmaddubsw m0, m5<br>
> -<br>
> - movh m1, [r6 + r1]<br>
> - punpcklbw m1, m4<br>
> - pmaddubsw m1, m5<br>
> -<br>
> - movh m2, [r6 + r1 * 2]<br>
> - punpcklbw m2, m4<br>
> - pmaddubsw m2, m5<br>
> -<br>
> - lea r6, [r6 + r1 * 2]<br>
> - movh m3, [r6 + r1]<br>
> - punpcklbw m3, m4<br>
> - pmaddubsw m3, m5<br>
> -<br>
> - add r5, 8<br>
> - cmp r5, r3<br>
> -<br>
> - movu [r2 + FENC_STRIDE * 0], m0<br>
> - movu [r2 + FENC_STRIDE * 2], m1<br>
> - movu [r2 + FENC_STRIDE * 4], m2<br>
> - movu [r2 + FENC_STRIDE * 6], m3<br>
> -<br>
> - je .nextH<br>
> - jmp .loopW<br>
> -<br>
> -<br>
> -.nextH:<br>
> - lea r0, [r0 + r1 * 4]<br>
> - add r2, FENC_STRIDE * 8<br>
> -<br>
> - sub r4d, 4<br>
> - jnz .loopH<br>
> - RET<br>
> -%endmacro<br>
> -PIXEL_WH_8xN 8, 8<br>
> -PIXEL_WH_8xN 8, 4<br>
> -PIXEL_WH_8xN 8, 16<br>
> -PIXEL_WH_8xN 8, 32<br>
> -<br>
> -<br>
> -;-----------------------------------------------------------------------------<br>
> -; void pixelToShort(pixel *src, intptr_t srcStride, int16_t *dst, int width, int height)<br>
> -;-----------------------------------------------------------------------------<br>
> -%macro PIXEL_WH_16xN 2<br>
> -INIT_XMM ssse3<br>
> -cglobal pixelToShort_%1x%2, 3, 7, 6<br>
> -<br>
> - ; load width and height<br>
> - mov r3d, %1<br>
> - mov r4d, %2<br>
> -<br>
> - ; load constant<br>
> - mova m4, [pb_128]<br>
> - mova m5, [tab_c_64_n64]<br>
> -<br>
> -.loopH:<br>
> - xor r5d, r5d<br>
> -.loopW:<br>
> - lea r6, [r0 + r5]<br>
> -<br>
> - movh m0, [r6]<br>
> - punpcklbw m0, m4<br>
> - pmaddubsw m0, m5<br>
> -<br>
> - movh m1, [r6 + r1]<br>
> - punpcklbw m1, m4<br>
> - pmaddubsw m1, m5<br>
> -<br>
> - movh m2, [r6 + r1 * 2]<br>
> - punpcklbw m2, m4<br>
> - pmaddubsw m2, m5<br>
> -<br>
> - lea r6, [r6 + r1 * 2]<br>
> - movh m3, [r6 + r1]<br>
> - punpcklbw m3, m4<br>
> - pmaddubsw m3, m5<br>
> -<br>
> - add r5, 8<br>
> - cmp r5, r3<br>
> -<br>
> - movu [r2 + r5 * 2 + FENC_STRIDE * 0 - 16], m0<br>
> - movu [r2 + r5 * 2 + FENC_STRIDE * 2 - 16], m1<br>
> - movu [r2 + r5 * 2 + FENC_STRIDE * 4 - 16], m2<br>
> - movu [r2 + r5 * 2 + FENC_STRIDE * 6 - 16], m3<br>
> - je .nextH<br>
> - jmp .loopW<br>
> -<br>
> -<br>
> -.nextH:<br>
> - lea r0, [r0 + r1 * 4]<br>
> - add r2, FENC_STRIDE * 8<br>
> -<br>
> - sub r4d, 4<br>
> - jnz .loopH<br>
> -<br>
> - RET<br>
> -%endmacro<br>
> -PIXEL_WH_16xN 16, 16<br>
> -PIXEL_WH_16xN 16, 8<br>
> -PIXEL_WH_16xN 16, 4<br>
> -PIXEL_WH_16xN 16, 12<br>
> -PIXEL_WH_16xN 16, 32<br>
> -PIXEL_WH_16xN 16, 64<br>
> -<br>
> -;-----------------------------------------------------------------------------<br>
> -; void pixelToShort(pixel *src, intptr_t srcStride, int16_t *dst, int width, int height)<br>
> -;-----------------------------------------------------------------------------<br>
> -%macro PIXEL_WH_32xN 2<br>
> -INIT_XMM ssse3<br>
> -cglobal pixelToShort_%1x%2, 3, 7, 6<br>
> -<br>
> - ; load width and height<br>
> - mov r3d, %1<br>
> - mov r4d, %2<br>
> -<br>
> - ; load constant<br>
> - mova m4, [pb_128]<br>
> - mova m5, [tab_c_64_n64]<br>
> -<br>
> -.loopH:<br>
> - xor r5d, r5d<br>
> -.loopW:<br>
> - lea r6, [r0 + r5]<br>
> -<br>
> - movh m0, [r6]<br>
> - punpcklbw m0, m4<br>
> - pmaddubsw m0, m5<br>
> -<br>
> - movh m1, [r6 + r1]<br>
> - punpcklbw m1, m4<br>
> - pmaddubsw m1, m5<br>
> -<br>
> - movh m2, [r6 + r1 * 2]<br>
> - punpcklbw m2, m4<br>
> - pmaddubsw m2, m5<br>
> -<br>
> - lea r6, [r6 + r1 * 2]<br>
> - movh m3, [r6 + r1]<br>
> - punpcklbw m3, m4<br>
> - pmaddubsw m3, m5<br>
> -<br>
> - add r5, 8<br>
> - cmp r5, r3<br>
> -<br>
> - movu [r2 + r5 * 2 + FENC_STRIDE * 0 - 16], m0<br>
> - movu [r2 + r5 * 2 + FENC_STRIDE * 2 - 16], m1<br>
> - movu [r2 + r5 * 2 + FENC_STRIDE * 4 - 16], m2<br>
> - movu [r2 + r5 * 2 + FENC_STRIDE * 6 - 16], m3<br>
> - je .nextH<br>
> - jmp .loopW<br>
> -<br>
> -<br>
> -.nextH:<br>
> - lea r0, [r0 + r1 * 4]<br>
> - add r2, FENC_STRIDE * 8<br>
> -<br>
> - sub r4d, 4<br>
> - jnz .loopH<br>
> -<br>
> - RET<br>
> -%endmacro<br>
> -PIXEL_WH_32xN 32, 32<br>
> -PIXEL_WH_32xN 32, 8<br>
> -PIXEL_WH_32xN 32, 16<br>
> -PIXEL_WH_32xN 32, 24<br>
> -PIXEL_WH_32xN 32, 64<br>
> -<br>
> -;-----------------------------------------------------------------------------<br>
> -; void pixelToShort(pixel *src, intptr_t srcStride, int16_t *dst, int width, int height)<br>
> -;-----------------------------------------------------------------------------<br>
> -%macro PIXEL_WH_64xN 2<br>
> -INIT_XMM ssse3<br>
> -cglobal pixelToShort_%1x%2, 3, 7, 6<br>
> -<br>
> - ; load width and height<br>
> - mov r3d, %1<br>
> - mov r4d, %2<br>
> -<br>
> - ; load constant<br>
> - mova m4, [pb_128]<br>
> - mova m5, [tab_c_64_n64]<br>
> -<br>
> -.loopH:<br>
> - xor r5d, r5d<br>
> -.loopW:<br>
> - lea r6, [r0 + r5]<br>
> -<br>
> - movh m0, [r6]<br>
> - punpcklbw m0, m4<br>
> - pmaddubsw m0, m5<br>
> -<br>
> - movh m1, [r6 + r1]<br>
> - punpcklbw m1, m4<br>
> - pmaddubsw m1, m5<br>
> -<br>
> - movh m2, [r6 + r1 * 2]<br>
> - punpcklbw m2, m4<br>
> - pmaddubsw m2, m5<br>
> -<br>
> - lea r6, [r6 + r1 * 2]<br>
> - movh m3, [r6 + r1]<br>
> - punpcklbw m3, m4<br>
> - pmaddubsw m3, m5<br>
> -<br>
> - add r5, 8<br>
> - cmp r5, r3<br>
> -<br>
> - movu [r2 + r5 * 2 + FENC_STRIDE * 0 - 16], m0<br>
> - movu [r2 + r5 * 2 + FENC_STRIDE * 2 - 16], m1<br>
> - movu [r2 + r5 * 2 + FENC_STRIDE * 4 - 16], m2<br>
> - movu [r2 + r5 * 2 + FENC_STRIDE * 6 - 16], m3<br>
> - je .nextH<br>
> - jmp .loopW<br>
> -<br>
> -<br>
> -.nextH:<br>
> - lea r0, [r0 + r1 * 4]<br>
> - add r2, FENC_STRIDE * 8<br>
> -<br>
> - sub r4d, 4<br>
> - jnz .loopH<br>
> -<br>
> - RET<br>
> -%endmacro<br>
> -PIXEL_WH_64xN 64, 64<br>
> -PIXEL_WH_64xN 64, 16<br>
> -PIXEL_WH_64xN 64, 32<br>
> -PIXEL_WH_64xN 64, 48<br>
><br>
> %macro PROCESS_LUMA_W4_4R 0<br>
> movd m0, [r0]<br>
> diff -r 22a312799bb0 -r 8797b42373de source/common/x86/ipfilter8.h<br>
> --- a/source/common/x86/ipfilter8.h Fri Mar 27 22:59:30 2015 -0500<br>
> +++ b/source/common/x86/ipfilter8.h Wed Mar 25 19:39:35 2015 +0530<br>
> @@ -289,8 +289,6 @@<br>
> SETUP_CHROMA_420_HORIZ_FUNC_DEF(64, 16, cpu); \<br>
> SETUP_CHROMA_420_HORIZ_FUNC_DEF(16, 64, cpu)<br>
><br>
> -void x265_chroma_p2s_sse2(const pixel* src, intptr_t srcStride, int16_t* dst, int width, int height);<br>
> -void x265_luma_p2s_sse2(const pixel* src, intptr_t srcStride, int16_t* dst, int width, int height);<br>
><br>
> CHROMA_420_VERT_FILTERS(_sse2);<br>
> CHROMA_420_HORIZ_FILTERS(_sse4);<br>
> @@ -624,28 +622,6 @@<br>
> LUMA_SP_FILTERS(_avx2);<br>
> LUMA_SS_FILTERS(_avx2);<br>
> void x265_interp_8tap_hv_pp_8x8_sse4(const pixel* src, intptr_t srcStride, pixel* dst, intptr_t dstStride, int idxX, int idxY);<br>
> -void x265_pixelToShort_4x4_ssse3(const pixel* src, intptr_t srcStride, int16_t* dst);<br>
> -void x265_pixelToShort_4x8_ssse3(const pixel* src, intptr_t srcStride, int16_t* dst);<br>
> -void x265_pixelToShort_4x16_ssse3(const pixel* src, intptr_t srcStride, int16_t* dst);<br>
> -void x265_pixelToShort_8x4_ssse3(const pixel* src, intptr_t srcStride, int16_t* dst);<br>
> -void x265_pixelToShort_8x8_ssse3(const pixel* src, intptr_t srcStride, int16_t* dst);<br>
> -void x265_pixelToShort_8x16_ssse3(const pixel* src, intptr_t srcStride, int16_t* dst);<br>
> -void x265_pixelToShort_8x32_ssse3(const pixel* src, intptr_t srcStride, int16_t* dst);<br>
> -void x265_pixelToShort_16x4_ssse3(const pixel* src, intptr_t srcStride, int16_t* dst);<br>
> -void x265_pixelToShort_16x8_ssse3(const pixel* src, intptr_t srcStride, int16_t* dst);<br>
> -void x265_pixelToShort_16x12_ssse3(const pixel* src, intptr_t srcStride, int16_t* dst);<br>
> -void x265_pixelToShort_16x16_ssse3(const pixel* src, intptr_t srcStride, int16_t* dst);<br>
> -void x265_pixelToShort_16x32_ssse3(const pixel* src, intptr_t srcStride, int16_t* dst);<br>
> -void x265_pixelToShort_16x64_ssse3(const pixel* src, intptr_t srcStride, int16_t* dst);<br>
> -void x265_pixelToShort_32x8_ssse3(const pixel* src, intptr_t srcStride, int16_t* dst);<br>
> -void x265_pixelToShort_32x16_ssse3(const pixel* src, intptr_t srcStride, int16_t* dst);<br>
> -void x265_pixelToShort_32x24_ssse3(const pixel* src, intptr_t srcStride, int16_t* dst);<br>
> -void x265_pixelToShort_32x32_ssse3(const pixel* src, intptr_t srcStride, int16_t* dst);<br>
> -void x265_pixelToShort_32x64_ssse3(const pixel* src, intptr_t srcStride, int16_t* dst);<br>
> -void x265_pixelToShort_64x16_ssse3(const pixel* src, intptr_t srcStride, int16_t* dst);<br>
> -void x265_pixelToShort_64x32_ssse3(const pixel* src, intptr_t srcStride, int16_t* dst);<br>
> -void x265_pixelToShort_64x48_ssse3(const pixel* src, intptr_t srcStride, int16_t* dst);<br>
> -void x265_pixelToShort_64x64_ssse3(const pixel* src, intptr_t srcStride, int16_t* dst);<br>
> #undef LUMA_FILTERS<br>
> #undef LUMA_SP_FILTERS<br>
> #undef LUMA_SS_FILTERS<br>
> diff -r 22a312799bb0 -r 8797b42373de source/test/ipfilterharness.cpp<br>
> --- a/source/test/ipfilterharness.cpp Fri Mar 27 22:59:30 2015 -0500<br>
> +++ b/source/test/ipfilterharness.cpp Wed Mar 25 19:39:35 2015 +0530<br>
> @@ -61,55 +61,6 @@<br>
> }<br>
> }<br>
><br>
> -bool IPFilterHarness::check_IPFilter_primitive(filter_p2s_wxh_t ref, filter_p2s_wxh_t opt, int isChroma, int csp)<br>
> -{<br>
> - intptr_t rand_srcStride;<br>
> - int min_size = isChroma ? 2 : 4;<br>
> - int max_size = isChroma ? (MAX_CU_SIZE >> 1) : MAX_CU_SIZE;<br>
> -<br>
> - if (isChroma && (csp == X265_CSP_I444))<br>
> - {<br>
> - min_size = 4;<br>
> - max_size = MAX_CU_SIZE;<br>
> - }<br>
> -<br>
> - for (int i = 0; i < ITERS; i++)<br>
> - {<br>
> - int index = i % TEST_CASES;<br>
> - int rand_height = (int16_t)rand() % 100;<br>
> - int rand_width = (int16_t)rand() % 100;<br>
> -<br>
> - rand_srcStride = rand_width + rand() % 100;<br>
> - if (rand_srcStride < rand_width)<br>
> - rand_srcStride = rand_width;<br>
> -<br>
> - rand_width &= ~(min_size - 1);<br>
> - rand_width = x265_clip3(min_size, max_size, rand_width);<br>
> -<br>
> - rand_height &= ~(min_size - 1);<br>
> - rand_height = x265_clip3(min_size, max_size, rand_height);<br>
> -<br>
> - ref(pixel_test_buff[index],<br>
> - rand_srcStride,<br>
> - IPF_C_output_s,<br>
> - rand_width,<br>
> - rand_height);<br>
> -<br>
> - checked(opt, pixel_test_buff[index],<br>
> - rand_srcStride,<br>
> - IPF_vec_output_s,<br>
> - rand_width,<br>
> - rand_height);<br>
> -<br>
> - if (memcmp(IPF_vec_output_s, IPF_C_output_s, TEST_BUF_SIZE * sizeof(int16_t)))<br>
> - return false;<br>
> -<br>
> - reportfail();<br>
> - }<br>
> -<br>
> - return true;<br>
> -}<br>
> -<br>
> bool IPFilterHarness::check_IPFilterChroma_primitive(filter_pp_t ref, filter_pp_t opt)<br>
> {<br>
> intptr_t rand_srcStride, rand_dstStride;<br>
> @@ -518,12 +469,13 @@<br>
> {<br>
> intptr_t rand_srcStride = rand() % 100;<br>
> int index = i % TEST_CASES;<br>
> + int16_t dstStride = MAX_CU_SIZE;<br>
><br>
> - ref(pixel_test_buff[index] + i, rand_srcStride, IPF_C_output_s);<br>
> + ref(pixel_test_buff[index] + i, rand_srcStride, IPF_C_output_s, dstStride);<br>
><br>
> - checked(opt, pixel_test_buff[index] + i, rand_srcStride, IPF_vec_output_s);<br>
> + checked(opt, pixel_test_buff[index] + i, rand_srcStride, IPF_vec_output_s, dstStride);<br>
><br>
> - if (memcmp(IPF_vec_output_s, IPF_C_output_s, TEST_BUF_SIZE * sizeof(pixel)))<br>
> + if (memcmp(IPF_vec_output_s, IPF_C_output_s, TEST_BUF_SIZE * sizeof(int16_t)))<br>
> return false;<br>
><br>
> reportfail();<br>
> @@ -532,18 +484,24 @@<br>
> return true;<br>
> }<br>
><br>
> -bool IPFilterHarness::check_IPFilterChromaP2S_primitive(filter_p2s_t ref, filter_p2s_t opt)<br>
> +bool IPFilterHarness::check_IPFilterChromaP2S_primitive(filter_p2s_t ref, filter_p2s_t opt, int csp)<br>
> {<br>
> for (int i = 0; i < ITERS; i++)<br>
> {<br>
> intptr_t rand_srcStride = rand() % 100;<br>
> int index = i % TEST_CASES;<br>
> + int16_t dstStride;<br>
><br>
> - ref(pixel_test_buff[index] + i, rand_srcStride, IPF_C_output_s);<br>
> + if (csp == X265_CSP_I444)<br>
> + dstStride = MAX_CU_SIZE;<br>
> + else<br>
> + dstStride = MAX_CU_SIZE >> (int16_t)(csp != X265_CSP_I444);<br>
><br>
> - checked(opt, pixel_test_buff[index] + i, rand_srcStride, IPF_vec_output_s);<br>
> + ref(pixel_test_buff[index] + i, rand_srcStride, IPF_C_output_s, dstStride);<br>
><br>
> - if (memcmp(IPF_vec_output_s, IPF_C_output_s, TEST_BUF_SIZE * sizeof(pixel)))<br>
> + checked(opt, pixel_test_buff[index] + i, rand_srcStride, IPF_vec_output_s, dstStride);<br>
> +<br>
> + if (memcmp(IPF_vec_output_s, IPF_C_output_s, TEST_BUF_SIZE * sizeof(int16_t)))<br>
> return false;<br>
><br>
> reportfail();<br>
> @@ -554,15 +512,6 @@<br>
><br>
> bool IPFilterHarness::testCorrectness(const EncoderPrimitives& ref, const EncoderPrimitives& opt)<br>
> {<br>
> - if (opt.luma_p2s)<br>
> - {<br>
> - // last parameter does not matter in case of luma<br>
> - if (!check_IPFilter_primitive(ref.luma_p2s, opt.luma_p2s, 0, 1))<br>
> - {<br>
> - printf("luma_p2s failed\n");<br>
> - return false;<br>
> - }<br>
> - }<br>
><br>
> for (int value = 0; value < NUM_PU_SIZES; value++)<br>
> {<br>
> @@ -622,11 +571,11 @@<br>
> return false;<br>
> }<br>
> }<br>
> - if (opt.pu[value].filter_p2s)<br>
> + if (opt.pu[value].luma_p2s)<br>
> {<br>
> - if (!check_IPFilterLumaP2S_primitive(ref.pu[value].filter_p2s, opt.pu[value].filter_p2s))<br>
> + if (!check_IPFilterLumaP2S_primitive(ref.pu[value].luma_p2s, opt.pu[value].luma_p2s))<br>
> {<br>
> - printf("filter_p2s[%s]", lumaPartStr[value]);<br>
> + printf("luma_p2s[%s]", lumaPartStr[value]);<br>
> return false;<br>
> }<br>
> }<br>
> @@ -634,14 +583,6 @@<br>
><br>
> for (int csp = X265_CSP_I420; csp < X265_CSP_COUNT; csp++)<br>
> {<br>
> - if (opt.chroma[csp].p2s)<br>
> - {<br>
> - if (!check_IPFilter_primitive(ref.chroma[csp].p2s, opt.chroma[csp].p2s, 1, csp))<br>
> - {<br>
> - printf("chroma_p2s[%s]", x265_source_csp_names[csp]);<br>
> - return false;<br>
> - }<br>
> - }<br>
> for (int value = 0; value < NUM_PU_SIZES; value++)<br>
> {<br>
> if (opt.chroma[csp].pu[value].filter_hpp)<br>
> @@ -694,7 +635,7 @@<br>
> }<br>
> if (opt.chroma[csp].pu[value].chroma_p2s)<br>
> {<br>
> - if (!check_IPFilterChromaP2S_primitive(ref.chroma[csp].pu[value].chroma_p2s, opt.chroma[csp].pu[value].chroma_p2s))<br>
> + if (!check_IPFilterChromaP2S_primitive(ref.chroma[csp].pu[value].chroma_p2s, opt.chroma[csp].pu[value].chroma_p2s, csp))<br>
> {<br>
> printf("chroma_p2s[%s]", chromaPartStr[csp][value]);<br>
> return false;<br>
> @@ -708,19 +649,10 @@<br>
><br>
> void IPFilterHarness::measureSpeed(const EncoderPrimitives& ref, const EncoderPrimitives& opt)<br>
> {<br>
> - int height = 64;<br>
> - int width = 64;<br>
> int16_t srcStride = 96;<br>
> int16_t dstStride = 96;<br>
> int maxVerticalfilterHalfDistance = 3;<br>
><br>
> - if (opt.luma_p2s)<br>
> - {<br>
> - printf("luma_p2s\t");<br>
> - REPORT_SPEEDUP(opt.luma_p2s, ref.luma_p2s,<br>
> - pixel_buff, srcStride, IPF_vec_output_s, width, height);<br>
> - }<br>
> -<br>
> for (int value = 0; value < NUM_PU_SIZES; value++)<br>
> {<br>
> if (opt.pu[value].luma_hpp)<br>
> @@ -777,23 +709,18 @@<br>
> pixel_buff + 3 * srcStride, srcStride, IPF_vec_output_p, srcStride, 1, 3);<br>
> }<br>
><br>
> - if (opt.pu[value].filter_p2s)<br>
> + if (opt.pu[value].luma_p2s)<br>
> {<br>
> - printf("filter_p2s [%s]\t", lumaPartStr[value]);<br>
> - REPORT_SPEEDUP(opt.pu[value].filter_p2s, ref.pu[value].filter_p2s,<br>
> - pixel_buff, srcStride, IPF_vec_output_s);<br>
> + printf("luma_p2s[%s]\t", lumaPartStr[value]);<br>
> + REPORT_SPEEDUP(opt.pu[value].luma_p2s, ref.pu[value].luma_p2s,<br>
> + pixel_buff, srcStride,<br>
> + IPF_vec_output_s, dstStride);<br>
> }<br>
> }<br>
><br>
> for (int csp = X265_CSP_I420; csp < X265_CSP_COUNT; csp++)<br>
> {<br>
> printf("= Color Space %s =\n", x265_source_csp_names[csp]);<br>
> - if (opt.chroma[csp].p2s)<br>
> - {<br>
> - printf("chroma_p2s\t");<br>
> - REPORT_SPEEDUP(opt.chroma[csp].p2s, ref.chroma[csp].p2s,<br>
> - pixel_buff, srcStride, IPF_vec_output_s, width, height);<br>
> - }<br>
> for (int value = 0; value < NUM_PU_SIZES; value++)<br>
> {<br>
> if (opt.chroma[csp].pu[value].filter_hpp)<br>
> @@ -836,13 +763,12 @@<br>
> short_buff + maxVerticalfilterHalfDistance * srcStride, srcStride,<br>
> IPF_vec_output_s, dstStride, 1);<br>
> }<br>
> -<br>
> if (opt.chroma[csp].pu[value].chroma_p2s)<br>
> {<br>
> printf("chroma_p2s[%s]\t", chromaPartStr[csp][value]);<br>
> REPORT_SPEEDUP(opt.chroma[csp].pu[value].chroma_p2s, ref.chroma[csp].pu[value].chroma_p2s,<br>
> pixel_buff, srcStride,<br>
> - IPF_vec_output_s);<br>
> + IPF_vec_output_s, dstStride);<br>
> }<br>
> }<br>
> }<br>
> diff -r 22a312799bb0 -r 8797b42373de source/test/ipfilterharness.h<br>
> --- a/source/test/ipfilterharness.h Fri Mar 27 22:59:30 2015 -0500<br>
> +++ b/source/test/ipfilterharness.h Wed Mar 25 19:39:35 2015 +0530<br>
> @@ -50,7 +50,6 @@<br>
> pixel pixel_test_buff[TEST_CASES][TEST_BUF_SIZE];<br>
> int16_t short_test_buff[TEST_CASES][TEST_BUF_SIZE];<br>
><br>
> - bool check_IPFilter_primitive(filter_p2s_wxh_t ref, filter_p2s_wxh_t opt, int isChroma, int csp);<br>
> bool check_IPFilterChroma_primitive(filter_pp_t ref, filter_pp_t opt);<br>
> bool check_IPFilterChroma_ps_primitive(filter_ps_t ref, filter_ps_t opt);<br>
> bool check_IPFilterChroma_hps_primitive(filter_hps_t ref, filter_hps_t opt);<br>
> @@ -63,7 +62,7 @@<br>
> bool check_IPFilterLuma_ss_primitive(filter_ss_t ref, filter_ss_t opt);<br>
> bool check_IPFilterLumaHV_primitive(filter_hv_pp_t ref, filter_hv_pp_t opt);<br>
> bool check_IPFilterLumaP2S_primitive(filter_p2s_t ref, filter_p2s_t opt);<br>
> - bool check_IPFilterChromaP2S_primitive(filter_p2s_t ref, filter_p2s_t opt);<br>
> + bool check_IPFilterChromaP2S_primitive(filter_p2s_t ref, filter_p2s_t opt, int csp);<br>
><br>
> public:<br>
><br>
><br>
</div></div>> _______________________________________________<br>
> x265-devel mailing list<br>
> <a href="mailto:x265-devel@videolan.org">x265-devel@videolan.org</a><br>
> <a href="https://mailman.videolan.org/listinfo/x265-devel" target="_blank">https://mailman.videolan.org/listinfo/x265-devel</a><br>
><br>
<span class="HOEnZb"><font color="#888888"><br>
--<br>
Steve Borho<br>
<br>
_______________________________________________<br>
x265-devel mailing list<br>
<a href="mailto:x265-devel@videolan.org">x265-devel@videolan.org</a><br>
<a href="https://mailman.videolan.org/listinfo/x265-devel" target="_blank">https://mailman.videolan.org/listinfo/x265-devel</a><br>
<br>
</font></span></blockquote></div><br></div>