<div dir="ltr"><div class="gmail_default" style="font-family:arial,helvetica,sans-serif;font-size:small">right.. you can send it to mailing list<br></div></div><div class="gmail_extra"><br><div class="gmail_quote">On Wed, Aug 19, 2015 at 3:26 PM,  <span dir="ltr"><<a href="mailto:aasaipriya@multicorewareinc.com" target="_blank">aasaipriya@multicorewareinc.com</a>></span> wrote:<br><blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><span class=""># HG changeset patch<br>
# User Aasaipriya Chandran <<a href="mailto:aasaipriya@multicorewareinc.com">aasaipriya@multicorewareinc.com</a>><br>
</span># Date 1439972978 -19800<br>
#      Wed Aug 19 13:59:38 2015 +0530<br>
# Node ID 8a45cff3182fa9f6e07493434711247d58f22cc4<br>
<span class=""># Parent  2980141a744a569ad6f60dbebdece76a4eababfd<br>
asm: disabled 10bpp AVX & AVX2 primitives having less than 10% speed up over SSE<br>
<br>
these primitives are slower than SSE primitives<br>
<br>
</span>diff -r 2980141a744a -r 8a45cff3182f source/common/x86/asm-primitives.cpp<br>
<span class="">--- a/source/common/x86/asm-primitives.cpp      Tue Aug 18 12:45:52 2015 +0530<br>
</span>+++ b/source/common/x86/asm-primitives.cpp      Wed Aug 19 13:59:38 2015 +0530<br>
<span class="">@@ -1185,7 +1185,6 @@<br>
         p.chroma[X265_CSP_I422].pu[CHROMA_422_8x32].satd = PFX(pixel_satd_8x32_avx);<br>
         p.chroma[X265_CSP_I422].pu[CHROMA_422_8x4].satd = PFX(pixel_satd_8x4_avx);<br>
<br>
-        ALL_LUMA_PU(satd, pixel_satd, avx);<br>
         p.chroma[X265_CSP_I420].pu[CHROMA_420_8x8].satd = PFX(pixel_satd_8x8_avx);<br>
         p.chroma[X265_CSP_I420].pu[CHROMA_420_8x4].satd = PFX(pixel_satd_8x4_avx);<br>
         p.chroma[X265_CSP_I420].pu[CHROMA_420_8x16].satd = PFX(pixel_satd_8x16_avx);<br>
@@ -1194,15 +1193,10 @@<br>
         p.chroma[X265_CSP_I420].pu[CHROMA_420_24x32].satd = PFX(pixel_satd_24x32_avx);<br>
         p.chroma[X265_CSP_I420].pu[CHROMA_420_4x16].satd = PFX(pixel_satd_4x16_avx);<br>
         p.chroma[X265_CSP_I420].pu[CHROMA_420_4x8].satd = PFX(pixel_satd_4x8_avx);<br>
-#if X265_DEPTH <= 10<br>
-        ASSIGN_SA8D(avx);<br>
-#endif<br>
+<br>
         p.chroma[X265_CSP_I420].cu[BLOCK_420_8x8].sa8d = PFX(pixel_sa8d_8x8_avx);<br>
         p.chroma[X265_CSP_I420].cu[BLOCK_420_16x16].sa8d = PFX(pixel_sa8d_16x16_avx);<br>
         p.chroma[X265_CSP_I420].cu[BLOCK_420_32x32].sa8d = PFX(pixel_sa8d_32x32_avx);<br>
-        LUMA_VAR(avx);<br>
-        p.ssim_4x4x2_core = PFX(pixel_ssim_4x4x2_core_avx);<br>
-        p.ssim_end_4 = PFX(pixel_ssim_end4_avx);<br>
<br>
         // copy_pp primitives<br>
         // 16 x N<br>
</span>@@ -1299,6 +1293,20 @@<br>
<span class="">         p.pu[LUMA_64x32].copy_pp = (copy_pp_t)PFX(blockcopy_ss_64x32_avx);<br>
         p.pu[LUMA_64x48].copy_pp = (copy_pp_t)PFX(blockcopy_ss_64x48_avx);<br>
         p.pu[LUMA_64x64].copy_pp = (copy_pp_t)PFX(blockcopy_ss_64x64_avx);<br>
+<br>
</span>+        /* The following primitives have been disabled since performance compared to SSE is negligible/negative */<br>
<span class="">+#if 0<br>
+        ALL_LUMA_PU(satd, pixel_satd, avx);<br>
+<br>
+        p.ssim_4x4x2_core = PFX(pixel_ssim_4x4x2_core_avx);<br>
+        p.ssim_end_4 = PFX(pixel_ssim_end4_avx);<br>
+<br>
+        LUMA_VAR(avx);<br>
+<br>
</span>+        #if X265_DEPTH <= 10<br>
+           ASSIGN_SA8D(avx);<br>
+        #endif<br>
+#endif<br>
     }<br>
     if (cpuMask & X265_CPU_XOP)<br>
     {<br>
@@ -1414,11 +1422,8 @@<br>
<span class="">         p.pu[LUMA_64x64].pixelavg_pp = PFX(pixel_avg_64x64_avx2);<br>
         p.pu[LUMA_48x64].pixelavg_pp = PFX(pixel_avg_48x64_avx2);<br>
<br>
-        p.pu[LUMA_8x4].addAvg   = PFX(addAvg_8x4_avx2);<br>
-        p.pu[LUMA_8x8].addAvg   = PFX(addAvg_8x8_avx2);<br>
         p.pu[LUMA_8x16].addAvg  = PFX(addAvg_8x16_avx2);<br>
         p.pu[LUMA_8x32].addAvg  = PFX(addAvg_8x32_avx2);<br>
-        p.pu[LUMA_12x16].addAvg = PFX(addAvg_12x16_avx2);<br>
         p.pu[LUMA_16x4].addAvg  = PFX(addAvg_16x4_avx2);<br>
         p.pu[LUMA_16x8].addAvg  = PFX(addAvg_16x8_avx2);<br>
         p.pu[LUMA_16x12].addAvg = PFX(addAvg_16x12_avx2);<br>
</span>@@ -1438,12 +1443,10 @@<br>
<span class="">         p.pu[LUMA_64x64].addAvg = PFX(addAvg_64x64_avx2);<br>
<br>
         p.chroma[X265_CSP_I420].pu[CHROMA_420_8x2].addAvg   = PFX(addAvg_8x2_avx2);<br>
-        p.chroma[X265_CSP_I420].pu[CHROMA_420_8x4].addAvg   = PFX(addAvg_8x4_avx2);<br>
         p.chroma[X265_CSP_I420].pu[CHROMA_420_8x6].addAvg   = PFX(addAvg_8x6_avx2);<br>
         p.chroma[X265_CSP_I420].pu[CHROMA_420_8x8].addAvg   = PFX(addAvg_8x8_avx2);<br>
         p.chroma[X265_CSP_I420].pu[CHROMA_420_8x16].addAvg  = PFX(addAvg_8x16_avx2);<br>
         p.chroma[X265_CSP_I420].pu[CHROMA_420_8x32].addAvg  = PFX(addAvg_8x32_avx2);<br>
-        p.chroma[X265_CSP_I420].pu[CHROMA_420_12x16].addAvg = PFX(addAvg_12x16_avx2);<br>
         p.chroma[X265_CSP_I420].pu[CHROMA_420_16x4].addAvg  = PFX(addAvg_16x4_avx2);<br>
         p.chroma[X265_CSP_I420].pu[CHROMA_420_16x8].addAvg  = PFX(addAvg_16x8_avx2);<br>
         p.chroma[X265_CSP_I420].pu[CHROMA_420_16x12].addAvg = PFX(addAvg_16x12_avx2);<br>
</span>@@ -1457,18 +1460,15 @@<br>
<span class="">         p.chroma[X265_CSP_I422].pu[CHROMA_422_8x16].addAvg = PFX(addAvg_8x16_avx2);<br>
         p.chroma[X265_CSP_I422].pu[CHROMA_422_16x32].addAvg = PFX(addAvg_16x32_avx2);<br>
         p.chroma[X265_CSP_I422].pu[CHROMA_422_32x64].addAvg = PFX(addAvg_32x64_avx2);<br>
-        p.chroma[X265_CSP_I422].pu[CHROMA_422_8x8].addAvg = PFX(addAvg_8x8_avx2);<br>
         p.chroma[X265_CSP_I422].pu[CHROMA_422_16x16].addAvg = PFX(addAvg_16x16_avx2);<br>
         p.chroma[X265_CSP_I422].pu[CHROMA_422_8x32].addAvg = PFX(addAvg_8x32_avx2);<br>
         p.chroma[X265_CSP_I422].pu[CHROMA_422_32x32].addAvg = PFX(addAvg_32x32_avx2);<br>
         p.chroma[X265_CSP_I422].pu[CHROMA_422_16x64].addAvg = PFX(addAvg_16x64_avx2);<br>
         p.chroma[X265_CSP_I422].pu[CHROMA_422_8x12].addAvg = PFX(addAvg_8x12_avx2);<br>
-        p.chroma[X265_CSP_I422].pu[CHROMA_422_8x4].addAvg = PFX(addAvg_8x4_avx2);<br>
         p.chroma[X265_CSP_I422].pu[CHROMA_422_16x24].addAvg = PFX(addAvg_16x24_avx2);<br>
         p.chroma[X265_CSP_I422].pu[CHROMA_422_16x8].addAvg = PFX(addAvg_16x8_avx2);<br>
         p.chroma[X265_CSP_I422].pu[CHROMA_422_8x64].addAvg = PFX(addAvg_8x64_avx2);<br>
         p.chroma[X265_CSP_I422].pu[CHROMA_422_24x64].addAvg = PFX(addAvg_24x64_avx2);<br>
-        p.chroma[X265_CSP_I422].pu[CHROMA_422_12x32].addAvg = PFX(addAvg_12x32_avx2);<br>
         p.chroma[X265_CSP_I422].pu[CHROMA_422_32x16].addAvg = PFX(addAvg_32x16_avx2);<br>
         p.chroma[X265_CSP_I422].pu[CHROMA_422_32x48].addAvg = PFX(addAvg_32x48_avx2);<br>
<br>
</span>@@ -1565,11 +1565,17 @@<br>
<span class="">         <a href="http://p.cu" rel="noreferrer" target="_blank">p.cu</a>[BLOCK_16x16].blockfill_s = PFX(blockfill_s_16x16_avx2);<br>
         <a href="http://p.cu" rel="noreferrer" target="_blank">p.cu</a>[BLOCK_32x32].blockfill_s = PFX(blockfill_s_32x32_avx2);<br>
<br>
-        ALL_LUMA_TU(count_nonzero, count_nonzero, avx2);<br>
-        ALL_LUMA_TU_S(cpy1Dto2D_shl, cpy1Dto2D_shl_, avx2);<br>
-        ALL_LUMA_TU_S(cpy1Dto2D_shr, cpy1Dto2D_shr_, avx2);<br>
-<br>
-        <a href="http://p.cu" rel="noreferrer" target="_blank">p.cu</a>[BLOCK_8x8].copy_cnt = PFX(copy_cnt_8_avx2);<br>
+        <a href="http://p.cu" rel="noreferrer" target="_blank">p.cu</a>[BLOCK_8x8].count_nonzero = PFX(count_nonzero_8x8_avx2);<br>
+        <a href="http://p.cu" rel="noreferrer" target="_blank">p.cu</a>[BLOCK_16x16].count_nonzero = PFX(count_nonzero_16x16_avx2);<br>
+        <a href="http://p.cu" rel="noreferrer" target="_blank">p.cu</a>[BLOCK_32x32].count_nonzero = PFX(count_nonzero_32x32_avx2);<br>
+<br>
+        <a href="http://p.cu" rel="noreferrer" target="_blank">p.cu</a>[BLOCK_16x16].cpy1Dto2D_shl = PFX(cpy1Dto2D_shl_16_avx2);<br>
+        <a href="http://p.cu" rel="noreferrer" target="_blank">p.cu</a>[BLOCK_32x32].cpy1Dto2D_shl = PFX(cpy1Dto2D_shl_32_avx2);<br>
+<br>
+        <a href="http://p.cu" rel="noreferrer" target="_blank">p.cu</a>[BLOCK_8x8].cpy1Dto2D_shr = PFX(cpy1Dto2D_shr_8_avx2);<br>
+        <a href="http://p.cu" rel="noreferrer" target="_blank">p.cu</a>[BLOCK_16x16].cpy1Dto2D_shr = PFX(cpy1Dto2D_shr_16_avx2);<br>
+        <a href="http://p.cu" rel="noreferrer" target="_blank">p.cu</a>[BLOCK_32x32].cpy1Dto2D_shr = PFX(cpy1Dto2D_shr_32_avx2);<br>
+<br>
         <a href="http://p.cu" rel="noreferrer" target="_blank">p.cu</a>[BLOCK_16x16].copy_cnt = PFX(copy_cnt_16_avx2);<br>
         <a href="http://p.cu" rel="noreferrer" target="_blank">p.cu</a>[BLOCK_32x32].copy_cnt = PFX(copy_cnt_32_avx2);<br>
<br>
</span>@@ -1766,48 +1772,29 @@<br>
<div><div class="h5">         p.chroma[X265_CSP_I420].pu[CHROMA_420_8x6].filter_hps = PFX(interp_4tap_horiz_ps_8x6_avx2);<br>
         p.chroma[X265_CSP_I420].pu[CHROMA_420_8x2].filter_hps = PFX(interp_4tap_horiz_ps_8x2_avx2);<br>
         p.chroma[X265_CSP_I420].pu[CHROMA_420_8x32].filter_hps = PFX(interp_4tap_horiz_ps_8x32_avx2);<br>
-        p.chroma[X265_CSP_I420].pu[CHROMA_420_16x16].filter_hps = PFX(interp_4tap_horiz_ps_16x16_avx2);<br>
         p.chroma[X265_CSP_I420].pu[CHROMA_420_16x8].filter_hps = PFX(interp_4tap_horiz_ps_16x8_avx2);<br>
-        p.chroma[X265_CSP_I420].pu[CHROMA_420_16x32].filter_hps = PFX(interp_4tap_horiz_ps_16x32_avx2);<br>
         p.chroma[X265_CSP_I420].pu[CHROMA_420_16x12].filter_hps = PFX(interp_4tap_horiz_ps_16x12_avx2);<br>
         p.chroma[X265_CSP_I420].pu[CHROMA_420_16x4].filter_hps = PFX(interp_4tap_horiz_ps_16x4_avx2);<br>
-        p.chroma[X265_CSP_I420].pu[CHROMA_420_32x32].filter_hps = PFX(interp_4tap_horiz_ps_32x32_avx2);<br>
-        p.chroma[X265_CSP_I420].pu[CHROMA_420_32x16].filter_hps = PFX(interp_4tap_horiz_ps_32x16_avx2);<br>
         p.chroma[X265_CSP_I420].pu[CHROMA_420_32x24].filter_hps = PFX(interp_4tap_horiz_ps_32x24_avx2);<br>
         p.chroma[X265_CSP_I420].pu[CHROMA_420_32x8].filter_hps = PFX(interp_4tap_horiz_ps_32x8_avx2);<br>
-        p.chroma[X265_CSP_I420].pu[CHROMA_420_24x32].filter_hps = PFX(interp_4tap_horiz_ps_24x32_avx2);<br>
-        p.chroma[X265_CSP_I420].pu[CHROMA_420_12x16].filter_hps = PFX(interp_4tap_horiz_ps_12x16_avx2);<br>
-        p.chroma[X265_CSP_I420].pu[CHROMA_420_6x8].filter_hps = PFX(interp_4tap_horiz_ps_6x8_avx2);<br>
         p.chroma[X265_CSP_I422].pu[CHROMA_422_8x8].filter_hps = PFX(interp_4tap_horiz_ps_8x8_avx2);<br>
         p.chroma[X265_CSP_I422].pu[CHROMA_422_8x16].filter_hps = PFX(interp_4tap_horiz_ps_8x16_avx2);<br>
         p.chroma[X265_CSP_I422].pu[CHROMA_422_8x32].filter_hps = PFX(interp_4tap_horiz_ps_8x32_avx2);<br>
         p.chroma[X265_CSP_I422].pu[CHROMA_422_8x12].filter_hps = PFX(interp_4tap_horiz_ps_8x12_avx2);<br>
-        p.chroma[X265_CSP_I422].pu[CHROMA_422_8x64].filter_hps = PFX(interp_4tap_horiz_ps_8x64_avx2);<br>
         p.chroma[X265_CSP_I422].pu[CHROMA_422_8x4].filter_hps = PFX(interp_4tap_horiz_ps_8x4_avx2);<br>
-        p.chroma[X265_CSP_I422].pu[CHROMA_422_16x32].filter_hps = PFX(interp_4tap_horiz_ps_16x32_avx2);<br>
-        p.chroma[X265_CSP_I422].pu[CHROMA_422_16x16].filter_hps = PFX(interp_4tap_horiz_ps_16x16_avx2);<br>
         p.chroma[X265_CSP_I422].pu[CHROMA_422_16x64].filter_hps = PFX(interp_4tap_horiz_ps_16x64_avx2);<br>
         p.chroma[X265_CSP_I422].pu[CHROMA_422_16x24].filter_hps = PFX(interp_4tap_horiz_ps_16x24_avx2);<br>
         p.chroma[X265_CSP_I422].pu[CHROMA_422_16x8].filter_hps = PFX(interp_4tap_horiz_ps_16x8_avx2);<br>
         p.chroma[X265_CSP_I422].pu[CHROMA_422_32x64].filter_hps = PFX(interp_4tap_horiz_ps_32x64_avx2);<br>
-        p.chroma[X265_CSP_I422].pu[CHROMA_422_32x32].filter_hps = PFX(interp_4tap_horiz_ps_32x32_avx2);<br>
-        p.chroma[X265_CSP_I422].pu[CHROMA_422_32x48].filter_hps = PFX(interp_4tap_horiz_ps_32x48_avx2);<br>
-        p.chroma[X265_CSP_I422].pu[CHROMA_422_32x16].filter_hps = PFX(interp_4tap_horiz_ps_32x16_avx2);<br>
-        p.chroma[X265_CSP_I422].pu[CHROMA_422_12x32].filter_hps = PFX(interp_4tap_horiz_ps_12x32_avx2);<br>
         p.chroma[X265_CSP_I422].pu[CHROMA_422_24x64].filter_hps = PFX(interp_4tap_horiz_ps_24x64_avx2);<br>
-        p.chroma[X265_CSP_I422].pu[CHROMA_422_6x16].filter_hps = PFX(interp_4tap_horiz_ps_6x16_avx2);<br>
         p.chroma[X265_CSP_I444].pu[LUMA_8x8].filter_hps = PFX(interp_4tap_horiz_ps_8x8_avx2);<br>
         p.chroma[X265_CSP_I444].pu[LUMA_8x4].filter_hps = PFX(interp_4tap_horiz_ps_8x4_avx2);<br>
         p.chroma[X265_CSP_I444].pu[LUMA_8x16].filter_hps = PFX(interp_4tap_horiz_ps_8x16_avx2);<br>
         p.chroma[X265_CSP_I444].pu[LUMA_8x32].filter_hps = PFX(interp_4tap_horiz_ps_8x32_avx2);<br>
-        p.chroma[X265_CSP_I444].pu[LUMA_16x16].filter_hps = PFX(interp_4tap_horiz_ps_16x16_avx2);<br>
         p.chroma[X265_CSP_I444].pu[LUMA_16x8].filter_hps = PFX(interp_4tap_horiz_ps_16x8_avx2);<br>
-        p.chroma[X265_CSP_I444].pu[LUMA_16x32].filter_hps = PFX(interp_4tap_horiz_ps_16x32_avx2);<br>
         p.chroma[X265_CSP_I444].pu[LUMA_16x12].filter_hps = PFX(interp_4tap_horiz_ps_16x12_avx2);<br>
         p.chroma[X265_CSP_I444].pu[LUMA_16x4].filter_hps = PFX(interp_4tap_horiz_ps_16x4_avx2);<br>
         p.chroma[X265_CSP_I444].pu[LUMA_16x64].filter_hps = PFX(interp_4tap_horiz_ps_16x64_avx2);<br>
-        p.chroma[X265_CSP_I444].pu[LUMA_32x32].filter_hps = PFX(interp_4tap_horiz_ps_32x32_avx2);<br>
-        p.chroma[X265_CSP_I444].pu[LUMA_32x16].filter_hps = PFX(interp_4tap_horiz_ps_32x16_avx2);<br>
         p.chroma[X265_CSP_I444].pu[LUMA_32x64].filter_hps = PFX(interp_4tap_horiz_ps_32x64_avx2);<br>
         p.chroma[X265_CSP_I444].pu[LUMA_32x24].filter_hps = PFX(interp_4tap_horiz_ps_32x24_avx2);<br>
         p.chroma[X265_CSP_I444].pu[LUMA_32x8].filter_hps = PFX(interp_4tap_horiz_ps_32x8_avx2);<br>
</div></div>@@ -1816,8 +1803,6 @@<br>
<span class="">         p.chroma[X265_CSP_I444].pu[LUMA_64x32].filter_hps = PFX(interp_4tap_horiz_ps_64x32_avx2);<br>
         p.chroma[X265_CSP_I444].pu[LUMA_64x16].filter_hps = PFX(interp_4tap_horiz_ps_64x16_avx2);<br>
         p.chroma[X265_CSP_I444].pu[LUMA_48x64].filter_hps = PFX(interp_4tap_horiz_ps_48x64_avx2);<br>
-        p.chroma[X265_CSP_I444].pu[LUMA_24x32].filter_hps = PFX(interp_4tap_horiz_ps_24x32_avx2);<br>
-        p.chroma[X265_CSP_I444].pu[LUMA_12x16].filter_hps = PFX(interp_4tap_horiz_ps_12x16_avx2);<br>
<br>
         p.chroma[X265_CSP_I420].pu[CHROMA_420_6x8].filter_hpp = PFX(interp_4tap_horiz_pp_6x8_avx2);<br>
         p.chroma[X265_CSP_I420].pu[CHROMA_420_8x2].filter_hpp = PFX(interp_4tap_horiz_pp_8x2_avx2);<br>
</span>@@ -1835,7 +1820,6 @@<br>
<span class="">         p.chroma[X265_CSP_I420].pu[CHROMA_420_32x16].filter_hpp = PFX(interp_4tap_horiz_pp_32x16_avx2);<br>
         p.chroma[X265_CSP_I420].pu[CHROMA_420_32x24].filter_hpp = PFX(interp_4tap_horiz_pp_32x24_avx2);<br>
         p.chroma[X265_CSP_I420].pu[CHROMA_420_32x32].filter_hpp = PFX(interp_4tap_horiz_pp_32x32_avx2);<br>
-        p.chroma[X265_CSP_I420].pu[CHROMA_420_12x16].filter_hpp = PFX(interp_4tap_horiz_pp_12x16_avx2);<br>
         p.chroma[X265_CSP_I420].pu[CHROMA_420_24x32].filter_hpp = PFX(interp_4tap_horiz_pp_24x32_avx2);<br>
         p.chroma[X265_CSP_I422].pu[CHROMA_422_6x16].filter_hpp = PFX(interp_4tap_horiz_pp_6x16_avx2);<br>
         p.chroma[X265_CSP_I422].pu[CHROMA_422_8x4].filter_hpp = PFX(interp_4tap_horiz_pp_8x4_avx2);<br>
</span>@@ -1853,7 +1837,6 @@<br>
<span class="">         p.chroma[X265_CSP_I422].pu[CHROMA_422_32x32].filter_hpp = PFX(interp_4tap_horiz_pp_32x32_avx2);<br>
         p.chroma[X265_CSP_I422].pu[CHROMA_422_32x48].filter_hpp = PFX(interp_4tap_horiz_pp_32x48_avx2);<br>
         p.chroma[X265_CSP_I422].pu[CHROMA_422_32x64].filter_hpp = PFX(interp_4tap_horiz_pp_32x64_avx2);<br>
-        p.chroma[X265_CSP_I422].pu[CHROMA_422_12x32].filter_hpp = PFX(interp_4tap_horiz_pp_12x32_avx2);<br>
         p.chroma[X265_CSP_I422].pu[CHROMA_422_24x64].filter_hpp = PFX(interp_4tap_horiz_pp_24x64_avx2);<br>
         p.chroma[X265_CSP_I444].pu[LUMA_8x4].filter_hpp = PFX(interp_4tap_horiz_pp_8x4_avx2);<br>
         p.chroma[X265_CSP_I444].pu[LUMA_8x8].filter_hpp = PFX(interp_4tap_horiz_pp_8x8_avx2);<br>
</span>@@ -1870,14 +1853,12 @@<br>
<span class="">         p.chroma[X265_CSP_I444].pu[LUMA_32x24].filter_hpp = PFX(interp_4tap_horiz_pp_32x24_avx2);<br>
         p.chroma[X265_CSP_I444].pu[LUMA_32x32].filter_hpp = PFX(interp_4tap_horiz_pp_32x32_avx2);<br>
         p.chroma[X265_CSP_I444].pu[LUMA_32x64].filter_hpp = PFX(interp_4tap_horiz_pp_32x64_avx2);<br>
-        p.chroma[X265_CSP_I444].pu[LUMA_12x16].filter_hpp = PFX(interp_4tap_horiz_pp_12x16_avx2);<br>
         p.chroma[X265_CSP_I444].pu[LUMA_24x32].filter_hpp = PFX(interp_4tap_horiz_pp_24x32_avx2);<br>
         p.chroma[X265_CSP_I444].pu[LUMA_64x16].filter_hpp = PFX(interp_4tap_horiz_pp_64x16_avx2);<br>
         p.chroma[X265_CSP_I444].pu[LUMA_64x32].filter_hpp = PFX(interp_4tap_horiz_pp_64x32_avx2);<br>
         p.chroma[X265_CSP_I444].pu[LUMA_64x48].filter_hpp = PFX(interp_4tap_horiz_pp_64x48_avx2);<br>
         p.chroma[X265_CSP_I444].pu[LUMA_64x64].filter_hpp = PFX(interp_4tap_horiz_pp_64x64_avx2);<br>
         p.chroma[X265_CSP_I444].pu[LUMA_48x64].filter_hpp = PFX(interp_4tap_horiz_pp_48x64_avx2);<br>
-<br>
         p.chroma[X265_CSP_I420].pu[CHROMA_420_4x2].filter_vpp = PFX(interp_4tap_vert_pp_4x2_avx2);<br>
         p.chroma[X265_CSP_I420].pu[CHROMA_420_4x2].filter_vps = PFX(interp_4tap_vert_ps_4x2_avx2);<br>
         p.chroma[X265_CSP_I420].pu[CHROMA_420_4x2].filter_vsp = PFX(interp_4tap_vert_sp_4x2_avx2);<br>
</span>@@ -1958,7 +1939,6 @@<br>
<span class="">         p.chroma[X265_CSP_I422].pu[CHROMA_422_8x64].filter_vpp = PFX(interp_4tap_vert_pp_8x64_avx2);<br>
         p.chroma[X265_CSP_I422].pu[CHROMA_422_8x64].filter_vps = PFX(interp_4tap_vert_ps_8x64_avx2);<br>
         p.chroma[X265_CSP_I422].pu[CHROMA_422_8x64].filter_vsp = PFX(interp_4tap_vert_sp_8x64_avx2);<br>
-        p.chroma[X265_CSP_I422].pu[CHROMA_422_8x64].filter_vss = PFX(interp_4tap_vert_ss_8x64_avx2);<br>
         p.chroma[X265_CSP_I444].pu[LUMA_4x4].filter_vpp = PFX(interp_4tap_vert_pp_4x4_avx2);<br>
         p.chroma[X265_CSP_I444].pu[LUMA_4x4].filter_vps = PFX(interp_4tap_vert_ps_4x4_avx2);<br>
         p.chroma[X265_CSP_I444].pu[LUMA_4x4].filter_vsp = PFX(interp_4tap_vert_sp_4x4_avx2);<br>
</span>@@ -1988,7 +1968,6 @@<br>
<span class="">         p.chroma[X265_CSP_I444].pu[LUMA_8x32].filter_vsp = PFX(interp_4tap_vert_sp_8x32_avx2);<br>
         p.chroma[X265_CSP_I444].pu[LUMA_8x32].filter_vss = PFX(interp_4tap_vert_ss_8x32_avx2);<br>
<br>
-<br>
         p.chroma[X265_CSP_I420].pu[CHROMA_420_6x8].filter_vss = PFX(interp_4tap_vert_ss_6x8_avx2);<br>
         p.chroma[X265_CSP_I420].pu[CHROMA_420_6x8].filter_vsp = PFX(interp_4tap_vert_sp_6x8_avx2);<br>
         p.chroma[X265_CSP_I420].pu[CHROMA_420_6x8].filter_vps = PFX(interp_4tap_vert_ps_6x8_avx2);<br>
</span>@@ -2037,10 +2016,6 @@<br>
<span class="">         p.chroma[X265_CSP_I420].pu[CHROMA_420_32x16].filter_vsp = PFX(interp_4tap_vert_sp_32x16_avx2);<br>
         p.chroma[X265_CSP_I420].pu[CHROMA_420_32x24].filter_vsp = PFX(interp_4tap_vert_sp_32x24_avx2);<br>
         p.chroma[X265_CSP_I420].pu[CHROMA_420_32x32].filter_vsp = PFX(interp_4tap_vert_sp_32x32_avx2);<br>
-        p.chroma[X265_CSP_I422].pu[CHROMA_422_6x16].filter_vss = PFX(interp_4tap_vert_ss_6x16_avx2);<br>
-        p.chroma[X265_CSP_I422].pu[CHROMA_422_6x16].filter_vsp = PFX(interp_4tap_vert_sp_6x16_avx2);<br>
-        p.chroma[X265_CSP_I422].pu[CHROMA_422_6x16].filter_vps = PFX(interp_4tap_vert_ps_6x16_avx2);<br>
-        p.chroma[X265_CSP_I422].pu[CHROMA_422_6x16].filter_vpp = PFX(interp_4tap_vert_pp_6x16_avx2);<br>
         p.chroma[X265_CSP_I422].pu[CHROMA_422_12x32].filter_vpp = PFX(interp_4tap_vert_pp_12x32_avx2);<br>
         p.chroma[X265_CSP_I422].pu[CHROMA_422_12x32].filter_vps = PFX(interp_4tap_vert_ps_12x32_avx2);<br>
         p.chroma[X265_CSP_I422].pu[CHROMA_422_12x32].filter_vss = PFX(interp_4tap_vert_ss_12x32_avx2);<br>
</span>@@ -2157,6 +2132,56 @@<br>
<span class="">         p.chroma[X265_CSP_I444].pu[LUMA_64x32].filter_vsp = PFX(interp_4tap_vert_sp_64x32_avx2);<br>
         p.chroma[X265_CSP_I444].pu[LUMA_64x48].filter_vsp = PFX(interp_4tap_vert_sp_64x48_avx2);<br>
         p.chroma[X265_CSP_I444].pu[LUMA_64x64].filter_vsp = PFX(interp_4tap_vert_sp_64x64_avx2);<br>
+<br>
</span>+        /* The following primitives have been disabled since performance compared to SSE is negligible/negative */<br>
<div><div class="h5">+#if 0<br>
+        p.chroma[X265_CSP_I422].pu[CHROMA_422_6x16].filter_vss = PFX(interp_4tap_vert_ss_6x16_avx2);<br>
+        p.chroma[X265_CSP_I422].pu[CHROMA_422_6x16].filter_vps = PFX(interp_4tap_vert_ps_6x16_avx2);<br>
+<br>
+        p.chroma[X265_CSP_I420].pu[CHROMA_420_12x16].filter_hpp = PFX(interp_4tap_horiz_pp_12x16_avx2);<br>
+        p.chroma[X265_CSP_I444].pu[LUMA_12x16].filter_hpp = PFX(interp_4tap_horiz_pp_12x16_avx2);<br>
+        p.chroma[X265_CSP_I422].pu[CHROMA_422_12x32].filter_hpp = PFX(interp_4tap_horiz_pp_12x32_avx2);<br>
+<br>
+        <a href="http://p.cu" rel="noreferrer" target="_blank">p.cu</a>[BLOCK_4x4].cpy1Dto2D_shl = PFX(cpy1Dto2D_shl_4_avx2);<br>
+        <a href="http://p.cu" rel="noreferrer" target="_blank">p.cu</a>[BLOCK_8x8].cpy1Dto2D_shl = PFX(cpy1Dto2D_shl_8_avx2);<br>
+        <a href="http://p.cu" rel="noreferrer" target="_blank">p.cu</a>[BLOCK_4x4].cpy1Dto2D_shr = PFX(cpy1Dto2D_shr_4_avx2);<br>
+        <a href="http://p.cu" rel="noreferrer" target="_blank">p.cu</a>[BLOCK_4x4].count_nonzero = PFX(count_nonzero_4x4_avx2);<br>
+        <a href="http://p.cu" rel="noreferrer" target="_blank">p.cu</a>[BLOCK_8x8].copy_cnt = PFX(copy_cnt_8_avx2);<br>
+<br>
+        p.chroma[X265_CSP_I444].pu[LUMA_12x16].filter_hps = PFX(interp_4tap_horiz_ps_12x16_avx2);<br>
+        p.chroma[X265_CSP_I420].pu[CHROMA_420_12x16].filter_hps = PFX(interp_4tap_horiz_ps_12x16_avx2);<br>
+        p.chroma[X265_CSP_I422].pu[CHROMA_422_12x32].filter_hps = PFX(interp_4tap_horiz_ps_12x32_avx2);<br>
+        p.chroma[X265_CSP_I422].pu[CHROMA_422_6x16].filter_hps = PFX(interp_4tap_horiz_ps_6x16_avx2);<br>
+        p.chroma[X265_CSP_I420].pu[CHROMA_420_6x8].filter_hps = PFX(interp_4tap_horiz_ps_6x8_avx2);<br>
+        p.chroma[X265_CSP_I444].pu[LUMA_16x32].filter_hps = PFX(interp_4tap_horiz_ps_16x32_avx2);<br>
+        p.chroma[X265_CSP_I420].pu[CHROMA_420_16x32].filter_hps = PFX(interp_4tap_horiz_ps_16x32_avx2);<br>
+        p.chroma[X265_CSP_I422].pu[CHROMA_422_16x32].filter_hps = PFX(interp_4tap_horiz_ps_16x32_avx2);<br>
+        p.chroma[X265_CSP_I420].pu[CHROMA_420_32x32].filter_hps = PFX(interp_4tap_horiz_ps_32x32_avx2);<br>
+        p.chroma[X265_CSP_I422].pu[CHROMA_422_32x32].filter_hps = PFX(interp_4tap_horiz_ps_32x32_avx2);<br>
+        p.chroma[X265_CSP_I444].pu[LUMA_32x32].filter_hps = PFX(interp_4tap_horiz_ps_32x32_avx2);<br>
+        p.chroma[X265_CSP_I420].pu[CHROMA_420_24x32].filter_hps = PFX(interp_4tap_horiz_ps_24x32_avx2);<br>
+        p.chroma[X265_CSP_I444].pu[LUMA_24x32].filter_hps = PFX(interp_4tap_horiz_ps_24x32_avx2);<br>
+        p.chroma[X265_CSP_I422].pu[CHROMA_422_32x48].filter_hps = PFX(interp_4tap_horiz_ps_32x48_avx2);<br>
+        p.chroma[X265_CSP_I422].pu[CHROMA_422_8x64].filter_hps = PFX(interp_4tap_horiz_ps_8x64_avx2);<br>
+        p.chroma[X265_CSP_I420].pu[CHROMA_420_32x16].filter_hps = PFX(interp_4tap_horiz_ps_32x16_avx2);<br>
+        p.chroma[X265_CSP_I422].pu[CHROMA_422_32x16].filter_hps = PFX(interp_4tap_horiz_ps_32x16_avx2);<br>
+        p.chroma[X265_CSP_I444].pu[LUMA_32x16].filter_hps = PFX(interp_4tap_horiz_ps_32x16_avx2);<br>
+        p.chroma[X265_CSP_I444].pu[LUMA_16x16].filter_hps = PFX(interp_4tap_horiz_ps_16x16_avx2);<br>
+        p.chroma[X265_CSP_I422].pu[CHROMA_422_16x16].filter_hps = PFX(interp_4tap_horiz_ps_16x16_avx2);<br>
+        p.chroma[X265_CSP_I420].pu[CHROMA_420_16x16].filter_hps = PFX(interp_4tap_horiz_ps_16x16_avx2);<br>
+        p.chroma[X265_CSP_I422].pu[CHROMA_422_8x64].filter_vss = PFX(interp_4tap_vert_ss_8x64_avx2);<br>
+        p.chroma[X265_CSP_I422].pu[CHROMA_422_6x16].filter_vsp = PFX(interp_4tap_vert_sp_6x16_avx2);<br>
+        p.chroma[X265_CSP_I422].pu[CHROMA_422_6x16].filter_vpp = PFX(interp_4tap_vert_pp_6x16_avx2);<br>
+<br>
+        p.pu[LUMA_8x4].addAvg   = PFX(addAvg_8x4_avx2);<br>
+        p.pu[LUMA_8x8].addAvg = PFX(addAvg_8x8_avx2);<br>
+        p.pu[LUMA_12x16].addAvg = PFX(addAvg_12x16_avx2);<br>
+        p.chroma[X265_CSP_I420].pu[CHROMA_420_8x4].addAvg = PFX(addAvg_8x4_avx2);<br>
+        p.chroma[X265_CSP_I420].pu[CHROMA_420_12x16].addAvg = PFX(addAvg_12x16_avx2);<br>
+        p.chroma[X265_CSP_I422].pu[CHROMA_422_8x8].addAvg = PFX(addAvg_8x8_avx2);<br>
+        p.chroma[X265_CSP_I422].pu[CHROMA_422_8x4].addAvg = PFX(addAvg_8x4_avx2);<br>
+        p.chroma[X265_CSP_I422].pu[CHROMA_422_12x32].addAvg = PFX(addAvg_12x32_avx2);<br>
+#endif<br>
 #endif<br>
<br>
         p.frameInitLowres = PFX(frame_init_lowres_core_avx2);<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" rel="noreferrer" target="_blank">https://mailman.videolan.org/listinfo/x265-devel</a><br>
</blockquote></div><br></div>