[x265] [PATCH 222 of 307] x86: Fix merge errors

mythreyi at multicorewareinc.com mythreyi at multicorewareinc.com
Sat Apr 7 04:33:40 CEST 2018


# HG changeset patch
# User Vignesh Vijayakumar<vignesh at multicorewareinc.com>
# Date 1512037629 -19800
#      Thu Nov 30 15:57:09 2017 +0530
# Node ID adbbeb6c2f1e3697515561179abe156bacc7cd1c
# Parent  834a8f52a976a6c5da294267392bcd6da1aa6d6e
x86: Fix merge errors

diff -r 834a8f52a976 -r adbbeb6c2f1e source/common/x86/asm-primitives.cpp
--- a/source/common/x86/asm-primitives.cpp	Wed Nov 29 16:28:04 2017 +0530
+++ b/source/common/x86/asm-primitives.cpp	Thu Nov 30 15:57:09 2017 +0530
@@ -2881,15 +2881,7 @@
         p.pu[LUMA_64x48].luma_vsp = PFX(interp_8tap_vert_sp_64x48_avx512);
         p.pu[LUMA_64x64].luma_vsp = PFX(interp_8tap_vert_sp_64x64_avx512);
         p.pu[LUMA_48x64].luma_vsp = PFX(interp_8tap_vert_sp_48x64_avx512);
-        p.pu[LUMA_32x64].luma_vsp = PFX(interp_8tap_vert_sp_32x64_avx512);
-        p.pu[LUMA_32x32].luma_vsp = PFX(interp_8tap_vert_sp_32x32_avx512);
-        p.pu[LUMA_32x24].luma_vsp = PFX(interp_8tap_vert_sp_32x24_avx512);
-        p.pu[LUMA_32x16].luma_vsp = PFX(interp_8tap_vert_sp_32x16_avx512);
-        p.pu[LUMA_32x8].luma_vsp = PFX(interp_8tap_vert_sp_32x8_avx512);
-        p.pu[LUMA_64x64].luma_vsp = PFX(interp_8tap_vert_sp_64x64_avx512);
-        p.pu[LUMA_64x48].luma_vsp = PFX(interp_8tap_vert_sp_64x48_avx512);
-        p.pu[LUMA_64x32].luma_vsp = PFX(interp_8tap_vert_sp_64x32_avx512);
-        p.pu[LUMA_64x16].luma_vsp = PFX(interp_8tap_vert_sp_64x16_avx512);
+
         p.cu[BLOCK_8x8].dct = PFX(dct8_avx512);
         p.cu[BLOCK_8x8].idct = PFX(idct8_avx512);
         p.cu[BLOCK_16x16].idct = PFX(idct16_avx512);
@@ -5010,8 +5002,18 @@
         p.pu[LUMA_64x48].luma_vpp = PFX(interp_8tap_vert_pp_64x48_avx512);
         p.pu[LUMA_64x32].luma_vpp = PFX(interp_8tap_vert_pp_64x32_avx512);
         p.pu[LUMA_64x16].luma_vpp = PFX(interp_8tap_vert_pp_64x16_avx512);
+
+        p.pu[LUMA_32x64].luma_vsp = PFX(interp_8tap_vert_sp_32x64_avx512);
+        p.pu[LUMA_32x32].luma_vsp = PFX(interp_8tap_vert_sp_32x32_avx512);
+        p.pu[LUMA_32x24].luma_vsp = PFX(interp_8tap_vert_sp_32x24_avx512);
+        p.pu[LUMA_32x16].luma_vsp = PFX(interp_8tap_vert_sp_32x16_avx512);
+        p.pu[LUMA_32x8].luma_vsp = PFX(interp_8tap_vert_sp_32x8_avx512);
+        p.pu[LUMA_64x64].luma_vsp = PFX(interp_8tap_vert_sp_64x64_avx512);
+        p.pu[LUMA_64x48].luma_vsp = PFX(interp_8tap_vert_sp_64x48_avx512);
+        p.pu[LUMA_64x32].luma_vsp = PFX(interp_8tap_vert_sp_64x32_avx512);
+        p.pu[LUMA_64x16].luma_vsp = PFX(interp_8tap_vert_sp_64x16_avx512);
+
         p.cu[BLOCK_8x8].dct    = PFX(dct8_avx512);
-
         /* TODO: Currently these kernels performance are similar to AVX2 version, we need a to improve them further to ebable
          * it. Probably a Vtune analysis will help here.
 


More information about the x265-devel mailing list