[x265] [PATCH] asm: luma_vpp[64x16] in avx2: improve 7245c->4910c

[Divya Manivannan]

# HG changeset patch
# User Divya Manivannan <divya at multicorewareinc.com>
# Date 1416980884 -19800
#      Wed Nov 26 11:18:04 2014 +0530
# Node ID baad907413bd3b711fb84090633df5b844701122
# Parent  879b986bbc706ecaf5eb93db5b5a381d9cea47e0
asm: luma_vpp[64x16] in avx2: improve 7245c->4910c

diff -r 879b986bbc70 -r baad907413bd source/common/x86/asm-primitives.cpp
--- a/source/common/x86/asm-primitives.cpp	Wed Nov 26 11:02:55 2014 +0530
+++ b/source/common/x86/asm-primitives.cpp	Wed Nov 26 11:18:04 2014 +0530
@@ -1812,6 +1812,7 @@
         p.luma_vpp[LUMA_32x32] = x265_interp_8tap_vert_pp_32x32_avx2;
         p.luma_vpp[LUMA_32x64] = x265_interp_8tap_vert_pp_32x64_avx2;
 
+        p.luma_vpp[LUMA_64x16] = x265_interp_8tap_vert_pp_64x16_avx2;
         p.luma_vpp[LUMA_64x32] = x265_interp_8tap_vert_pp_64x32_avx2;
         p.luma_vpp[LUMA_64x48] = x265_interp_8tap_vert_pp_64x48_avx2;
         p.luma_vpp[LUMA_64x64] = x265_interp_8tap_vert_pp_64x64_avx2;
diff -r 879b986bbc70 -r baad907413bd source/common/x86/ipfilter8.asm
--- a/source/common/x86/ipfilter8.asm	Wed Nov 26 11:02:55 2014 +0530
+++ b/source/common/x86/ipfilter8.asm	Wed Nov 26 11:18:04 2014 +0530
@@ -5846,6 +5846,33 @@
 FILTER_VER_LUMA_AVX2_64xN 64, 48
 FILTER_VER_LUMA_AVX2_64xN 64, 64
 
+INIT_YMM avx2
+%if ARCH_X86_64 == 1
+cglobal interp_8tap_vert_pp_64x16, 4, 10, 15
+    mov             r4d, r4m
+    shl             r4d, 7
+
+%ifdef PIC
+    lea             r5, [tab_LumaCoeffVer_32]
+    add             r5, r4
+%else
+    lea             r5, [tab_LumaCoeffVer_32 + r4]
+%endif
+
+    lea             r4, [r1 * 3]
+    sub             r0, r4
+    lea             r6, [r3 * 3]
+    mova            m14, [pw_512]
+    mov             r9d, 4
+.loopW:
+    PROCESS_LUMA_AVX2_W16_16R
+    add             r2, 16
+    add             r0, 16
+    dec             r9d
+    jnz             .loopW
+    RET
+%endif
+
 ;-------------------------------------------------------------------------------------------------------------
 ; void interp_8tap_vert_%3_%1x%2(pixel *src, intptr_t srcStride, pixel *dst, intptr_t dstStride, int coeffIdx)
 ;-------------------------------------------------------------------------------------------------------------