[x265] [PATCH] asm: luma_vpp[32x16] in avx2: improve 3808c->2491c

chen chenm003 at 163.com
Tue Nov 25 20:30:01 CET 2014


right

At 2014-11-25 14:04:54,"Divya Manivannan" <divya at multicorewareinc.com> wrote:
># HG changeset patch
># User Divya Manivannan <divya at multicorewareinc.com>
># Date 1416895454 -19800
>#      Tue Nov 25 11:34:14 2014 +0530
># Node ID d62fabc8f48fdb9992b828c55caca9738f8a0de5
># Parent  46ece877c99f866a08cb29a7d9d1a9d5bf1dc7d4
>asm: luma_vpp[32x16] in avx2: improve 3808c->2491c
>
>diff -r 46ece877c99f -r d62fabc8f48f source/common/x86/asm-primitives.cpp
>--- a/source/common/x86/asm-primitives.cpp	Tue Nov 25 11:10:44 2014 +0530
>+++ b/source/common/x86/asm-primitives.cpp	Tue Nov 25 11:34:14 2014 +0530
>@@ -1804,6 +1804,7 @@
>         p.luma_vpp[LUMA_16x16] = x265_interp_8tap_vert_pp_16x16_avx2;
>         p.luma_vpp[LUMA_16x32] = x265_interp_8tap_vert_pp_16x32_avx2;
>         p.luma_vpp[LUMA_16x64] = x265_interp_8tap_vert_pp_16x64_avx2;
>+        p.luma_vpp[LUMA_32x16] = x265_interp_8tap_vert_pp_32x16_avx2;
>         p.luma_vpp[LUMA_32x32] = x265_interp_8tap_vert_pp_32x32_avx2;
>         p.luma_vpp[LUMA_32x64] = x265_interp_8tap_vert_pp_32x64_avx2;
> #endif
>diff -r 46ece877c99f -r d62fabc8f48f source/common/x86/ipfilter8.asm
>--- a/source/common/x86/ipfilter8.asm	Tue Nov 25 11:10:44 2014 +0530
>+++ b/source/common/x86/ipfilter8.asm	Tue Nov 25 11:34:14 2014 +0530
>@@ -5369,6 +5369,33 @@
> FILTER_VER_LUMA_AVX2_32xN 32, 32
> FILTER_VER_LUMA_AVX2_32xN 32, 64
> 
>+INIT_YMM avx2
>+%if ARCH_X86_64 == 1
>+cglobal interp_8tap_vert_pp_32x16, 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, 2
>+.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)
> ;-------------------------------------------------------------------------------------------------------------
>_______________________________________________
>x265-devel mailing list
>x265-devel at videolan.org
>https://mailman.videolan.org/listinfo/x265-devel
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.videolan.org/pipermail/x265-devel/attachments/20141126/a0176ebb/attachment.html>


More information about the x265-devel mailing list