[x265] [PATCH] assembly code for intra_pred_ang8_4
chen
chenm003 at 163.com
Tue Jan 21 07:51:29 CET 2014
code is right
At 2014-01-21 14:36:15,yuvaraj at multicorewareinc.com wrote:
># HG changeset patch
># User Yuvaraj Venkatesh <yuvaraj at multicorewareinc.com>
># Date 1390286104 -19800
># Tue Jan 21 12:05:04 2014 +0530
># Node ID 051c3938d6e797d75c06792ccbf8fb2326576b92
># Parent ce41ee0f5c8c6bf0a8b5ab268c172c7d6194f1d7
>assembly code for intra_pred_ang8_4.
>
>diff -r ce41ee0f5c8c -r 051c3938d6e7 source/common/x86/asm-primitives.cpp
>--- a/source/common/x86/asm-primitives.cpp Tue Jan 21 10:11:44 2014 +0530
>+++ b/source/common/x86/asm-primitives.cpp Tue Jan 21 12:05:04 2014 +0530
>@@ -1013,11 +1013,13 @@
> SETUP_INTRA_ANG4(32, 4, sse4);
> SETUP_INTRA_ANG4(33, 3, sse4);
>
>+ SETUP_INTRA_ANG8(3, 3, sse4);
>+ SETUP_INTRA_ANG8(4, 4, sse4);
>+ SETUP_INTRA_ANG8(32, 4, sse4);
>+ SETUP_INTRA_ANG8(33, 3, sse4);
>+
> SETUP_INTRA_ANG32(17, 17, sse4);
>
>- SETUP_INTRA_ANG8(3, 3, sse4);
>- SETUP_INTRA_ANG8(33, 3, sse4);
>-
> p.dct[DCT_8x8] = x265_dct8_sse4;
> }
> if (cpuMask & X265_CPU_AVX)
>diff -r ce41ee0f5c8c -r 051c3938d6e7 source/common/x86/intrapred8.asm
>--- a/source/common/x86/intrapred8.asm Tue Jan 21 10:11:44 2014 +0530
>+++ b/source/common/x86/intrapred8.asm Tue Jan 21 12:05:04 2014 +0530
>@@ -1196,6 +1196,7 @@
> pmulhrsw m2, m7
> packuswb m1, m2
>
>+.transpose8x8:
> jz .store
>
> ; transpose 8x8
>@@ -1229,6 +1230,65 @@
>
> RET
>
>+cglobal intra_pred_ang8_4, 3,5,8
>+ cmp r4m, byte 32
>+ cmove r2, r3mp
>+ lea r3, [ang_table + 19 * 16]
>+ mova m7, [pw_1024]
>+
>+ movu m0, [r2 + 1] ; [16 15 14 13 12 11 10 9 8 7 6 5 4 3 2 1]
>+ palignr m1, m0, 1 ; [x 16 15 14 13 12 11 10 9 8 7 6 5 4 3 2]
>+
>+ punpckhbw m2, m0, m1 ; [x 16 16 15 15 14 14 13 13 12 12 11 11 10 10 9]
>+ punpcklbw m0, m1 ; [9 8 8 7 7 6 6 5 5 4 4 3 3 2 2 1]
>+ palignr m1, m2, m0, 2 ; [10 9 9 8 8 7 7 6 6 5 5 4 4 3 3 2]
>+ mova m5, m1
>+
>+ movu m3, [r3 + 2 * 16] ; [21]
>+ movu m6, [r3 - 9 * 16] ; [10]
>+
>+ pmaddubsw m4, m0, m3
>+ pmulhrsw m4, m7
>+ pmaddubsw m1, m6
>+ pmulhrsw m1, m7
>+ packuswb m4, m1
>+
>+ movu m3, [r3 + 12 * 16] ; [31]
>+ pmaddubsw m5, m3
>+ pmulhrsw m5, m7
>+
>+ palignr m6, m2, m0, 4 ; [11 10 10 9 9 8 8 7 7 6 6 5 5 4 4 3]
>+
>+ movu m3, [r3 + 1 * 16] ; [ 20]
>+ pmaddubsw m6, m3
>+ pmulhrsw m6, m7
>+ packuswb m5, m6
>+
>+ palignr m1, m2, m0, 6 ; [12 11 11 10 10 9 9 8 8 7 7 6 6 5 5 4]
>+
>+ movu m3, [r3 - 10 * 16] ; [ 9]
>+ pmaddubsw m6, m1, m3
>+ pmulhrsw m6, m7
>+
>+ movu m3, [r3 + 11 * 16] ; [30]
>+ pmaddubsw m1, m3
>+ pmulhrsw m1, m7
>+ packuswb m6, m1
>+
>+ palignr m1, m2, m0, 8 ; [13 12 12 11 11 10 10 9 9 8 8 7 7 6 6 5]
>+
>+ movu m3, [r3] ; [19]
>+ pmaddubsw m1, m3
>+ pmulhrsw m1, m7
>+
>+ palignr m2, m0, 10 ; [14 13 13 12 12 11 11 10 10 9 9 8 8 7 7 8]
>+
>+ movu m3, [r3 - 11 * 16] ; [8]
>+ pmaddubsw m2, m3
>+ pmulhrsw m2, m7
>+ packuswb m1, m2
>+ jmp mangle(private_prefix %+ _ %+ intra_pred_ang8_3 %+ SUFFIX %+ .transpose8x8)
>+
> ;-----------------------------------------------------------------------------
> ; void intraPredAng16(pixel* dst, intptr_t dstStride, pixel *refLeft, pixel *refAbove, int dirMode, int bFilter)
> ;-----------------------------------------------------------------------------
>_______________________________________________
>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/20140121/7833fe81/attachment-0001.html>
More information about the x265-devel
mailing list