Without this, code like "shufps m0, m1, m2, 0xaa" would not work in
CPUs not supporting SSE2.

This is needed for my IMDCT36() patch (committed in ffmpeg, WIP for libav).

-Vitor
From 711295e4e3a82eb47c20b87d391220d533c154e5 Mon Sep 17 00:00:00 2001
From: Vitor Sessak <[email protected]>
Date: Sun, 27 Nov 2011 15:41:56 +0100
Subject: [PATCH] x86inc: Flag shufps as an floating-point instruction for the AVX emulation code.

Without this, code like "shufps m0, m1, m2, 0xaa" would not work in CPUs
not supporting SSE2.
---
 libavutil/x86/x86inc.asm |    2 +-
 1 files changed, 1 insertions(+), 1 deletions(-)

diff --git a/libavutil/x86/x86inc.asm b/libavutil/x86/x86inc.asm
index 31f74b9..475e70e 100644
--- a/libavutil/x86/x86inc.asm
+++ b/libavutil/x86/x86inc.asm
@@ -1016,7 +1016,7 @@ AVX_INSTR punpcklwd, 0, 0, 0
 AVX_INSTR punpckldq, 0, 0, 0
 AVX_INSTR punpcklqdq, 0, 0, 0
 AVX_INSTR pxor, 0, 0, 1
-AVX_INSTR shufps, 0, 1, 0
+AVX_INSTR shufps, 1, 1, 0
 AVX_INSTR subpd, 1, 0, 0
 AVX_INSTR subps, 1, 0, 0
 AVX_INSTR subsd, 1, 0, 0
-- 
1.7.0.4

_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel

Reply via email to