diff options
| author | Lynne <dev@lynne.ee> | 2023-02-01 02:26:20 +0100 |
|---|---|---|
| committer | Lynne <dev@lynne.ee> | 2023-02-01 04:23:55 +0100 |
| commit | bbe95f7353a972f28a48be8da883549f02c59e4b (patch) | |
| tree | 08841c9da55e7f076f6046d1dbd70f49d74c0ec0 /libavcodec/x86/aacpsdsp.asm | |
| parent | fc9a3b584da3cf3fc1f00036be2eaf5dff903ccf (diff) | |
| download | ffmpeg-bbe95f7353a972f28a48be8da883549f02c59e4b.tar.gz | |
x86: replace explicit REP_RETs with RETs
From x86inc:
> On AMD cpus <=K10, an ordinary ret is slow if it immediately follows either
> a branch or a branch target. So switch to a 2-byte form of ret in that case.
> We can automatically detect "follows a branch", but not a branch target.
> (SSSE3 is a sufficient condition to know that your cpu doesn't have this problem.)
x86inc can automatically determine whether to use REP_RET rather than
REP in most of these cases, so impact is minimal. Additionally, a few
REP_RETs were used unnecessary, despite the return being nowhere near a
branch.
The only CPUs affected were AMD K10s, made between 2007 and 2011, 16
years ago and 12 years ago, respectively.
In the future, everyone involved with x86inc should consider dropping
REP_RETs altogether.
Diffstat (limited to 'libavcodec/x86/aacpsdsp.asm')
| -rw-r--r-- | libavcodec/x86/aacpsdsp.asm | 10 |
1 files changed, 5 insertions, 5 deletions
diff --git a/libavcodec/x86/aacpsdsp.asm b/libavcodec/x86/aacpsdsp.asm index 105e1af5c5..cc496d4df8 100644 --- a/libavcodec/x86/aacpsdsp.asm +++ b/libavcodec/x86/aacpsdsp.asm @@ -49,7 +49,7 @@ align 16 add dstq, mmsize add nq, mmsize*2 jl .loop - REP_RET + RET %endmacro INIT_XMM sse @@ -83,7 +83,7 @@ align 16 add src2q, mmsize add nq, mmsize*2 jl .loop - REP_RET + RET ;*********************************************************************** ;void ff_ps_stereo_interpolate_sse3(float (*l)[2], float (*r)[2], @@ -116,7 +116,7 @@ align 16 movhps [rq+nq], m2 add nq, 8 jl .loop - REP_RET + RET ;*************************************************************************** ;void ps_stereo_interpolate_ipdopd_sse3(float (*l)[2], float (*r)[2], @@ -164,7 +164,7 @@ align 16 movhps [rq+nq], m2 add nq, 8 jl .loop - REP_RET + RET ;********************************************************** ;void ps_hybrid_analysis_ileave_sse(float out[2][38][64], @@ -484,7 +484,7 @@ align 16 add outq, strideq add nq, 64 jl .loop - REP_RET + RET %endmacro INIT_XMM sse |