[MPlayer-dev-eng] [PATCH]: af_resample SIMD optimization(MMX/SSE2/3DNOW!/SSE/SSE3)

[Zuxy Meng]

2009/3/12 Andrew Savchenko <bircoph at gmail.com>:
> Hi,
>
> On Thursday 12 March 2009 10:09, Zhou Zongyi wrote:
>> I tested them on Athlon XP. SSE3 seems a bit faster than SSE.
>> Perhaps lddqu is faster than movups?
>
> Huh? Athlon XP doesn't support neither SSE3, nor SSE2, only SSE.
>
> # x86info
> [...]
> Family: 6 Model: 10 Stepping: 0
> CPU Model : Athlon XP (Barton)
> Feature flags:
>  fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat
> pse36 mmx fxsr sse
> Extended feature flags:
>  fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat
> pse36 mmxext mmx fxsr 3dnowext 3dnow
>
> And the following code fails with SIGILL on my Athlon XP:
> int main(void) {
>    __asm__ __volatile__ ("addsubpd %%xmm0, %%xmm1":::"memory");
>    return 0;
> }
>
> lddqu also fails with SIGILL.
> Am I missing something?

I bet he meant K8/Athlon 64.

>
> Anyway it is always a good idea to have SSE code as fallback for
> the case when SSE3 is not available.
>


-- 
Zuxy
Beauty is truth,
While truth is beauty.
PGP KeyID: E8555ED6