Re: [libav-devel] [PATCH] dsputil: SIMD for bswap_buf

2012-01-29 Thread Diego Biurrun
On Sat, Jan 28, 2012 at 11:36:12AM +0100, Christophe Gisquet wrote: > 2012/1/25 Diego Biurrun : > > Never mind, I looked at the subject line of the mail, but you > > attached the patch and did not send git-send-email... > > Ping? Queued. Diego ___ liba

Re: [libav-devel] [PATCH] dsputil: SIMD for bswap_buf

2012-01-28 Thread Christophe Gisquet
Hi, 2012/1/25 Diego Biurrun : > Never mind, I looked at the subject line of the mail, but you > attached the patch and did not send git-send-email... Ping? ___ libav-devel mailing list libav-devel@libav.org https://lists.libav.org/mailman/listinfo/libav

Re: [libav-devel] [PATCH] dsputil: SIMD for bswap_buf

2012-01-25 Thread Diego Biurrun
On Wed, Jan 25, 2012 at 06:43:22PM +0100, Diego Biurrun wrote: > On Wed, Jan 25, 2012 at 06:40:17PM +0100, Christophe Gisquet wrote: > > 2012/1/25 Loren Merritt : > > >>+ mov %+ %1  m0, [r1 +  0] > > > > > > mov%1 > > > > Indeed. > > > > Also changed a bit the patch title. > > It would be nice t

Re: [libav-devel] [PATCH] dsputil: SIMD for bswap_buf

2012-01-25 Thread Diego Biurrun
On Wed, Jan 25, 2012 at 06:40:17PM +0100, Christophe Gisquet wrote: > 2012/1/25 Loren Merritt : > >>+ mov %+ %1  m0, [r1 +  0] > > > > mov%1 > > Indeed. > > Also changed a bit the patch title. It would be nice to have "x86" in the patch title. Diego _

Re: [libav-devel] [PATCH] dsputil: SIMD for bswap_buf

2012-01-25 Thread Christophe Gisquet
Hi, 2012/1/25 Loren Merritt : >>+ mov %+ %1  m0, [r1 +  0] > > mov%1 Indeed. Also changed a bit the patch title. Best regards, Christophe From c8ce1bf47248ea2cef8d762d58c7a20693470af4 Mon Sep 17 00:00:00 2001 From: Christophe GISQUET Date: Thu, 19 Jan 2012 21:48:39 +0100 Subject: [PATCH 4/4] x

Re: [libav-devel] [PATCH] dsputil: SIMD for bswap_buf

2012-01-25 Thread Loren Merritt
On Sat, 21 Jan 2012, Christophe Gisquet wrote: >+ mov %+ %1 m0, [r1 + 0] mov%1 --Loren Merritt ___ libav-devel mailing list libav-devel@libav.org https://lists.libav.org/mailman/listinfo/libav-devel

Re: [libav-devel] [PATCH] dsputil: SIMD for bswap_buf

2012-01-25 Thread Christophe Gisquet
2012/1/21 Christophe Gisquet : > Anyway, here's an updated patch taking into account Loren's comments. Ping? ___ libav-devel mailing list libav-devel@libav.org https://lists.libav.org/mailman/listinfo/libav-devel

Re: [libav-devel] [PATCH] dsputil: SIMD for bswap_buf

2012-01-21 Thread Christophe Gisquet
Hi, 2012/1/19 Jason Garrett-Glaser : > Is the SSE2 version faster than using 64-bit bswap on x86_64?  If not, > it shouldn't be used on 64-bit. If you mean something like the attached patch (MD5 validated), then it's almost a tie: on a longer sequence to reduce error measures and a 64bits system,

Re: [libav-devel] [PATCH] dsputil: SIMD for bswap_buf

2012-01-19 Thread Loren Merritt
On Thu, 19 Jan 2012, Christophe Gisquet wrote: > +%if %1 == 0 > +%define MOVR mova > +%else > +%define MOVR movu > +%endif If you passed u/a instead of 1/0, you wouldn't need to define MOVR. --Loren Merritt ___ libav-devel mailing list libav-devel@li

Re: [libav-devel] [PATCH] dsputil: SIMD for bswap_buf

2012-01-19 Thread Jason Garrett-Glaser
On Thu, Jan 19, 2012 at 1:05 PM, Christophe Gisquet wrote: > Hi, > > Note: > - I have tried SSE4 (and pextrd for the remaining cases) with a > testcase on fewer elements (eg mpc7), but that was not faster > - Unrolling SSSE3 only up to 4 is slower > - Unrolling up to 16 doesn't change things > > I

[libav-devel] [PATCH] dsputil: SIMD for bswap_buf

2012-01-19 Thread Christophe Gisquet
Hi, Note: - I have tried SSE4 (and pextrd for the remaining cases) with a testcase on fewer elements (eg mpc7), but that was not faster - Unrolling SSSE3 only up to 4 is slower - Unrolling up to 16 doesn't change things I'm not sure it's worth writing a MMX2 version. Maybe building several versio