hi, I got not stable results, which means the optimization is not as good as expected. All results based on plain c volume function, not touch sse part. All tests are for format s16ne. Result1: channels 4, with same volume, with patch; samples 1021, loop 1000; Result2: channels 4, with same volume, without patch; samples 1021, loop 1000; Result1: channels 4, with different volumes; samples 1021, loop 1000; First try: I: svolume_sse.c: ##########New: s16ne test result########## I: svolume_sse.c: ref: 4292 usec. --> as good as expect I: svolume_sse.c: ##########Before patch########## I: svolume_sse.c: ref: 14568 usec. I: svolume_sse.c: ##########OLD: s16ne test result########## I: svolume_sse.c: ref: 12069 usec. Second try I: svolume_sse.c: ##########New: s16ne test result########## I: svolume_sse.c: ref: 7671 usec. --> not stable result I: svolume_sse.c: ##########Before patch########## I: svolume_sse.c: ref: 6674 usec. I: svolume_sse.c: ##########OLD: s16ne test result########## I: svolume_sse.c: ref: 7580 usec. Attached the raw patch for the test. As Maarten had done some work, i decide to check them first. --xingchao 2011/10/18 Wang Xingchao <wangxingchao2011 at gmail.com>: > 2011/10/18 Tanu Kaskinen <tanu.kaskinen at digia.com>: >> On Tue, 2011-10-18 at 14:49 +0300, Wang Xingchao wrote: >>> Tanu, i've not done much performance test, do you have some tools for >>> svolume test? All suggestions are appreciated. >> >> No, I don't know any particular tools. I was asking just because I don't >> think the patches can be merged before there is some proof that they are >> actually helpful (I believe that it's not uncommon for optimization >> efforts to not actually help as much as expected). >> >> -- >> Tanu >> >> > yeah, that's true, thanks. > > --xingchao > -------------- next part -------------- A non-text attachment was scrubbed... Name: softvolume-optimization.patch Type: text/x-patch Size: 7240 bytes Desc: not available URL: <http://lists.freedesktop.org/archives/pulseaudio-discuss/attachments/20111019/2d0cb18f/attachment.bin>
