2011/10/20 Arun Raghavan <arun.raghavan at collabora.co.uk>: > On Tue, 2011-10-18 at 17:09 -0400, Wang Xingchao wrote: >> if all channels have same volume setting, use fast way to >> do volume change. this patch intended to work for two formats: >> s16ne/s16re. >> >> Signed-off-by: Wang Xingchao <xingchao.wang at intel.com> >> --- > > As Tanu points out, some hard data here would be good. But just to set > your expectations correctly, this code is actually not used on any > recent x86 or ARM machine since there are optimised routines for these. > I'm certainly not against optimisations to these paths, since > there /are/ other architectures that PA runs on. That's only for s16. And actually that exact same optimisation (treating the stream as mono if all channel volumes are equal) could also work for the assembly functions, just as I had already added it to the Orc functions in my orcify branch. Maarten
