On Tue, Jul 19, 2011 at 10:37:47PM -0400, Sean Collins wrote: > > I am measuring latency's of 108-300+msec on my machine, which uses hd > audio. Ouch! (I thought you had at one time said you were getting low latency?) But maybe that explains why Axel is seeing it now. Maybe HDA's buffer size got increased recently, so the same effects are seen there. > .Are you on a single or multi core machine ? I am on a pretty fast > multi core machine and while I don't get glitchy sound, I do see very > long buffers sizes and long latency's, latency's which are reported much > lower on my windows install with the same hardware. Mine's a single-CPU box, but it's a lot faster than my old BeOS twin-CPU "Ming Special", which did most things quite fast enough, so I'm not too worried. (:-)) It does seem that the glitches are masked by large buffers. I think they can still occur rarely, and latency increases, but the buffers are longer than the blackouts, so they cover them quite nicely. > > I know it'd be nice to see Haiku fulfill its role as the BeOS > replacement with the BeOS's amazing latency times which were borderline > bare metal. Actually when I have my fixes in place, and am careful not to move windows, I get better latencies than on my Ming Special (<20ms vs. >30ms). And on my own apps, I've started to set the mode to B_RECORDING; I still get the glitches of course, but latency stays low. I'd like to be able to conveniently use SqueekySynth, but the only way to set the mode of that is to catch it with Cortex before things go haywire and set it myself. Maybe Cyan will resurface at some point, and update it. > > If there anything I can help with please let me know. I will > test/report . As it sits this latency issue on audio is going to be very > problematic for anyone looking to develop media software especially if > its real time "harhar" type software. Thanks. Yes, I really want to get this fixed, as -- aside from general utility -- real-time audio is my main reason for wanting Haiku. -- Pete --