I wanna add my feedback about perfomance. Someone (sorry I don't remember) wrote me that Mixxx should take about 15% of CPU. Yes, my Mixxx does! Only 15-16%. But it is really slow and lagy now. When I shade it's window Mixxx produces less clicks. But still unable to work with it. Latency is set to 100 now. Mixxx clicks permanently - for example, when I switch to another window or when some hint apears and so on. I'll try to use OSS as you advice. And will send results here.
On Monday 07 January 2008 2:55 am, Albert Santoni wrote: > Hello, > > On Sun, 2008-01-06 at 22:30 +0000, Notorious wrote: > > Hi > > > > Configuration(not mentioned before): > > Xorg 7.3 with Kde 3.5.8 > > Nvidia TI 4200 using 96.43.01 > > Gcc 4.2.3 > > > > Some things I forgot to mention: > > When mixxx gets killed, It loses it's Sound Hardware settings. > > Settings only get saved on (clean) exit. This is something we might look > into changing. > > > Full screen mode doesn't increase size of the interface itself. > > This is something we're also slowly trying to address. Basically, we > know it sucks, but there's no easy solution to it. It's something we > want to fix though, and I've talked to DJ Synthe about several possible > solutions. If we're lucky, maybe we can fix this over the summer. > > > What about the ability to use the scratch wheel to scroll through the > > song while playing? > > If you feel like writing some code, you could hack it in... ? The > scratch wheel is supposed to perform a sort of temporary pitch bend > while the song is playing, which is quite helpful when you're > beatmatching. We try to offer to keep the controls consistent across > hardware controllers as well. > > > Response: > > > > Dj Control MP3 Leds: > > Installed svn version, no changes > > In the djconsole legacy code I see that nearly everything concerning the > > leds is commented out, libdjconsole is not compatible with my Dj control > > mp3. > > > > Tying to patch the 2.6.23.8 kernel with the hid patch generates errors. > > > > Performance: > > The hifi-eq is disabled > > > > As Wesley said, disabling the rendering of the waveform drastically > > reduces the cpu usage although sudden high spikes resulting into "clicks" > > & lagged response remain. > > I must stress the fact that other programs don't show this behavior. > > Could there be something with the threading inside of mixxx? > > > > I can't wrap my head around the fact that this little bit of rendering > > could cause 30+ % more cpu usage. > > According to the Nvidia tool, direct rendering is enabled. > > Thanks again for the useful feedback. Does your CPU usage really drop > 30% when you disable the waveform rendering? Mine drops, but I don't > think it's that drastic. As for the threading issue, yes, there could be > something wrong inside Mixxx. There's one issue with PortAudio's ALSA > code (the lack of requesting a realtime thread) that may be hurting ALSA > performance, so try selecting OSS as an API. (I can get much better > latency with OSS compared to ALSA in Mixxx.) > > Again, the CPU usage is somewhat justified, but unintentional. What > you're seeing sounds like a combination of a couple of problems, and > hopefully we can sort them out. > > Thanks, > Albert > > > > ------------------------------------------------------------------------- > This SF.net email is sponsored by: Microsoft > Defy all challenges. Microsoft(R) Visual Studio 2005. > http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ > _______________________________________________ > Mixxx-devel mailing list > [email protected] > https://lists.sourceforge.net/lists/listinfo/mixxx-devel ------------------------------------------------------------------------- This SF.net email is sponsored by: Microsoft Defy all challenges. Microsoft(R) Visual Studio 2005. http://clk.atdmt.com/MRT/go/vse0120000070mrt/direct/01/ _______________________________________________ Mixxx-devel mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/mixxx-devel
