@Shonumi: now that the PA backend is officially the best audio backend for Dolphin on Linux, time to try installing it again?
degasus fixed latency and it should now be much better than the ALSA backend (close to the performance of XAudio on Windows).
@delroth - No. Not until PulseAudio comes with the latest Slackware install. I've used the Slackbuild, built it correctly, but I only got it to work on my laptop, not my desktop. By work I mean having sound output. Maybe if it's part of the distro it will "just work" for me
Aside from not being able to stream sound over a network (natively that is) I've had no trouble with ALSA in Dolphin. Too lazy to look at recent changes, but now that you mention PulseAudio and Dolphin, any special commits to dolphin-emu that I should be aware of?
EDIT - Too lazy to read apparently. That's great that PA has reduced latency, but I've really never noticed any latency under ALSA.
You aren't listening well then. The ALSA backend has something like ~100ms latency.
Not hearing it. At all. I'm serious.
I've been playing Dolphin for almost 2 years now. I would have known if there was audio lag that significant (I can hear the audio lag in OpenAL for example, as low as 10ms).
You wouldn't likely be able to notice an actual 10ms of latency. OpenAL's audio latency is actually 300-1000ms (or some other range of shit-to-extremely-shit like that) *plus* whatever you set the slider to; the timestretching plus OpenAL's inherent shittiness slows it down that much. ALSA probably actually does have ~100ms of latency (which isn't that noticeable – it's a tenth of a second, e.g. 6 frames at 60fps and 3 at 30). There are issues in Dolphin's mixer that add 40-80ms when running any games under LLE/new-ax-hle at full speed – and degasus is currently fixing those, I hear.
*Disclaimer: I haven't measured these numbers. JMC47 has, but I don't remember the exact numbers.
1/10th of a second is actually a lot. I've been working with audio programming lately, having to deal with sets of samples smaller than that for comparison in something like Audacity.
Perhaps you could post a comparison of the audio dumps from Dolphin. Something simple would suffice, like switching between the 1P Game and VS menus (with the control stick) very quickly in Melee. The sounds should happen quick enough were any significant variations in latency (100ms vs 300ms) would be clear enough looking at the waveform spacing.
EDIT - Or rather than using Dolphin to do the dumping, it would probably be better to use another program to capture all system audio output. Assuming Dolphin just dumps the samples it has generated, that wouldn't show any latency in the waveforms.
If Nintendo did that then they're doomed. Everyone owning a WiiU will be pissed.
(02-08-2014, 03:49 AM)Shonumi Wrote: [ -> ]1/10th of a second is actually a lot. I've been working with audio programming lately, having to deal with sets of samples smaller than that for comparison in something like Audacity.
Perhaps you could post a comparison of the audio dumps from Dolphin. Something simple would suffice, like switching between the 1P Game and VS menus (with the control stick) very quickly in Melee. The sounds should happen quick enough were any significant variations in latency (100ms vs 300ms) would be clear enough looking at the waveform spacing.
EDIT - Or rather than using Dolphin to do the dumping, it would probably be better to use another program to capture all system audio output. Assuming Dolphin just dumps the samples it has generated, that wouldn't show any latency in the waveforms.
Since our audio dumping is done after going through the AudioCommon Mixer you would miss ~50ms of latency there as well. All backends currently have a fixed latency bound because of this (which degasus is also working to fix).
100ms of audio latency is terribly easy to notice, by the way. The difference between 50 and 100 also is. And no, OpenAL does not have 300-1000ms of audio latency. Do you really think people would use an audio API that lags by 1s behind video? ...