Page 1 of 1

Request for Windows testers

Posted: Thu Jul 14, 2011 11:54 am
by sixxie
I've uploaded a new Windows XRoar snapshot here: xroar-snap-20110714-windows32.zip (just the SDL version)

This has a Windows-specific audio driver added, but I've only been able to test it under Wine. Now obviously I get good results there, as I wouldn't upload it otherwise, but the real thing may well differ, so if anyone has the time, I'd be really grateful if you could test this.

I'm after any noticeable differences between it and the last release, so assuming it works at all:
  • Does it "stutter" at all? I expect it to if you move windows around etc., but in normal operation?
    • and if it does, is this alleviated with -ao-buffer-ms 30 (or 40, or 50...)?
  • Is there any obvious difference in latency? If you tended to notice the audio lagging in any particular program before, does this make it any better? Or worse?
  • CPU usage - more or less?
If this makes things generally better, I'll leave it as the default - fingers crossed.

Re: Request for Windows testers

Posted: Fri Jul 15, 2011 2:05 pm
by robcfg
Hi Ciaran!

I just tested the snapshot and it runs incredibly slower than the last release version.

I uploaded a small video to youtube so you can see how much slower it runs.

Hope this helps. I'll try to test it on a more powertful machine, but I think there must be something fishy for such a performance loss.

Regards,
Rob

Re: Request for Windows testers

Posted: Fri Jul 15, 2011 2:19 pm
by sixxie
Ooh, painful!

(Although to my ear the "normal" one sounded a little bit choppy too).

Okay, back to the drawing board, I think. Thanks for trying it! There's probably some minimum buffer size Windows demands that Wine doesn't (does -ao-buffer-ms 30 change anything at all?). I'll endeavour to find a reliable way of testing myself before I start bugging you here again :)

Edit: actually, there are many things it could be... If for some reason it obtained a different sample rate to that requested that would screw things too. Or if the call I use to obtain the playback time fails...

Re: Request for Windows testers

Posted: Fri Jul 15, 2011 2:23 pm
by robcfg
I tried the AO-buffer switch with no difference at all.

I also tried some playback rates, no effect again.

And, as you can see, the whole emulation is a lot slower than usual, see how cursor blinks in both windows.

I'll be glad to help you by testing any new versions.

Re: Request for Windows testers

Posted: Sat Jul 16, 2011 10:39 pm
by Alastair
sixxie wrote: (Although to my ear the "normal" one sounded a little bit choppy too).
The "normal" Windows one has always been a bit choppy.

Further to Rob's comments, I will mention that on my system when running xroar-0.25-windows32 (plus a few other programs) that CPU usage is between ~50 and ~85%, whereas when running xroar-snap-20110714-windows32 (plus the same other programs) that CPU usage is between ~45 and ~60%.