tk421 wrote: ↑Fri 12 Jun, 2020 6:15 am
Hopefully someone finds a solution for this issue soon. May I ask if you have tried another video card, such as the S3 Trio 64 or the ATI Video Xpression? I typically set the Video speed to default.
Are you using a large virtual hard disk? Mine are normally 2gb in size. Also, how much RAM have you assigned to each PCem VM? I have rarely gone above 64mb, even for a K6-2 configuration. For a fast 486 VM I normally do not exceed 16mb of RAM. I use about 4mb of RAM for a slow 486.
Appreciate you trying to help.
Just tried the ATI Video Xpression with a Pentium 75MHz, 2GB HDD and 16MB RAM but it still has sound stuttering. I also set the CD drive model to 'AZT CDA 468-021 4X' because i used the 'PCemCD' drive before. Think i've exhausted all options with the PCem settings now apart from 486 and slower machines, but NT31 used a 486 anyway, and they still have this sound problem.
Emulating a CPU that's too fast for the physical CPU will make the sound stuttering worse, but that's an entirely separate problem that is caused by current hardware just not being fast enough to emulate the faster CPU options (in my case a 300MHz AMD K6-2) and that's probably why you had sound stuttering before you upgraded.
NT31 wrote: ↑Fri 12 Jun, 2020 7:05 pm
This may sound a bit far-fetched, but to those who see this...what CPU are you using on your host system? I'm starting to wonder if it might be an issue with certain CPUs. I'm using an AMD Ryzen 7 3700U.
PoloniumRain, I hope I am not asking the wrong questions. I think your analysis and assessment of PCem is good, but I admit I am using an Intel CPU on my laptop. My v16 tests have yet to produce this serious sound stuttering problem.
May I ask if you have tried earlier versions of PCem? Versions 13.1 through v16 offer many of the same features, minus the K6-2 of course. I find that most of my programs do not need a Pentium II, so the K6-2 is almost always passed over in favour of the Pentium MMX. Even today I use v13.1 for many of my programs. I noticed that it matters which version I use, depending on whether I use a joystick or a certain program that just works better with a different version of PCem. Perhaps if you tried an older version of PCem, the sound stuttering issue will not occur.
If necessary, you could use v12. It does not have Voodoo2 SLI mode or the ATI Video Xpression, but the Voodoo 2 is there. My experience with the Voodoo2 without SLI mode has always been positive. I have still been impressed with V2 SLI mode, but I rarely use it. I always try to make sure that my CPU is running PCem at best performance.
As suggested, i just tried out PCem v13.1 and the sound works perfectly.
It's working great in Windows 98, in DOS, and in any game i try. But in v16 it stutters/skips with all of these things.
I also noticed that v13.1 is faster than v16. Windows installed faster, starts up/shuts down faster, and games install faster. This version certainly seems better so far, but i miss some of the new features in v16.
So maybe it's not a Windows 10 problem as previously suspected, because everything else is identical. For example, i set up an identical machine in v13.1, with the same hardware as one of the machines i have on v16 that has sound stuttering, with the same OS (Win98). Right away the sound worked flawlessly on v13.1, even with the DOS game FX Fighter which is particularly bad on v16.
Being as the sound works fine for NT31 on v14 as well, i'd say it's a safe bet that the sound will also work fine for me if i was to try that version. It's clear i have the same problem as NT31. So something that's been changed in PCem since v14 is causing the sound to stutter/skip for some people.
Sorry for the bump, as I know those tend to get annoying, but I've decided to test this again by switching the recompiler on and off. I tested it by using an SB16 and playing a MIDI file. I was surprised by the results.
386DX/40 + No recompiler = No skipping
486DX/50 + No recompiler = No skipping
486DX/50 + Recompiler = Skipping every few seconds
Pentium 75 + Recompiler (Forced) = Skipping every few seconds
I know that we said at the beginning that the recompiler had been ruled out, but based on this, I'm willing to bet that the recompiler might in fact be the cause.
Another curious thing I've noticed is that the extent of the skipping changes depending on what host I use. On a Core 2 Quad machine, as well as an AMD A8 laptop, the stuttering is constant and really bad, making the music sound like it's playing at half speed. On my current laptop, which has an AMD Ryzen 7, the stuttering only happens every 5 seconds or so. In all of these situations, the emulation speed stayed at 100%. I don't know if this might indeed be a flaw in the recompiler or not, and I know I've hinted at it before, but now that I have a machine capable on running Pentium mode without lag, I'm pretty confident at this point that it's that. Please correct me if you think I'm wrong.
Another thing I found out during my testing: at least on my end, the stuttering only affects FM synthesis ( e.g. SB16). I set the sound card to SB AudioPCI and no skipping is present.
NT31 wrote: ↑Sun 23 Aug, 2020 3:52 am
I set the sound card to SB AudioPCI and no skipping is present.
Do you mean Sound Blaster PCI 128 or Ensonic AudioPCI? Because there's no SB AudioPCI option.
And does it still skip with DOS games? SB PCI 128 only seems to fix the stuttering in Windows games for me (although i haven't tried many yet) but it makes no difference with DOS stuff.
I was referring to Ensoniq, although the two options are almost identical, and I didn't test DOS games because I don't need the recompiler to run those at 100% speed. My priority here was Pentium-class Windows 9x emulation.
MIDI audio on the AudioPCI is done in software while in the other Soundlblasters is done by an emulated chip ( an OPL on most of them, and the EMU8000 on the AWE32/64).
Also, there are two OPL emulators, DBOPL and NukeOPL (Selectable from the soundcard options in pcem setup).
DBOPL is like the one in DoxBox. NukeOPL is a higher precision one, which also uses more CPU, but shouldn't really affect your case.
The audio on them is mixed into the soundblaster and then output to the soundcard.
If sampled audio works but midi skips, it would seem that there is a problem in the mixing stage or some kind of wait that doesn't produce enough audio at time of mixing.
