Groovymame Win 10 Sound Dips (Slowdown)

[trev1976]

Hi
Recently I rebuilt 2 groovymame computers , both run win 10 and GM 0.257 , I've noticed on both that when playing neo geo games the sound will slowdown for a second randomly (could happen on other games , I havnt tested) the visuals and gameplay seem fine.

I've put both computers in High performance mode but hasn't helped.

1x Q9550 , 4gb Ram ,HD7570
1x I5 3470 , 8gb Ram, HD7570

I tried out some games on retroarch and havnt noticed and dips ? Any ideas ?

I didn't have this problem with my old XP , E8500 , HD4350, 0.147 build, but that died

Thanks

[lilshawn]

the core 2 duo E8500 actually has a faster single core speed (3.1ghz) over the q9550 (2.83ghz)  in synthetic loads... the only thing the quad core does better is multicore loads due to the larger L2 cache 12mb vs 6mb but due to the way emulators (and software in general), mame only really takes advantage of dual core cpus (one thread game emulation and a seperate thread to handle sound (unless that has since changed...)) which doesn't really apply here so having more cores doesn't really help.

overall the q9550 is only about 17% faster than the E8500... and that only because of its multicore performance.

you want the fastest single core speed CPU you can shoehorn in there.

if you are set in the 775 socket, check your compatibility with your mobo... but see if you can get a hold of QX9770 - 3.2 GHz ( with FSB 400 x8)

The QX9770 is obviously the best 775 chip but it will require a high end motherboard which supports an effective FSB of 1600 MT/s (400 x 4)...this chip is still extremely expensive even on the used market RN... it's likely just cheaper and easier to go with an 1150 socket motherboard. hell, even keep a look out for a used AMD AM3+ motherboard and CPU combo with like a FX8350. saw one go on the local craigslist for 50 bucks for the mobo and cpu and 8gb ram and it's single core speed\ about 25% faster than either of those CPUs you got there.

[burn_654]

I would start with lowering your framedelay a little bit (if enabled) to see if it helps. You may lose some of its benefit but it can be pretty performance intensive - I noticed sound dips that were a result of the game running a few percent slower in parts because the higher framedelay stressed that build too much.

[trev1976]

OK thanks guys , I'll try a few things out.

[Zebidee]

If you're using an older case with newer mobo, make sure that you have the right front panel audio (FP_AUDIO) connector. Intel started using HD_Audio connectors instead of AC'97 around or before the times of Core2Duo. Both are 2x5 pin headers, so easy to mix up. If a HD_Audio connector is unknowingly connected to an AC'97 header or cable, it can have unintended consequences. Sparks or explosions unlikely, but (for example) plugging a pair of headphones into the headphone jack may not mute the speakers as expected.

This article discusses the matter: https://superuser.com/questions/419038/hd-audio-or-ac97-connector-which-to-use-when-and-whats-the-difference, and there is plenty more if you do an internet search.

You may also want to check your BIOS settings.

As you reported no problems using Retroarch, maybe it is software, check sound settings in your mame.ini. "portaudio" has been known to cause some issues.

Re: doubling down on lilshawn's comments, I still use Core2Duo machines with 775 mobos and E7X00 - E8600 CPUs for mame - I wouldn't bother with the quad core CPUs. A E8500 will give you better mame performance than a Q9550.

The E8XXX CPUs are very good performers in general and very cheap these days, although the E8600 3.33Ghz is a bit harder to find as it is the top of the line (though I've heard about an unofficial E8700, but rare like hen's teeth). As a testament to their single-core performance and durability, the E8600 are (were?) popular with overclockers, presumably adding to their rarity (wonder how many have been burned up?).

EDIT: comparing E8500 with Q9550, they both have very similar single core and dual-core performance, however the E8500 is rated at 65W vs the Q9550 rated at 95W. So the Q9550 uses nearly 50% more power to do the same thing as the E8500, produces more waste heat too.

EDIT2: I've overclocked an E8600 to over 4Ghz and was able to run Gauntlet Legends at "full speed" with no latency/stutter. Could have gone much higher apparently, but that was just one-time for the giggles, generally I don't overclock CPUs.

[Calamity]

Set syncaudio 0 in mame.ini.

[lilshawn]

it's really hard to apples to apple compare CPU's especially with how synthetic testing is. (test optimization by certain manufacturers has been known to happen and benchmarking CPU's is really only for clout and not really representative of actual workloads CPUs do on a daily basis.

about the best you can do is compare single core performance on any CPU you choose to consider.

i always come back to "cpu.userbenchmark.com" to do cpu comparisons. it's testing methodology is a little more down to earth with testing that caters to both the strengths of CPU's and their weaknesses.

take for instance, comparing the Q6600 and the E8500. https://cpu.userbenchmark.com/Compare/Intel-Core2-Quad-Q6600-vs-Intel-Core2-Duo-E8500/1980vsm13


most benchmarks (including this one) put the CPU's nearly neck and neck in terms of overall performance... and the consensus of the people seems to agree. I have seen people reccomend either one of these CPU's for this and that... The E8500 effective speed is 1% faster than the Q6600 according to userbenchmark so at first glance it seems like it's a non-issue and it seems like they would perform the same...

but if you look at the single core performance.. the E8500 has single core performance 50% faster than the Q6600, and dual core speeds 30% faster than the Q6600 despite the E8500 being a dual core CPU and the Q6600 being quad core. and only a 600 mhz difference in the clock speed.

the only reason the Q6600 makes up this huge gap is in parallel computing where multicore processors really shine... where despite the core speeds being only 600mhz slower than the E8500, it has more of them, and is able to use those cores to do the same "work" with 2x the cores as the E8500.

[trev1976]

Thanks again , I do have a E8500 knocking about so might pop that back in and try the sync audio setting

[Zebidee]

Yes, more cores doesn't necessarily mean increased performance, and you have to look at how the computer will be used. The CPU benchmark sites are great.

For all those reasons, last year I replaced a Q6600 CPU (came essentially "free" with a second-hand AsRock 775 mainboard I'd purchased) with an E8500 (and new heat sink) before fitting it all into a MAME PC I was building.

I've built about 5 MAME PCs using similar parts (and running Windows 7 and CRT_emudriver) as general purpose gaming systems that I can just throw into a cab. Found permanent homes for 2-3 so far.

Thanks again , I do have a E8500 knocking about so might pop that back in and try the sync audio setting

Good plan 

[lilshawn]

with testing that caters to both the strengths of CPU's and their weaknesses.

sorry, my ADHD is going to make me clarify this statement more.

if i'm using a computer for mainly re-coding audio or video files or I'm using a multi-core/multi-thread application... I would compare cpu's and look at their multithreaded and multicore performance... since it's possible a technically "slower" multicore speed can still outperform another CPU. (due to process changes, L2 amounts, etc) this is where the multi core testing comparison between CPU's  at userbenchmark comes in handy.

if i'm using a computer for single threaded applications, more cores can be useful in that i can process multiple threads at the same time. but i am more interested in the single core speeds to get those threads moving than trying to do multiple things at the same time. this is where the single thread comparisons come in handy.

most sites just plunk a "benchmark" number like "6246" next to a cpu, and that's what it would be "rated" at, and people would use these numbers to compare one cpu to another.

... but that number doesn't tell the whole story and is subjective to the test(s) performed. (just the same as the Q6600 vs E8500 comparison i example'd above.) They would be assigned "benchmark numbers" comparatively close to each other... in fact passmark scores "1813" for the Q6600 and "1257" for the E8500... and in the grand scheme of things, you'd be tempted to think the Q6600 was some 15% better... but these 2 CPU's are optimised for 2 different tasks and the E8500 outperforms the Q6600 handily in single core loads because it's strong in those suits... and it would only be in multi-core loads that the Q6600 is strong in, would perform better.

this is why apples to apples comparisons are so hard to get... and why i appreciate in particular the work and testing that userbenchmark does in supplying this kind of information.

TL;DR - more not always betterer

[trev1976]

Thanks again , I do have a E8500 knocking about so might pop that back in and try the sync audio setting

Just an update on this , I've adjusted the sync audio setting and so far so good. 👌