81 replies to this topic

[destruk]

Sounds like a really cheap ass card design to me.

Edited by destruk, 20 April 2011 - 08:01 AM.

[SnailGary]

Ok, just let me add my latest findings from the last evening. Maybe this will take some blame away from the card makers...

As most others I was totally convinced this is all about graphics cards or maybe drivers not being utilized correctly. However what I learned yesterday made me think again. We may actually be gazing at a secondary or even tertiary side effect of something completely different. What I am trying to say is: it could actually be related to the sound, not the graphics!
(Please don't call for an ambulance just yet, I will explain in a second!)

@maceman and others on Win7/Nvidia: Please try to run the Windows "Sound Recorder" (to be found under Accessories) and watch the framerate. There you go. No graphics display involved.


What I did then was fire up this nice tool called Process Explorer. With this, you can get performance info on the single threads inside an application:

- In a typical low-FPS scenario the ratio between VPinball and vpinmame is roughly 1:1 on my system. Each thread draws around 10% total CPU time (total 20%) on my 3-core AMD. This is still far from the theoretical 33% it could reach, but ok.

- In a high-FPS scenario, the ratio is changed to 3:1 with VPinball taking up 15%, vpinmame 5%. vpinmame still running smooth, no stutter!

Now the sound recorder and many other applications seem to magically change the CPU usage of the vpinmame thread.

In my mind, pieces slowly start falling into place now: rob046 has reported some messengers gave him a boost: messengers also use sound, right? media players are already on the list, so what about FF4? don't know, but there is probably some sound functions in there, too. However, with Winamp this effect did not happen. So it may be related to the type of sound output? Or it may even require some sound input?

But even the sound access may only be one side effect in this puzzle. What I still need to find out is why does vpinmame suddenly draw less CPU power, giving way to VPinball for a speedup? And why are those two threads not utilizing the full CPU? Could be some thread sync problems? Or waitstates introduced by graphics display who then make the scheduler act funny?
Getting closer...

Edited by SnailGary, 20 April 2011 - 09:14 AM.

[kruge99]

To add another thing to this equation, I recently replaced my pci Audigy sound card with a newer pci-express Soundblaster X-Fi Fatal1ty card (just the card not the extra peripherals from that series) to my pinball pc and there was a dramatic performance increase and reduction of ball stutter. The performance improvement is from tables that I have used UVP and have multiball modes. Tables I used to have stutter on like Scared Stiff, Theatre of Magic, Tales of the Arabian Nights, and Monster Bash have nearly smooth ball movement with UVP enabled. With UVP disabled, they play even better.

This is not a recommendation nor an endorsement for pci-express sound cards so don't go out and spend upwards of a $100.00 on one of these because it may not do anything for people who already have "high-end" systems.

I will be testing this thing tonight with running Windows sound recorder to see if there is any further improvement or degradation of performance.


Best Regards,
Todd.

[blur]

great find

sound (or lack of it) is a known source of some problems with hyperspin

[cupid]

Yes, really great find! Thanks Snail!

That sounds promising... I think that VP could need a new Sound Lib. Maybe with directional sound and loudness control. ;-)

I thought that it had somethings to do with optimized graphics Drivers or the DirectX-System: like This: Hey, only DX7 is used... then it can be nothing serious, lets save some performance.

[maceman]

Thanx guys,

If you read my last post...I had mentioned about sound already yesterday. This was already discovered after one day research.

...by turning off the MMCSS service(this is sound!!) which controls all multimedia, there is a drastic jump in performance. You can't run like this of course, but it got me thinking. I didn't have time (or brain-power left) to follow up, so thanx Gary- that is good additional info that save me time I was on the chat last night having KEG try a few things and we were talking about his sound as well...because that is when i found the thread about sound from the other forum.

This doesn't explain to me why opening FF to vpforums (or the like) makes an additional boost, unless something with HD onboard sound is either disabled or enable when opening these programs.

What i think we need further as information before testing further is the following:

1. Does anyone here use an Nvdia newer card with good performance(fps). and if so, what are your SOUND specs, ie. are you using integrated or EXT.

2. Does anyone have or can get an EXT card to try diabling their onboard HD soundand try to see if that works as Todd experienced.

What I can tell you on my slow fps system is that I have an integrated HD sound. I am not using any of it. I have an old rca jack type stereo plugged into my card, always been this way....so i mention it as perhaps not using the onboard properly is limiting something.


i want to hear from Noah especially about his sound, since he is running smooth....andanyone else who has a good working nvidia based system....

Maceman

Edited by maceman, 20 April 2011 - 04:52 PM.

[settingsons]

It is funny how I thought this thread at the start was just a crazy joke... but now it has turned into one of the most fascinating threads for a long time.

Just to add a bit of my experience. When I added a Xonar sound card to my setup, and disabled the onboard sound I did get about 10% FPS increase but the main benefit was sound quality. Also although I don't get hardly any stutter nowadays (after much tweaking and component upgrading) what I still notice in particular on AFM and Indianna Jones (Lord Hiri's), is that when there is a lot of sound activity, I get some brief excessive stuttering for a couple of seconds, but then it seems once the sounds have played once it is as if they are cached and the same sounds play fine subsequently.

I use XP 32-bit SP3 with only about 6/7 services running, an SSD harddrive, and GTX 260 896MB (Playfield and Backglass) with Geforce 8400 GS (DMD). Get about 800-850 FPS on AFM with UVP and 3rd monitor showing DMD. Both backglass and DMD are set to 16-bit, and ddraw is set to off (zero) on all pinmame tables.

Edited by settingsons, 20 April 2011 - 09:07 PM.

[maceman]

Thanx SS,

you know it is funny how we never put these facts together before either. I mean, like you...i have noticed the glitch when sound happens in a game, and i never put that together until your post! Thanx.

It excites me to think that whatever we discover will most likely get fixed for the next realease! This will be major!

Maceman

[moondrummer78]

I wonder if updated Direct X support would offer any improvements. True, all the cards are backwards compatible, and so is Windows with it's compatibility Modes, but who knows!

[rob046]

I was wondering about possible sound issues causing this too, because a few months back my sound card (creative X-fi Titanium) stopped being able to record, like using "what u hear". I don't know why it stopped working, though I do know that x64 creative driver's haven't always been the best.
But when Gary brought up the windows sound recorder thing, it immediately reminded me of my own issue with sound recording. Followed by another thing I noticed a few weeks back, which was the "audiodg.exe" process now hogging usually 4% cpu resources.
Where before this wasn't something I ever noticed.

So I did some testing. I disabled (in device manager) my creative card, disabled my 4 "nvidia high definition" audio things (those are just for if you use HDMI if your card supports it), & enabled my onboard sound. Unfortunately none of them changed anything.
Enabling my onboard sound did however fix the "audiodg.exe" issue, it went back to 0% & my computer idles at 0%. Not that 4% cpu usage is a big deal, but it kinda is when I know it wasn't there before. I don't think creative has new drivers, but I'm gonna check for new ones or maybe uninstall/reinstall current ones.
Though considering enabling my onboard audio didn't help, I don't know that doing this will affect VP at all. It might only fix my cpu usage issue.

Yeah, both through my sound card & onboard, I pretty much get the same fps in VP, & things are otherwise no better, same issue is still there unless I enable one of the programs I've talked about. Maybe even a couple extra fps gained with onboard audio, but not nearly enough to warrant me not using the creative card since that card helps performance on other gaming & has some great features I use. Though I do hope for better drivers soon.

I'll keep messing around, but it is good to hear we might be getting somewhere on this issue.

As for clock issues, I don't think that is my issue here. When I first noticed this problem that was the 1st thing I checked. I use MSI Afterburner to see what my clocks are doing when I run VP or other games, & according to that, even in adaptive move my clocks always go to full speed when running VP both when I do & don't have my odd fix enabled. So that doesn't seem to be related unless my Afterburner software is wrong, which it shouldn't be, its a pretty good program.

[maceman]

Thanx again Rob!

Yeah, I agree that clocks are not really the issue... i mean, yes, if we get higher clocks with both cpu and video card, the obvious thing expected would be higher fps's, but that doesn't fix the current low fps issue.

I am glad so many people are diving into this and trying, every little bit helps!

[rob046]

The latest Nvidia drivers seemed to have fixed this issue for me! I did a few other things as well, but I suspect the drivers did the fix. You guys try it!
Not only did the new drivers seem to fix this issue, but I even get a few more fps than I did before! & the drivers I were using weren't that old, like 267.xx, I just updated to 270.xx, whatever it is.

Before I'm 100% convinced things are fixed, I want to maybe restart a couple more times (my pc), & keep launching a few different tables to make sure.
It would be great if this driver fixes this issue for everybody, but it also makes me even more wary of Nvidia in that you can't trust their drivers. There is another thread going on now where one guy is experience much worse performance with the new drivers, meanwhile I'm experiencing the opposite. Geez.
If this is a good fix, I sure hope that upcoming drivers continue to have this fix.
I still recommend people might want to go with ATI these days when choosing a card for VP use.

[maceman]

I knew the fps would be higher with the newer drivers... problem m is that I play cab with 60/40 Vp to FP ratio. FP will only work with 260.99, so i hope to get the problem solved while using that driver. Good to know though!!

Maceman

[rob046]

well given the history of this issue & other issues, even though this has seemed to work for me (I think), who knows if its gonna work for everybody. I'm sure snail & others will give it a try & if it does fix things, maybe lets still dig in & find out why with some drivers this issue occurs. I'd still like answers as to why it happens. Maybe look at driver release notes to see what they did?
There may have been some pretty major changes since the drivers are now in the 270.xx range.

So far its great news though, this seems fixed for me, plus I got a nice little fps boost. Not much but I usually don't expect any fps boost with new drivers with VP, so its a nice bonus.
Now between that & disabling desktop composition (maybe a 5% boost?), even JP/Grizz's TOTAN is running perfectly smooth at almost all times. & if I do see jitters it is very minor.

[toxie]

that other guy experiencing slowdowns was me (in fact its really drastic, not like 10% slowdown, more like 80% or so)..
but maybe its only for the laptop variants of 270.XX? i know that they have a lot of additional trickery in them for saving power, the optimus thing (switching seamlessly between onboard and additional gfx board), etc.

[rob046]

that other guy experiencing slowdowns was me (in fact its really drastic, not like 10% slowdown, more like 80% or so)..
but maybe its only for the laptop variants of 270.XX? i know that they have a lot of additional trickery in them for saving power, the optimus thing (switching seamlessly between onboard and additional gfx board), etc.

You sure you got the right driver? I'm thinking that the desktop drivers might not be the same as laptop drivers, which might be the issue right there. How are other games performing?

[toxie]

yup, of course.. i'm always trying to use the latest WHQL one (i.e. also never experienced such a slowdown with VP before), especially to detect issues like this one before other users would have too..
and the strange thing is that no other game showed slowdowns (tested with only a handful of rather dull ones though, the newest being SORRv5)..
maybe it will resolve itself with the next driver update again, who knows? maybe its simply a bug in the driver..

[kruge99]

To add another thing to this equation, I recently replaced my pci Audigy sound card with a newer pci-express Soundblaster X-Fi Fatal1ty card (just the card not the extra peripherals from that series) to my pinball pc and there was a dramatic performance increase and reduction of ball stutter. The performance improvement is from tables that I have used UVP and have multiball modes. Tables I used to have stutter on like Scared Stiff, Theatre of Magic, Tales of the Arabian Nights, and Monster Bash have nearly smooth ball movement with UVP enabled. With UVP disabled, they play even better.

This is not a recommendation nor an endorsement for pci-express sound cards so don't go out and spend upwards of a $100.00 on one of these because it may not do anything for people who already have "high-end" systems.

I will be testing this thing tonight with running Windows sound recorder to see if there is any further improvement or degradation of performance.


Best Regards,
Todd.

Running Windows sound recorder offered no visible improvement to VP912. The only table I could test was Roller Coaster Tycoon due to time constraints.


Best Regards,
Todd.

[bent98]

yup, of course.. i'm always trying to use the latest WHQL one (i.e. also never experienced such a slowdown with VP before), especially to detect issues like this one before other users would have too..
and the strange thing is that no other game showed slowdowns (tested with only a handful of rather dull ones though, the newest being SORRv5)..
maybe it will resolve itself with the next driver update again, who knows? maybe its simply a bug in the driver..

Did you do a clean install of the driver? Sometimes the old driver info is still stuck in windows registry or windows still uses old .dll. The simplest way to do it is when installing these new drivers do an advanced install and check "clean install". Its a new feature on nividia drives.

If that doesnt work there are driver cleaners you can download from guru.com that will clean out your reg files and remove all old video driver files.

[toxie]

i never had to do that before (at least not in the recent years), but i'll give it a shot over the weekend..