12 replies to this topic

[ParadigmCalibre]

This has been a problem making a game with great potential such a burden to play. Every now and again, my GPU usage drops from 99% to 0% and MWO seems to freeze whilst my CPU and network usage putter away as the game goes on without me. The remedy has always been to ctrl-alt-del, bring up the task manager in the foreground, wait until I see smooth movement again in the game behind it, and then alt-tab back (hopefully before I've died or run off a cliff).

This happens in no other game or under any other circumstance and plagues me regardless of graphics quality settings.

I'm running two GTX 550Ti's in SLI (though I know SLI isn't supported yet, so only the first GPU actually gets used), which seems to handle the game perfectly for the most part - it's only these periodic drops (once every few matches) that bungle it all up.

Is this happening to anyone else? Does anyone have any idea of cause and/or solution?

Happy to provide more information if needed.

[Catamount]

and yet another Nvidia user has problems with MWO; seriously, this is getting ridiculous.

The only thing I can suggest, if you aren't already, is running the game in Full Window mode instead of Fullscreen mode. Fullscreen is really buggy in this game, always has been.

[NocturnalBeast]

I had a real degredation in performance after the last few patches with my 680 GTX. I was able to get my stable framerates back by rolling my drivers back to the latest WHQL version and completely uninstalling, redownloading and reinstalling MWO.

[ParadigmCalibre]

I have been playing in Full Window mode - seems that was the default setting. Perhaps I'll try playing in Fullscreen mode (though they appear exactly the same). Then I'll try Windowed mode. I'll post back after trying a few matches in each setting.

Thanks for the idea.

[ParadigmCalibre]

All three modes elicit the problem - GPU usages drops to 0% and I'm left SOL.

Any other ideas?

[Catamount]

Could this be some kind of extreme thermal throttling? How hot does your GPU get?

[Durant Carlyle]

MW:O freezes for me when my GPU objects to too-high overclocking. I have a new setup and I'm working to find the best overclock for my GPU. MW:O will freeze and I'll have to try to Alt-Tab out. Most times it will let me, and I'll see the "The graphics driver has stopped working and rebooted itself" message in the bottom-right corner of the screen. It could also be caused by the GPU overheating, which would make the graphics driver reboot itself.

First, try Alt-Tab (multiple times) instead of Ctrl-Alt-Del to see if that works, and see if you can spot the driver reboot message. If you do, then something's not right in GPU-land. Maybe your GPU fan is clogged or has stopped working. Or maybe there isn't enough space in between the graphics cards and the one is overheating because of it.

[Shamous13]

I Found this in support workarrounds Sounds like ATI cards are being throttled at 70deg temp and the fan curve isn't aggressive enough to keep it cool. It would be interesting if this solved ati user FPS problems

Roland, on 10 October 2013 - 07:34 PM, said:

Ok, so I recently got a new machine, and had a GTX 670 in it. The behavior I witnessed was strange.. I could put everything on max, and it would run perfectly at 60 fps... for a little bit. Then, it would totally {Scrap} out, and the frame rate would drop heavily, barely able to pull 30.

I was like, "WTF? WHY U SO BAD MWO?"

However, after much testing and digging, I found the cause, and figured it may help folks.

The 670, and possibly all kepler chips, have built in throttling when temperatures get high.

Info explaining some of this comes from here.

Only, when I say "high", I mean really not that high at all... Throttling starts at 70degress. Since the card can go up to 90, while 70 was hot, it wasn't so hot that I thought it was the problem.. however, this is what was happening...

So, the fan had a really lazy speed curve by default. Basically, it would be off completely until the thing got up to like 40... then it would increase, but not fast enough to keep the card from hitting that threshold. This was exacerbated by the fact that the case didn't have really good airflow.

What happened is that the card would get hot, and then the fan would eventually kick on, but by then the case was full of heat and it would take a bit to cool things down. In the meantime, the card would bust through the throttling threshold and start getting all stupid. At 70, it supposedly only throttles down 13 Mhz, but whatever, the effect was huge... I don't think it was hitting the 80 degree throttle point ever.

Anyway, this was making everything terrible. I ran some 3dMark II tests, and was getting like 2000, instead of the expected 8.

The ultimate fix was to create a more aggressive fan curve, which basically keeps the temperature below 70, always. Now, the card works perfectly. Everything is cool.

Hopefully, this helps someone else who had the same issue with this card... I know I saw some folks who had complained about the 670 with what seemed like similar issues.

[Big Giant Head]

Driver issues?
http://www.geforce.c...orce-experience

[Big Giant Head]

if you activated Vsync try to disable it then try to set it to windowed mode (or full windowed)

[CooloutAC]

did you check the temps? Nvidia and amd cards are the same in that, if the card really does get very hot, the auto fan control really doesn't go higher then 35%. You have to manually up it past that yourself. Try putting it to 50% and see what happens. I like to keep temps under 70 degrees maximum.

A strange issue i have noticed, ist he mechlab gets hotter then the game itself. Total twilight zone status.

So a good tip is if you walk away form the pc don't leave the mechlab open haha. close out or at least minimize it if only for a short while. It feels like it secretly folds or mines for bitcoins. lmao.

Edited by CooloutAC, 11 October 2013 - 12:53 PM.

[ParadigmCalibre]

No drivers issues - everything is up to date and there's never a driver failure message.

Checking the temperatures though, I am seeing my cards to be 70-80degC. That's running them standard, I don't have anything overclocked. It's curious to me, though, why those temperatures would cause throttling only in MWO and nowhere else - shouldn't throttling be program-independent and instead be based only on GPU state? I will still try "underclocking" to bring those temperatures down and work on bolstering fan speed.

Vsync was disabled long ago as I lowered graphics settings hoping to resolve the issue. The window modes, I've already discussed.

[CooloutAC]

ParadigmCalibre, on 12 October 2013 - 11:09 AM, said:

No drivers issues - everything is up to date and there's never a driver failure message.

Checking the temperatures though, I am seeing my cards to be 70-80degC. That's running them standard, I don't have anything overclocked. It's curious to me, though, why those temperatures would cause throttling only in MWO and nowhere else - shouldn't throttling be program-independent and instead be based only on GPU state? I will still try "underclocking" to bring those temperatures down and work on bolstering fan speed.

Vsync was disabled long ago as I lowered graphics settings hoping to resolve the issue. The window modes, I've already discussed.

raise fan speed to like 40% thats what i do. to get it under 70. but your right i don't see that as being the problem but worth a shot.