Rapid FPS drop in game... any ideas?

M

msjones

Weaksauce
Joined
May 21, 2008
Messages
71
Hi,

With money being rather tight these days I have been unable to fix my bust intel box so I have gone back to my AMD rig (see sig for specs). I have updated the graphics card to the 4670, which I have to say is a great budget card. While this is not the worlds best box it still has some life left in it. However I am having FPS issues in games. Mainly UT3.

I have removed the HSF, cleaned and re-greased the CPU & HSF. I have also cleaned/re-greased the cooler on the 4670. The problem is when I am in game after about 5-10 minutes the FPS drop from 60 to around 5-10 for a good few minutes and then back up again, and this happens quite often. So as you can imagine this has an impact on my fragging.

My first port of call was to check the temps on CPU & GPU. My CPU's loads is never more than 55C and my GPU has peaked at 60C but 9 times out of 10 is in the low 50's. So this is not the issue.

I am powering the box with a noname 460w PSU, which should be more than adequate as the GPU doesn't require a PCI-E power connector.

My motherboard is only PCI-E 1.0 and the card is 2.0 but again this shouldnt be the problem as the card is backwards compatible with earlier standards.

I really am at a loss, I have defragged the drive, even a fresh install of XP but I am still having the same issues. I am also noticing this drop in other games like The Witcher & Left 4 Dead.

Any ideas?
 
What resolution are you playing the game at? If you're running at a high res that may become a problem for that 4670.
 
I am playing at my monitors native resolution which is 1680x1050. I get the same issue when dropped down to 1024x768.

So its looking to be my PSU then?
 
silent-circuit said:
A no name PSU could easily be rated at 3x its capacity...and then it wouldn't be enough.
Click to expand...

Sudden drops in FPS are not a sympton of PSU issues. Trust me. I've experienced similar issues with ATI cards in the past. Unfortunately, I have not seen the problem in a hardware configuration remotely similar to this one. The incident I'm referrring to was most likely caused by a very specific combination of hardware that doesn't match the OP's rig in the slightest.
 
Are you sure it is the timing and not the location in the game you are at? Perhaps your video card is barely running UT3 (I know my 8600 GTX has difficulty), but when you hit a graphically intense place (could be unnoticeable, like high polygon count) it could choke.
 
MindChild said:
Are you sure it is the timing and not the location in the game you are at? Perhaps your video card is barely running UT3 (I know my 8600 GTX has difficulty), but when you hit a graphically intense place (could be unnoticeable, like high polygon count) it could choke.
Click to expand...

This is true. It could be just that simple. The system isn't exactly high end. (Not trying to be mean or anything) and while UT3 isn't the most demanding game, it really can put the hurt on older machines.
 
No offence taken :)

Well I have played a UT3 deathmatch maxxed out on bots and it ran fine untill the FPS drop, it picks back up about 1-2 mins after.

Even on games like the witcher the FPS drop can be when im stood in the middle of a field with nothing going on around.
 
Yep.

I've experienced it with a 4870-512 in both Stalker games. Playing along fine and fps just drops to single digits or teens and stays there. Couple other games as well though I don't remember having trouble with UT3, which I did play at that time. Alt-Tabbing out and back in (in the games that let you do that) usually returns fps back to normal for a little while. With Stalker I had to pull down the console and type vid_restart and that would fix it for a while.
512Mb ATI cards combined with Vista 64 seems to be a common thread with this problem. I didn't notice what OS your running, just saying what I've found as my OCD drove me to find answers for this immensly frustrating problem. I literally wasted 2 months of my life lurking and searching the web. I even wrote ATI but they never responded. Thought about posting a video of it doing it on YouTube but I never did.

Only thing that truly fixed it for me was finally buying a GTX 260. I heard that people who went to the 1Gb version of the 4870 saw the problem go away as well. I hope so cause I just bought a 4890 to see how they do.
(I know, not that much better than a 260 but I was itchin for something new plus I have a Crossfire board and want to eventually use it at least once. LOL)
If it's doing the same thing, off it goes and I'll limp along till December with my retro A64/WinXP system I just built for my daughter and play all my old games again.

Good Luck!
 
Well my issues were with three different 4870 X2's. I had two of them in CrossfireX but I tried all three individually. I even tried pre-production and newer BIOS ROMs for them as well. Eventually i developed a theory as to why I had the problem. In any case the fix was to ditch the two 4870 X2's for three Geforce GTX 280 OC cards. I've not had any problems since making the switch.
 
this is not looking good, should I really just give up with this rig and force my budget to accommodate?

I have noticed during the time of the lag the CPU jumps to 100%. This seems to be causing the drop, however there is nothing in task manager saying its using the CPU other than the game. Its a fresh XP install with only drivers and the games installed, no AV etc. I had vista on this box before XP and it did the same FPS drop in games and that had AV etc running.
 
That my CPU is shite :) and is causing issues for this card. Bottlenecking? But then why does it play fine >50FPS when its not lagging?
 
It's not your cpu.

I meant Dan. Your theory man. I must have your theory damnit.
 
subscribed



Still waiting on Dan's Theory. Any thoughts on this could help. Even if it doesn't seem probable or likely. The more things people try, the more can be eliminated and the closer we are to a solution. My ultimate dream was that some really talented programmer/troubleshooter would get a card with this problem and publish his fix for all the web to see bringing joy to the masses as we all sing glorious songs of praise. Alas, not to be.

My theory is an evil plot by nVidia.

See how easy that is?
 
Last edited:
Nope still doing it, but disabling AI has gained me 1-3FPS in the drop :) not enought to be playable though.

I will just add that im still getting this running at 1024x768 with low settings...
 
I'm not running any AV at all. I am running on a fresh install of XP with only my drivers and games installed.
 
wrangler said:
What was your theory?
Click to expand...

wrangler said:
It's not your cpu.

I meant Dan. Your theory man. I must have your theory damnit.
Click to expand...

My theory only had to do with my system. In my case I'm running the Intel D5400XS which uses Intel's 5400 "Seaburg" chipset. The board also has two nForce 100 series MCP's onboard. The 5400 chipset is PCI-Express 2.0 compliant and the nForce 100 MCP's are not. They are PCI-Express 1.0/1.0a compliant. As many of you know the nForce MCP's don't really effectively add PCI-Express lanes. They still go through the chipset's PCI-Express lanes. In essence, they interface with the chipset through the chipsets native PCI-Express lanes. In this case the signals have to be translated from PCI-Express 2.0 to 1.0/1.0a and back again. Speed isn't an issue as PCI-Express 1.0/1.0a has more than enough bandwidth for the 4870 X2. With that said it does add latency. I believe that this process of multiplexing the chipset's native PCI-Express lanes caused a considerable amount of latency which caused the random FPS fluctuations I experienced with the 4870 X2. This theory is substantiated by test results in other machines that were not equipped with the nForce MCP's and other reports of the exact same issue being reported by individuals using 680i SLI, and 780i SLI chipset based motherboards. The latter uses the nForce 200 MCP. I do not yet know if this problem will persist with nForce 200 equipped X58 chipset based boards. However, that's not quite the end of it either. Not everyone with a nForce MCP equipped system has reported the same issues. Also this issue was hard to narrow down due to the extremely low market penetration of Intel D5400XS motherboards.

I don't know that this was specifically hardware related, or that it couldn't have been solved with driver updates or BIOS updates. Though those issues never have been solved to my knowledge. AMD didn't respond to our inquiries and none of the Intel BIOS or driver updates ever resolved this. Again it was just a theory, though it happens to fit established facts in all the cases I read about where this problem was reported.
 
One thing I never tried, mainly because I was Crossfiring 4870's and you have to use CCC to enable Crossfire, was installing only the drivers without CCC and using a program like Ati Tray Tools or RivaTuner as my control panel which from what I understand now support Crossfire. I know your only single card but........worth a shot. Both easiest found @ Guru3D.

You might try that. Totally clean out the drivers using Driver Cleaner then install the bare drivers. You might not even need a control panel if your not OC'ing or trying to force AA or AF.
 
Ah

Thank you for that.

I sent a detailed letter and many driver bug reports and never got a response as well.

If this new card does it, I AM going to do the video thing and post it.
 
I am looking at getting a cheap AM2 chip and board, can pick both up from microdirect for about 80 quid see if this changes anything.
 
Just a thought:

The HD 4670 I have just put in my AMD box is new, yesterday infact. Before I was running an nVidia 8400GS and games where steady playable (without fps drops) with the old card. So that really narrows it down to the new radeon. Could it possibly be my no name PSU?

Also I have recently updated the motherboard bios to the latest version, that wouldn't affect anything would it?
 
Come to think of it it can't be the PSU. This same PSU has powered a P4 3GHZ and an AGP Radeon HD 3850 which requires PCI-E power.
 
MindChild said:
No, it couldn't. If your PSU dips below the power needed, you'll crash or freeze. Plain and simple. Your hardware will NOT gracefully deal with it and scale back.
Click to expand...

100% on the money. Many people seem to want to blame power supplies for things they can't logically be responsible for.
 
New development:

I have just re-flashed the bios on my motherboard with the new Radeon 4670 in, booted windows and fire up UT3. No frame rate drops? Could this really of been the issue?

This was the same bios version (A8V702) as I was using when I was having the FPS issue.
 
Also, I have been looking into my PSU, even though its not the issue. The model number is MPT-460XP. I found this chart on the web:

http://database.ul.com/cgi-bin/XYV/template/LISEXT/1FRAME/showpage.html?name=E199529&ccnshorttitle=Power+Supplies,+Information+Technology+Equipment+Including+Electrical+Business+Equipment+-+Component&objid=1073887472&cfgid=1073741824&version=versionless&parent_id=1073787374&sequence=1

If you search for my PSU (near the bottom) Am I right in thinking this is telling me I have 34 amps on the 12v?
 
Just like to chime in and say that I had a friend who had similar issues and it was a PSU related problem, modern video cards that require multiple power connectors will behave oddly with low power, I think certain cards wont enter the 3d clock speeds and remain at 2d clock speeds if there is no power or insufficient power to the card.

Obviously it depends on what hardware you have and how its used, I had some strange behaviour recently and found out my 850W was struggling with adding in another Tb HDD so I had to fall back to 3 HDDs for the time being. Mostly just BSODs but everything from applicatione errors to devices behaving erraticly. It would be OK most of the time but say copying between 2 of the drives it meant they spin up and consume more power pushing the PSU over its limits.
 
OK so I have the PC next to me now at work. I have just fired up UT2004 to have a quick game with the boss (I love my job) and the low FPS kicked in.

I havent noticed this before but when it start with the low fps a can hear like a high pitched squeal, hard to describe but I only hear when the low fps kick in.
 
msjones said:
OK so I have the PC next to me now at work. I have just fired up UT2004 to have a quick game with the boss (I love my job) and the low FPS kicked in.

I havent noticed this before but when it start with the low fps a can hear like a high pitched squeal, hard to describe but I only hear when the low fps kick in.
Click to expand...

Have you tried different driver versions? Have you removed the drivers with driver cleaner and then tried again? Have you tried the card in another PC? What's the story with your Intel box?
 
Frostex said:
Just like to chime in and say that I had a friend who had similar issues and it was a PSU related problem, modern video cards that require multiple power connectors will behave oddly with low power, I think certain cards wont enter the 3d clock speeds and remain at 2d clock speeds if there is no power or insufficient power to the card.

Obviously it depends on what hardware you have and how its used, I had some strange behaviour recently and found out my 850W was struggling with adding in another Tb HDD so I had to fall back to 3 HDDs for the time being. Mostly just BSODs but everything from applicatione errors to devices behaving erraticly. It would be OK most of the time but say copying between 2 of the drives it meant they spin up and consume more power pushing the PSU over its limits.
Click to expand...

Sorry, but this theory doesn't pan out. An insufficient, unstable or otherwise faulty PSU will NOT cause the system to act that way. It will become unstable, BSOD, crash, hard lock or whatever. The video cards do not have such complex power management capabilities. They can not do what you credit them with doing. They simply don't have that capability.
 
Ahem.... Well after sorting out the cables inside the case, I booted back up and went into UT3, half hours worth of high detail fragging and no frame drop. Same with the Witcher. Both ran smooth.

Do you know what different? The side if off of the case. So I guess after all that the is a heat issue. I guess what I really need is a bigger. better cooled case.

I just can't understand it, using Everest to monitor temps and everything is well within range?

Sorry guys :)
 
msjones said:
Ahem.... Well after sorting out the cables inside the case, I booted back up and went into UT3, half hours worth of high detail fragging and no frame drop. Same with the Witcher. Both ran smooth.

Do you know what different? The side if off of the case. So I guess after all that the is a heat issue. I guess what I really need is a bigger. better cooled case.

I just can't understand it, using Everest to monitor temps and everything is well within range?

Sorry guys :)
Click to expand...

What are your video card temps?
 
Under load it never peaks over 50-55C since reapplying thermal paste. CPU never above 55C, but this was the first thing I did before starting the thread.
 
Video cards won't reduce their speed for lack of power, but they will underclock themselves if they get too hot. However, your card shouldn't be underclocking at those temperatures. If you've got a second monitor, you can run GPU-Z on the second monitor and see what the clocks and temps do while you are playing games.
 
No problem with my 4890 in several hours of gaming today.

Whoo Hoo!!!

Gonna get another one.

Glad to hear you've solved your problem.
 
Dan_D said:
Sorry, but this theory doesn't pan out. An insufficient, unstable or otherwise faulty PSU will NOT cause the system to act that way. It will become unstable, BSOD, crash, hard lock or whatever. The video cards do not have such complex power management capabilities. They can not do what you credit them with doing. They simply don't have that capability.
Click to expand...

A quick google showed me there was plenty of cases where people mentioned low power modes for video cards not reciving sufficient power, and I have read about video cards only running in 2d clock modes when only 1 power cable (of 2) was connected, although i cannot find any articles specifically stating that I've definately read it in the past.

Showing that video cards have some form of rudamentry failsafe behaviour when insufficient power is supplied.

I'm not saying its "complex" nor am i saying that it's the specific problem here (i dont have enough information to be sure) it just shouldn't be ruled out because you think it's impossible.

Check here for an example: http://forums.nvidia.com/index.php?showtopic=95843&pid=536515&st=0&#entry536515
 
You must log in or register to reply here.
Back
Top