SLI lock ups? Overheating?

Prod1gy

Member
May 13, 2005
31
0
0
I bought a new SLI system and had 2 BFG 6800GT's in the computer. Games would lock up about 10-15 min after starting, when logging the temps they would always be around 83C.

The company replaced both BFG cards with eVGA saying the BFG cards had alot of issues with overheating in SLI.

I got the eVGA cards and they are doing the same thing.

Still crashing.

Is 83-85C normal, even if it is somewhat hot is that hot enough to cause games to freeze up?

Thanks
 

deadseasquirrel

Golden Member
Nov 20, 2001
1,736
0
0
Overheating is just one thing that can cause such behavior. 85C seems a little high, so we can focus on that first. It sounds like this was a prebuilt system you bought and not one you built yourself. Do you know what kind of case and cooling it is? Other specs would help too, such as motherboard, CPU, memory, heatsink and cpu fan.

Each system can behave differently, and I am sure my SLi GTs have reached 85C here in the summer (Texas), and have had no freezing or lockups of any kind. Granted, I haven't logged any temps, but there hasn't been a reason to do so. Mine show 67C right now at idle. So throw some stressful 3D at them and they should be hitting 80C in no time.

What drivers are you using?
 

Prod1gy

Member
May 13, 2005
31
0
0
My systems specs are as follows:

Enermax 600W Noisetaker PS
Antec Plusview 1000AMG Case (with 5 case fans - 2 front intake,2 rear exhaust, 1 side intake)
Athlon 64 4000+ 90nm (Heatsink is ThermalTake A1772)
2GB (2 1-GB) Corsair TWINX2048-3200PRO) RAM
MSI K8N Neo4 Platnium
2-eVGA BFG 6800GT's
2-WD 74GB Raptors in RAID 0

Wiring is system is actually quite nice and tied off very well.

I ran memtest86 for a pass at 45 min long and 0 errors.

-62-63C on the primary card idle (ambient temp is around 40C)
-58-60C on the secondary card idle
-42-43C on the CPU idle (not sure underload)

I also tested the system with a single video card, no SLI, it locked up in game as well after about 15 min.

Using the 77.30 drivers, also tried the 71.89. Lock ups in all.

Thanks again for helping deadseasquirrel and anyone else that can shed some light.

The weird part is I was getting a BSOD on boot up right before the XP login screen, until I set the CPU VID to 1.45v.

That seemed to fix that issue.

Im in Texas as well, Austin here.
 

Nomada

Banned
Apr 27, 2005
967
0
0
Sounds to me like a RAM issue. Maybe the voltage needs a little nudge up. Is it overclocked.
EDIT: It being the system RAM. Maybe the RAM timings are too aggressive.
 

deadseasquirrel

Golden Member
Nov 20, 2001
1,736
0
0
Well, my fellow Austinite, your temps seem fine. Here are my GTs right now:

1-- 68C, ambient 46C
2-- 66C, ambient 50C

Yours are actually LOWER than mine, and I get no lockups at all, even after several hours of BF2.

What are your GPUs clocked at?
Your cpu?
Your RAM?
 

Acanthus

Lifer
Aug 28, 2001
19,915
2
76
ostif.org
It is my understanding that Corsair memory does not like MSI or DFI Nforce 4 motherboards. Id try bumping the memory voltage a tad and checking on the memory timings to see if they are in spec.

When you say "lock up" is it just the game freezing, or do you crash to desktop? Does it halt the whole system requiring a power down?
 

Prod1gy

Member
May 13, 2005
31
0
0
The machine freezes completely requiring a reboot.

Im not overclocking anything.

The memory settings are I think 3-3-3-8, but not 100% positive.

Everything else is stock.

So how would I go about bumping the memory voltage a little, what setting should I be looking for in the BIOS?

The GPUs are stock speed 350/1000.

What other information could help here?

Thanks..
 

fstime

Diamond Member
Jan 18, 2004
4,382
5
81
I think 85C is enough for the GFX cards to crash.

My 6 series always had 75C or less load, no crashes.
 

Acanthus

Lifer
Aug 28, 2001
19,915
2
76
ostif.org
Originally posted by: BouZouki
I think 85C is enough for the GFX cards to crash.

My 6 series always had 75C or less load, no crashes.

NVIDIA says they are safe well over 100C, and there is thermal throttling to prevent crashing that would just cause severe loss in performance.
 

Prod1gy

Member
May 13, 2005
31
0
0
Ok I think we can rule out overheating of the video cards as the problem.

I just got home and while I was gone ran 3dMark05 for 20 passes so it ran for about 200 minutes. Temps got as high as 88C with no crashes.

Then I go an play HL2 or BF2 demo and lock up within 15 min. I moved the computer to a better ventilated area. Temps in BF2 demo hit 80C one time and was mostly in the upper 70's during gameplay.

I tried adjusting the voltage some on the memory but still crashed after about 10-15 min in both games.

Any more suggestion of things to try before I RMA the entire system?

Thanks!
 

Acanthus

Lifer
Aug 28, 2001
19,915
2
76
ostif.org
can you see north bridge or mosfet temps on that mobo?

your temps do seem unusually high. although they are not dangerous to the graphics cards you may be overheating somewhere else.

I would also make sure you have chipset drivers installed for nforce 4, id get the latest from Nvidias website. Missing or outdated chipset drivers can cause all kinds of stability and performance problems.

Hope this helps.
 

Continuity27

Senior member
May 26, 2005
516
0
0
Originally posted by: BouZouki
I think 85C is enough for the GFX cards to crash.

My 6 series always had 75C or less load, no crashes.

You're not even close, before I installed better HSFs on the video cards, both breached 100C and were still perfectly stable. Throttling is set at 135C to mine, so unless they really screwed my throttling settings, they should be good for about 135C before getting into severe issues. (eVGA 6800GT)
 

Prod1gy

Member
May 13, 2005
31
0
0
Also trying some more memory tests. Running SuperPI right now. Just want to try and rule that out.
 

Prod1gy

Member
May 13, 2005
31
0
0
Ran SuperPI no errors took about 30 min.

I opened Sandra and watched MSI Core Center app, while I ran the Arithmetic CPU benchmark and the CPU multimedia benchmark and noticed something weird.

In the middle of the screen of Core Center it gives the current Mhz that the processor is running at.

At various times during the test it would report really bizarre stuff.

It would show 200Mhz x 12.0, then a second later change to 200Mhz x 48.0 then go back to normal then go to 200Mhz x 4.5 and so on.

Is this normal? Maybe bad CPU or motherboard?

Also is there a program that will log the CPU temps while in games etc?
 

Prod1gy

Member
May 13, 2005
31
0
0
I was watching the temps while the CPU benchmarks were running and the highest I saw the CPU get was 46C.

The weird part was I saw it go really high sometimes like 48 multiplier.
 

Acanthus

Lifer
Aug 28, 2001
19,915
2
76
ostif.org
That just sounds like a software bug unrelated to your issue to me.

Have you tried memtest86 and prime95?

Edit: my last shot in the dark guess would be plain old windows file corruption, you said you were getting BSODs earlier. I would try a format before RMAing as a last resort.
 

Prod1gy

Member
May 13, 2005
31
0
0
I ran memtest86 for one pass, took like 45 min, 0 errors. I am running prime95 torture test right now.

Windows file corruption could cause games to lock up requiring a reboot?

I checked the event log to see if anything weird. Here is the only thing out of the ordinary that I saw, this was an error that looked like it happened everytime I restart the machine.

The COM+ Event System detected a bad return code during its internal processing. HRESULT was C0000005 from line 44 of d:\qxp_slp\com\com1x\src\events\tier1\eventsystemobj.cpp. Please contact Microsoft Product Support Services to report this error.
 

Acanthus

Lifer
Aug 28, 2001
19,915
2
76
ostif.org
Originally posted by: Prod1gy
I ran memtest86 for one pass, took like 45 min, 0 errors. I am running prime95 torture test right now.

Windows file corruption could cause games to lock up requiring a reboot?

I checked the event log to see if anything weird. Here is the only thing out of the ordinary that I saw, this was an error that looked like it happened everytime I restart the machine.

The COM+ Event System detected a bad return code during its internal processing. HRESULT was C0000005 from line 44 of d:\qxp_slp\com\com1x\src\events\tier1\eventsystemobj.cpp. Please contact Microsoft Product Support Services to report this error.

Ive had windows corruption cause everything from random app crashes to slow perfmormance to disappearing files. Im honestly not sure what that COM+ error means.
 
Mar 30, 2005
60
0
0
do you happen to have a SB card in that system?
I was getting random game freezes with my 6800 GT because of an IRQ sharing between my graphics and Audigy card. Also, make sure you have cool and quiet disabled, that could be the cause of your multiplier changing.
 

Prod1gy

Member
May 13, 2005
31
0
0
I will check into the IRQ thing. I am using the onboard sound which is Creative but who knows that could still be happening.

I will also ask the tech support guy at the place I bought it about reinstalling Windows XP.

Other than that I am going to RMA the system.

BTW: Prime95 ran for over 4 hours with 0 errors 0 warnings.

Also, the CPU temp reached a max 49C the entire test.
 

Acanthus

Lifer
Aug 28, 2001
19,915
2
76
ostif.org
Being as this computer was built for you by a company, at obviously great expense, i would just RMA it to be honest.

Youre going through too much trouble for something that shouldve worked in the 1st place, they can reformat it if they want for free.
 

deadseasquirrel

Golden Member
Nov 20, 2001
1,736
0
0
Originally posted by: Acanthus
Being as this computer was built for you by a company, at obviously great expense, i would just RMA it to be honest.

Youre going through too much trouble for something that shouldve worked in the 1st place, they can reformat it if they want for free.

:thumbsup:

Make the builder stand by their work. If they can't get it working to your satisfaction, get your money back and build yourself. I'm here in Austin too and I'll even meet up and help ya. It's not difficult; you'll save a lot of money; and if any issues come up in the future, you'll know your system inside and out and troubleshooting will be much easier.
 

Dman877

Platinum Member
Jan 15, 2004
2,707
0
0
If this system was prebuilt, why don't you just call them up and bitch about it until you get a new one? That's the best part of buying prebuilt... you don't have to deal with these headaches :).