Overclock.net › Forums › Graphics Cards › NVIDIA › NVIDIA Cooling › Gainward GeForce GTX 690 temperature issue
New Posts  All Forums:Forum Nav:

Gainward GeForce GTX 690 temperature issue

post #1 of 4
Thread Starter 
My problem is that GPU 1 is running 30-40C higher than GPU2. Is this acceptable or should I worry? I've included screenshots below.

This is under load. Idle, the temperatures are somewhat similar, usually within 0-5C of eachother, but after stress testing today it seems the temperature gap has permanently increased to about 10C for some reason. Right now they are idle at 45C and 34C.

Before I start the process of reseating the waterblock, I was wondering if there could possibly be another explanation to this other than a hardware problem.

I have a GTX 690, with a Heatkiller waterblock, swapped thermal compound with Liquid Ultra. Apart from the GPU, the loop also includes the CPU, one 240mm radiator and one 200mm radiator.

Picture of setup: http://i.imgur.com/lqDbybR.jpg Front fan is intake, top intake and back exhaust - if it matters, with air flow in mind.

I'm not sure what other information to give you, but I've got these screenshots:

NVIDIA Inspector 1: http://i.imgur.com/ICkIYqY.png
NVIDIA Inspector 2: http://i.imgur.com/7tDEcZV.png

MSI Kombustor: http://i.imgur.com/Rdbs4ml.png
MSI Afterburner: http://i.imgur.com/T4rvB29.png

Just as a side note, my CPU seems to be cooling fine. It's an i7 3770k 3.5ghz (4.2ghz) ivy bridge and it's stable at about 53C when stress testing. However it is delidded, original thermal compound replaced with Liquid Ultra (on both the die and ihs) and water cooled so I don't know, is it good or bad? I probably should research it myself.

Any help is appreciated.

EDIT: So I failed to mention that I just recently turned SLI on in Nvidia control panel, and that's when these problems started occuring. Beforehand I guess the card just acted like a GTX 680 (?). Also I was and still am running 2 monitors, main is 120hz, secondary 60hz.

The framerate and temperatures were A-OK before I enabled SLI. As an example I was running Black Ops 2 at 120fps stable, everything maxed and GPU 1 would be in the low to mid 40s. GPU2 in the low 30s but under no load. With SLI enabled I can't go over 60 FPS, both GPUs are under similar load but GPU 1 is at 65C and GPU 2 is at 35C.

Another thing that happened during SLI is that Google Chrome started flashing black continuously like rapidly shutting and opening a blind - intensified by scrolling down the page. This only occured withing the window frame and did not affect the adress-, status bar etc.

After disabling SLI the FPS went up to where it had been but the temperature on GPU 1 is now at about 66C from the previous low mid 40C. The black flashing on Google Chrome also seized.
Edited by onlinez - 11/30/13 at 12:10pm

Gear mentioned in this thread:

post #2 of 4
I agree you should pull the card and reseat the block. There has to be a problem with the thermal paste or block fitment.
Headshot
(15 items)
 
  
CPUMotherboardGraphicsRAM
Intel Core I7 2600K@4.7ghz Tb 5.0ghz  Asus Z77 Sabertooth Zotac GTX 1070 ZT-P107001-10P (16gb) GSkill Ares DDR3 2400mhz @2133mhz 
Hard DriveHard DriveOptical DriveCooling
Adata 240gb SP550 SSD WD Blue WD10EZEX 1TB LG BD-ROM Thermalright Venomous X 
OSMonitorKeyboardPower
Windows 10 Home 64 bit Asus MG248Q 144Hz 1920x1080 Gamdias Ares Cougar GX G1050w 
CaseMouseMouse Pad
Corsair Air 540 Gskill Ripjaws MX780 Bloody Defense Armor B-071 
  hide details  
Reply
Headshot
(15 items)
 
  
CPUMotherboardGraphicsRAM
Intel Core I7 2600K@4.7ghz Tb 5.0ghz  Asus Z77 Sabertooth Zotac GTX 1070 ZT-P107001-10P (16gb) GSkill Ares DDR3 2400mhz @2133mhz 
Hard DriveHard DriveOptical DriveCooling
Adata 240gb SP550 SSD WD Blue WD10EZEX 1TB LG BD-ROM Thermalright Venomous X 
OSMonitorKeyboardPower
Windows 10 Home 64 bit Asus MG248Q 144Hz 1920x1080 Gamdias Ares Cougar GX G1050w 
CaseMouseMouse Pad
Corsair Air 540 Gskill Ripjaws MX780 Bloody Defense Armor B-071 
  hide details  
Reply
post #3 of 4
Thread Starter 
That's what I feared.

This is really doing me in, though.

Like right now, SLI disabled I can stress test in MSI Kombustor and not even break 40C on GPU 1. Enable SLI and boom, temperature rockets up to 60C during the first few seconds all the way up to 75-80C, but only on GPU 1. Shouldn't the two cores sharing the load have a positive effect on temperatures? And if there's a hardware problem why does it only manifest itself during SLI?
Edited by onlinez - 11/30/13 at 6:11am
post #4 of 4
Thread Starter 
I said I had FPS problems while running SLI. Seems that I solved it by switching "SLI rendering mode" from "Force alternate frame rendering 1" to "Force alternate frame rendering 2", which seems weird to me. I had the impression that there wasn't much of a difference between the two? That they merely switch the order which frames are rendered in between the GPUs; 1, 3, 5 vs 2, 4, 6 or vice versa (sorry if I'm being unclear, I'm very tired).

I used this guide http://www.geeks3d.com/20101209/tips-how-to-enable-sli-and-crossfire-support-for-opengl-applications/ during some point which kinda illustrates what I was talking about above although you all know what the nvidia control panel looks like, I guess.

Anyways. I have no idea what SLI is doing to Google Chrome and Spotify. I've been trying to find out what they have in common to make them both flicker/ flash with black but have yet to be successful. Both Chrome and Spotify seem to display an FPS marker when Fraps is open, Spotify only under the "Discovery" tab, however. Does that narrow it down?

And the temperature. I know I'm repeating myself but how can GPU 1 be at 40C under load when running alone and then shoot up to 80C when running SLI under the same load? GPU 2 looks fine because it's at half the temperature at seemingly the same load and if the heatsink isn't seated properly shouldn't the temperatures on GPU 1 show that when I'm running it outside of SLI? Does this indicate a hardware problem at all? It doesn't make sense to me.

I've been googling my ass off for hours now and have yet to come across anything conclusive. If anyone can help me succesfully troubleshoot this I'd be willing to pay for their services rolleyes.gif Just PM me or post here.


EDIT: Hmmm.. It seems as if I got my cores mixed up. My apologies. The GPU that's active when SLI is disabled is indeed not the one that's "overheating". I got mixed up because they're called GPU 0 and GPU 1 aswell as GPU 1 and GPU 2 in MSI Kombustor.

SLI: http://i.imgur.com/1OpgJF7.png
NO SLI: http://i.imgur.com/CDKqdJe.png

CPU 2 working fine. CPU 1 overheating.

Can I conclude with this being an assembly error?
Edited by onlinez - 11/30/13 at 12:48pm
New Posts  All Forums:Forum Nav:
  Return Home
  Back to Forum: NVIDIA Cooling

Gear mentioned in this thread:

Overclock.net › Forums › Graphics Cards › NVIDIA › NVIDIA Cooling › Gainward GeForce GTX 690 temperature issue