Sign in with
Sign up | Sign in
Your question
Closed

1 Card heating significantly more with Zotac GTX 550Ti AMP! Edition

Last response: in Graphics & Displays
Share
February 18, 2012 2:15:40 AM

I very recently began running two identical Zotac GTX 550Ti AMP! edition cards in SLI configuration. I've been running a single one for nearly a year, and decided, come tax time, to invest in another, as I'd been very pleased with it as a single card. Here's the problem.

The card in the primary (top) slot, is running at the normal temperatures I received as a single-card setup, 30c idling, and 81c under load. The secondary card (bottom, brand new) is idling at 68c, and under load, is reaching around 95-96c. This is also the one the Nvidia software has designated for PhysX processing. I have tried multiple solutions to attempt to improve airflow in the case, including completely removing the PSU from the case, and running with the side panel open, and this does not seem to affect the temperature. The fan is operating, and graphically, I don't see any hiccups in gameplay,

I'm out of ideas, and I can't seem to locate anything telling me if this is normal behavior for this setup. It seems a rather large disparity between cards. It may be the case that my Google-fu has failed me, if so, I apologize. Any insight into the matter is appreciated.

Athlon II X4 3.0Ghz
Gigabyte 990XA-UD3
12GB 1333 DDR3
Zotac GTX 550Ti AMP! edition 2-way SLI
600W Thermaltake TR2 PSU
February 18, 2012 2:21:38 AM

Damn that is very hot. You can definitely ruin the card easily at that temp. What is weird is that the bottom card is hotter than the top. Usually the top one would be running hotter as it pulls in the hot air that the bottom card expels. Can you make sure the bottom card's fan is running properly. Also, check the fan speeds each card is running at.
Score
0
a b Î Nvidia
February 18, 2012 2:26:42 AM

Interesting. Did you ran the new GPU in single configuration?
Score
0
Related resources
February 18, 2012 2:29:23 AM

saint19 said:
Interesting. Did you ran the new GPU in single configuration?

That is a good idea too. If it runs with problems unlike your older card then it might be a defect with the card. In that case you should opt for RMA.
Score
0
a b Î Nvidia
February 18, 2012 2:30:20 AM

Try this out, check the temps, post GPUz screenshots here before and after, interchanging the cards.
Run the lower card in the upper slot and the upper card in the lower slot.
Remember to open 2 copies of GPUz side by side and select the second card in the second instance of GPUZ running before taking the screenshot.
Score
0
February 18, 2012 2:30:59 AM

Idle, Fan speed for top card is 1290@29c, Bottom card is 1590@68c.
Top card, under load, 2400@83c, Bottom card is 4200@96c.

Fan on the bottom is indeed speeding up with the temperature.

Edit: Doing card swaps, checking some BIOS settings, and getting GPUz Screenies now.
Score
0
February 18, 2012 2:33:41 AM

Crash0073 said:
Idle, Fan speed for top card is 1290@29c, Bottom card is 1590@68c.
Top card, under load, 2400@83c, Bottom card is 4200@96c.

Fan on the bottom is indeed speeding up with the temperature.

Edit: Doing card swaps, checking some BIOS settings, and getting GPUz Screenies now.

Your card's fans seem to be in order. It is more likely that their effectiveness, or the heatsink and/or thermal compound might be defective.
Score
0
February 18, 2012 3:43:51 AM

Original Problem config.
16x
8x

GPU positions swapped
16x
8x

New card running in Single-GPU configuration.
16x

And, back to SLI config.
16x
8x

The new card worked, still heated up more than the original. But when I placed the older card (confirmed to run at normal temperatures, in the 8x slot for SLI, I got a higher than normal temperature from it, and lower one from the new card. I have tried both configurations swapping out PSU rails, with no noticeable difference according to GPUz... Bios settings aren't indicating different power usages.
Score
0
February 18, 2012 3:50:23 AM

Maybe the 8x slot is not properly functioning with two cards.
Score
0
February 18, 2012 3:52:59 AM

Is your PSU working good? Maybe it is malfunctioning when a higher power is needed from it for SLI.
Score
0
February 18, 2012 3:59:41 AM

Motherboard specifies Nvidia SLI compatibility so if the 16x/8x thing is my issue, I'm going to be very upset at Gigabyte.

PSU is 2 months old, shows no signs of wear and tear, under warranty.

When I was checking on the requirements for a SLI system, the chart on Nvidia's SLI page indicated I needed at least 500w to run the GTX 550ti in SLI.

For the life of me, i can't figure it out.

I have had a marked increase in graphics processing, it doubled my 3dMark11 FPS. Everything appears to be functioning properly, and by reputation, the AMP series runs hot to start with. I am okay with higher than a non-overclocked 550ti, but the AMP edition comes factory OCed, and I haven't had any problems out of it before this.
Score
0
a b Î Nvidia
February 18, 2012 4:01:01 AM

Ok, you gave the wrong screens but you got the point basically.
The screens that I needed were to check whether both the cards were working in 8X or 61x, or one at 8x and he other at 16x.
I had that problem once.
Your card need to be functioning both at 16x.
The screens needed were the first page screens of GPUz.
Score
0
a b Î Nvidia
February 18, 2012 4:01:48 AM

Now open GPUz and check both the firmware versions and revisions of the cards... I have a feeling they are different and not the same.
Score
0
February 18, 2012 4:05:07 AM

alyoshka said:
Ok, you gave the wrong screens but you got the point basically.
The screens that I needed were to check whether both the cards were working in 8X or 61x, or one at 8x and he other at 16x.
I had that problem once.
Your card need to be functioning both at 16x.
The screens needed were the first page screens of GPUz.

Yeah that was my initial thought although the cards should theoretically run fine even though they are each on different slots, you may only see a slight performance drop. However, it can also be possible that the heating issue may have to do with this though it's unlikely. :??: 
Score
0
a b Î Nvidia
February 18, 2012 4:06:07 AM

let the op answer, we're not interested in what you thought.
Score
0
February 18, 2012 4:06:09 AM

If you have another PC available then try the cards on that one.
Score
0
February 18, 2012 4:09:54 AM

GPU tab of GPUz.
16x
8x

Okay, it's showing the bus interface on my second slot as PCI-E 1.1x16.... That's my problem, isn't it...

I do not have another PC capable of testing the cards unfortunately.
Score
0
a b Î Nvidia
February 18, 2012 4:10:31 AM

Akhilcool said:
If you have another PC available then try the cards on that one.

For gods sake stop giving stupid advise. The OP will end up messing things further. These aren't dead cards, this is a mismatch issue which can be set right without increasing the problems further. And everyone doesn't have 'other rigs' lying around to test things out. People here, generally can't pay for services to troubleshoot their rigs, please keep that in mind. There are other enthusiat section where you can 'discuss' your thoughts and other fantasies... not while solving problems.
Score
0
February 18, 2012 4:11:24 AM

alyoshka said:
let the op answer, we're not interested in what you thought.

I'm helping him understand what running the cards in different slots can do if he already doesn't know. ;) 
Score
0
a b Î Nvidia
February 18, 2012 4:12:58 AM

Yes, that is your problem. You have stuck the card in the wrong slot.
To top it all , that's the card that has been reserved for PhysX calculations too.
This is your board right?
Gigabyte - 990FXA-UD3
Score
0
February 18, 2012 4:14:04 AM

alyoshka said:
For gods sake stop giving stupid advise. The OP will end up messing things further. These aren't dead cards, this is a mismatch issue which can be set right without increasing the problems further. And everyone doesn't have 'other rigs' lying around to test things out. People here, generally can't pay for services to troubleshoot their rigs, please keep that in mind. There are other enthusiat section where you can 'discuss' your thoughts and other fantasies... not while solving problems.

I said "if" he had another rig. Meaning that if he doesn't then he can say "NO" and if he does than it helps.
Score
0
February 18, 2012 4:14:04 AM

Hehehe.. Yeah, running the SLI at 8x was a "Wallet vs. Performance" issue, which should be pretty evident by doing a SLI rig with 2 550ti's as opposed to 4 580's.... lol

Yep, that's the board. The third PCI-E slot reads as a 4x though... Bios update maybe?
Score
0
a b Î Nvidia
February 18, 2012 4:18:38 AM

Akhilcool said:
I'm helping him understand what running the cards in different slots can do if he already doesn't know. ;) 

And you're helping him by guesswork?
Guess#1: "Can you make sure the bottom card's fan is running properly. Also, check the fan speeds each card is running at."
Guess#2: "It is more likely that their effectiveness, or the heatsink and/or thermal compound might be defective."
Guess#3: "Maybe the 8x slot is not properly functioning with two cards. "
Guess#4: "Is your PSU working good? Maybe it is malfunctioning when a higher power is needed from it for SLI.If you have another PC available then try the cards on that one."

And I don't know what else you could come up with....



Score
0
a b Î Nvidia
February 18, 2012 4:23:01 AM

Crash0073 said:
Hehehe.. Yeah, running the SLI at 8x was a "Wallet vs. Performance" issue, which should be pretty evident by doing a SLI rig with 2 550ti's as opposed to 4 580's.... lol

Yep, that's the board. The third PCI-E slot reads as a 4x though... Bios update maybe?


A BIOS update might be there but it's not required.
Score
0
February 18, 2012 4:23:42 AM

Troubleshooting is process of elimination... ;)  The render test is indeed showing the second slot running as PCI-E 1.1X16, even though the blurb on the manufacturer's website says...

1 x PCI Express x16 slot, running at x16 (PCIEX16) (Note 3)
1 x PCI Express x16 slot, running at x8 (PCIEX8) (Note 4)
1 x PCI Express x16 slot, running at x4 (PCIEX4)
2 x PCI Express x1 slots
(All PCI Express slots conform to the PCI Express 2.0 standard.)
2 x PCI slots

So... can I solve this problem, or am I destined to be a victim of not paying enough attention and deceptive marketing practices?
Score
0
February 18, 2012 4:25:20 AM

alyoshka said:
And you're helping him by guesswork?
Guess#1: "Can you make sure the bottom card's fan is running properly. Also, check the fan speeds each card is running at."
Guess#2: "It is more likely that their effectiveness, or the heatsink and/or thermal compound might be defective."
Guess#3: "Maybe the 8x slot is not properly functioning with two cards. "
Guess#4: "Is your PSU working good? Maybe it is malfunctioning when a higher power is needed from it for SLI.If you have another PC available then try the cards on that one."

And I don't know what else you could come up with....

You seem to be spending more time on me than on helping him with his problems. I am giving suggestions that may or may not help him but have a good possibility of fixing his problem. Most issues that can be addressed with common solutions. If not than we can get into the nitty gritty stuff but be a sport and focus on the problem at hand, you can work with me later. ;) 
Score
0

Best solution

a b Î Nvidia
February 18, 2012 4:25:59 AM

This is the where it should be.
Share
February 18, 2012 4:26:40 AM

Crash0073 said:
Troubleshooting is process of elimination... ;)  The render test is indeed showing the second slot running as PCI-E 1.1X16, even though the blurb on the manufacturer's website says...

1 x PCI Express x16 slot, running at x16 (PCIEX16) (Note 3)
1 x PCI Express x16 slot, running at x8 (PCIEX8) (Note 4)
1 x PCI Express x16 slot, running at x4 (PCIEX4)
2 x PCI Express x1 slots
(All PCI Express slots conform to the PCI Express 2.0 standard.)
2 x PCI slots

So... can I solve this problem, or am I destined to be a victim of not paying enough attention and deceptive marketing practices?

Correct and that was what I was trying to do until this veteran grandpa started worrying about me. :non: 
Score
0
February 18, 2012 4:29:05 AM

Best answer selected by Crash0073.
Score
0
February 18, 2012 4:30:21 AM

Well, you all have solved my problem it appears, I will fix this and get back to you in very short order with the results, thank you both very much for your time and effort, it is indeed appreciated!
Score
0
a b Î Nvidia
February 18, 2012 4:30:47 AM

Welcome
Score
0
February 18, 2012 4:57:18 AM

Yeah I think the BIOS is being faulty. It isn't properly designating the cards during SLI which is quite weird. It could have to do with the GPU drivers or the motherboard. Are both cards properly connected by molex cable to the PSU?
Score
0
February 18, 2012 5:12:05 AM

GPU Drivers are the latest available from Nvidia, Nope, not connected by Molex, but by PCI-E 6-slot PSU lines. ;)  Drivers are up to date, Just updated the BIOS. Google has reported that GPUz sometimes incorrectly reports system BUS specifications. Nvidia Control Panel is reporting that it's detecting both GPUs operating in PCIx8 Gen2 slots, so that should be correct, ruling out the improper PCI-E slot detection.

Le sigh... Back to the drawing board...
Score
0
February 18, 2012 5:14:46 AM

LOL sorry yes PCI-E 6 pin connector... uhm the last resort is wondering if the bottom card itself is just being faulty with SLI. But of course you wanna at least try anything else before you try to RMA.
Score
0
February 18, 2012 5:16:53 AM

Did you overclock the cards at all. Maybe one of the cards is running at a much higher core voltage. Even if you didn't mess with anything like that, consider trying to find out what voltages both cards are running at.
Score
0
February 18, 2012 5:31:26 AM

The cards are both factory OCed, and run hotter than normal versions of the same chipset.

http://www.legitreviews.com/article/1688/12/ Independent testing of wattage used by the AMP cards.

Just realized, the older card is running at a nice, comfortable .950v, the new one is running at 1.162v... Looking now for a way to stabilize the voltage down to the level of the original card.


Score
0
February 18, 2012 5:36:11 AM

Crash0073 said:
The cards are both factory OCed, and run hotter than normal versions of the same chipset.

http://www.legitreviews.com/article/1688/12/ Independent testing of wattage used by the AMP cards.

Just realized, the older card is running at a nice, comfortable .950v, the new one is running at 1.162v... Looking now for a way to stabilize the voltage down to the level of the original card.

Bingo! I think you've found the problem. Try using MSI Afterburner.
Score
0
February 18, 2012 5:37:12 AM

Although the reason why the card is at that high of a voltage is still a mystery.
Score
0
February 18, 2012 5:38:54 AM

BTW, 1.162V vs 0.950V is a big difference, I believe the max the cards should handle is 1.150V : / surprising that the cards haven't died out yet.
Score
0
February 18, 2012 5:47:48 AM

The newer card is fresh out of the box today. I like my systems to last and be consistent, so I do alot of testing when I first get a new component to see what its capabilities and limits are, it was during the testing that I found the problem, so it's literally got less than 16 hours of use, most idling and rebooting while I've been troubleshooting today.

Nvidia Inspector is showing that even though I'm not running any programs more complicated than my usual monitoring software, card 2 is pinned at 2000Mhz of shader usage, which might explain the voltage draw during idle... Off to find an answer for this question now...
Score
0
February 18, 2012 5:52:43 AM

I'm sure you can probably fix this issue now that it's been narrowed down. :) 
Score
0
February 18, 2012 5:54:38 AM

Good that you haven't done much with the cards because if you didn't monitor the temps the card would have been fried. Even still, if you can't fix it yourself, at least you can now be specific when asking for an RMA if you need one.
Score
0
a b Î Nvidia
February 18, 2012 6:13:10 AM

Try setting the clocks with this

And if that doesn't work you can always try the MSI Afterburner.
Score
0
February 18, 2012 6:36:05 AM

http://www.overclock.net/t/964370/fix-info-dual-monitor-downclocking-fix-for-nvidia-cards#post12719142

This link had a fix, and so far, it's working, Idle temps are down to 40c (still dropping) and 28c, which, considering they're in a mid-tower case, is alright by me. Getting ready to test them under load after a reboot and see how things work.

Another thing that dropped my temps, oddly enough, was plugging one monitor into each card, as opposed to both monitors on the primary card. Who knew, huh?

Have to say, for a company showboating their factory overclocking, Firestorm is a particularly unimpressive piece of software... lol I don't plan on pushing past the factory overclocking, so I'm not expecting to run anything other than the forced-state process in Nvidia Inspector. I'll let you guys know how it goes after some Skyrim... ;) 

Edit: Or not.. I just looked at the clock... I'll post the results of that tomorrow... lol.
Score
0
February 18, 2012 5:57:04 PM

Alrighty, Idle temps are under control. Resting at 36c and 32c respectively, nice and in the comfort zone.

During gameplay (Skyrim, High-Res texture pack running at High settings) It runs smooth as glass, one card in the low 80's, the other heating to the low 90's, peaking at 96c while rendering graphic-intensive zones (transitions with multiple sets of textures, areas with lard amounts of specular lighting). Looking for some kind of load-balancer to split the difference, as the cooler card is running at about 70% usage, the hotter one at 95% constantly. After extended gameplay, (2 hours+) I got flickering on some textures and objects. The card appears to be properly throttling itself back when it gets in that top-end range, but frame rate and rendering suffers.

Fallout: NV presents even more encouraging news. The less graphic-intensive game does not heat the cards as much, 73c and 86c, respectively. This tells me I've got the cards themselves under control, and now need to find a cooling solution. Once again, thanks for everything!
Score
0
February 18, 2012 8:53:41 PM

Crash0073 said:
Alrighty, Idle temps are under control. Resting at 36c and 32c respectively, nice and in the comfort zone.

During gameplay (Skyrim, High-Res texture pack running at High settings) It runs smooth as glass, one card in the low 80's, the other heating to the low 90's, peaking at 96c while rendering graphic-intensive zones (transitions with multiple sets of textures, areas with lard amounts of specular lighting). Looking for some kind of load-balancer to split the difference, as the cooler card is running at about 70% usage, the hotter one at 95% constantly. After extended gameplay, (2 hours+) I got flickering on some textures and objects. The card appears to be properly throttling itself back when it gets in that top-end range, but frame rate and rendering suffers.

Fallout: NV presents even more encouraging news. The less graphic-intensive game does not heat the cards as much, 73c and 86c, respectively. This tells me I've got the cards themselves under control, and now need to find a cooling solution. Once again, thanks for everything!

I'm happy to see it is going well for you now. The card that peaks 96*C is something I'd worry about. Hopefully an aftermarket cooler can fix that. Anyways good luck!
Score
0
April 30, 2013 3:46:54 AM

switch with better case or thermal paste should help actually
Score
0
!