Heat Issues With SLI

Hey,

I overclocked my GPUs for the first time last night, and everything seemed to run fine, but I noticed a significant different in the temps of my GPUs. These were the max temps for both my cards, the average was about 10C difference as well:

9800m GTX #1: 84C
9800m GTX #2: 67C

Is this common in SLI, does the main card usually produce way more heat than the secondary one?

 

Is the First card closer to the CPU or HDD? This I know can raise the temps by alot

 

when you OC'ed, was for both cards, or just one?

but in that system each videocard has its own heatsink and fan... so the temps should not be too different.

around 10-15C difference should not be an issue.

 

I OC'd both to 600/900/1550. It just seems a little weird that one is running so much hotter than the other.

@mrbutterbiscuits They appear to both be equal distance from the HDD, one might be closer to the CPU though. I don't have it with me at this moment, and I don't have the tools to take it apart with me (I actually haven't even looked at the hardware yet, I'm assuming I need special tools to take the casing apart like all my other laptops).

 

I would not worry about it... as long as the temps stay below 90C, you should be perfectly fine.

 

=GamingACU;4845221]Hey,

I overclocked my GPUs for the first time last night, and everything seemed to run fine, but I noticed a significant different in the temps of my GPUs. These were the max temps for both my cards, the average was about 10C difference as well:

9800m GTX #1: 84C
9800m GTX #2: 67C <---- this means your running your applications in single gpu mode. change it to one of these...sfr/afr/afr2. this will make the cards work together for most of the time.

you still have up to 98 degrees before you system starts a forced down clock
Is this common in SLI, does the main card usually produce way more heat than the secondary one? <--- this is because the master card is doing most of the work

 

Johnksss' answer sounds like the right one; however, given the relatively tight thermal budget constraints notebooks operate under (yes, even the fabled D901C is not as thermal-friendly as your average desktop), it's possible that overclocking your GPUs may be generating enough heat that slight differences in the amount of airflow over each GPU (due to the internal configuration of the various components) are becoming noticeable. Also, since each GPU has its own heatsink, your overclocking might be indicating a problem with the contact between the processor and the heatsink in the hotter GPU.

 

one tiny little problem.....this is something i tested for hours on end, trying to understand and make sure i had the facts right. so when anyone asking..i could help them in the right direction.

thanks shyster! i almost forgot
it matters allot on which driver you use. if your driver does not down clock it self after your done really using it....it's not the right driver for your system. it means the powermizer options are not working correctly.

right now, both my cards have down clocked themselves to 200/100/400 and the temps are riding at 42'c on one and 40c on the slave.

 

Thanks for the info Johnksss, but how do I go about changing the 2nd card to ..sfr/afr/afr2.

My current driver is the DOX 185.68. It runs really well, just from changing to that from my stock drivers I gained about 1k score in vantage.

Also, my GPU idles around 54 degrees, that doesn't seem to hot, should I worry about it downclocking itself properly or not?

 

Oh, I've no doubt that your answer is the right one; however, I still have to justify my existence, don't I? So I picked up on what might be a remote possibility, that the heatsink wasn't properly attached to the hotter processor.

 

never shyster!! your answers are always welcomed and very appreciated as far as im concerned! you just passing through to say hi is good enough justification for me..

pssssss...(whispering) i dont think he took it apart yet...shhhhhhhh....(whispering)


oh...hey gamingACU...didn't see you sitting there...lol

1: click nvidia control panel
2: click manage 3d settings
3: click sli performance mode
4: click drop down menu and play with the settings there....

 

Haha, it's ok.

Which drivers do you use. I figure since we're using about the same setup you'd know the best drivers for our rig.

Oh...yea I'm at work right now, I won't be messing with it for a few hours, so no, I didn't tear it apart yet.

 

well hit me up when you get home(even if you dont see my light on), so we can get you up to speed on what is what with these machines...i can't be the only one up this high...i need to have some other faithful np9262 owners up here as well....

 

Definitely, however I don't get home for another 4 hours. Don't know if you'll be up or not.

I'm willing to try (or at least learn) if you're willing to teach.

 

always worth a shot...

 

just switched to sfr, the max temp dropped to 81, but my p score also dropped by about 100

 

your p score?

 

The temps of my cards differ a little, but I've never seen over a 5C difference, even when OCed.

For reference: When I run Crysis (or any other very demanding game) on high fans (Fn+1) I get temps in the high 60's/low 70's, roughly what you got with GPU#2. Without high fans though, my temps get close to what you got on GPU#1.

This suggests that you might want to assure that both GPU fans are capable of switching to high mode.

You've run benchmarks, and if you're getting scores similar to comparable machines, your SLI is working OK. A quick way to check if you're really running in SLI mode is to turn on the "SLI Visual Indicator" in nVidia control panel and play a game.

 

vantage pscore.
Both the GPU and CPU dropped slightly. The gpu lost about 100 and the cpu lost about 2k.

Also, with fans on this significantly drops my temps, about 42C idle and 60-70C max.

 

Just an update, I installed the 185.85 official drivers and my temps have started to vary even more dramatically.

With the fan on low and running any intense game I max out around 90C on GPU #1 while only hitting high 60's with GPU #2.

With the fan on high GPU #1 maxes around 65C and GPU #2 maxes around 55C.

Think it's time to check the heat sinks/apply new thermal paste? This thing's only a week old though, it shouldn't have worn out that quick.

 

Have you cleaned out the inside of your system with compressed air lately?

 

No...It's brand new, and has been seeing temps like that since I got it.

 

Doesnt matter if you got it right off the factory floor, dust especially when it travels accumulates quickly. It only takes a week for stuff to get in there and cause problems. Take your bottom off of the machine and take a look. Spray it out and good you will be surprised what gets caught in there.

Then if its giving you problems then check the thremal paste. If you are in Irag it could have cracked due to environment. Just reapply.

 

Alright, pulled it apart, and the fans are clean and working fine.

So I went to check the thermal paste...and it turns out im a total noob when it comes to notebook hardware.
Where's the paste supposed to go/be? I took off the fan to the gpu and all there was, was a copper plate chilling there. Do I have to take that thing off to get to the actual gpu, and is the paste supposed to go on the copper plate, or is there something else to apply it to?

edit: another odd thing I just noticed was that GPU #2 is slightly hotter during startup, but gpu #1 is always at least moderately hotter during idle/gameplay.

 

i would be very carefully pasting the gpu...if you get it on the memory chips or components..it will render the card useless....