Sli gaming temp issue

draterrojam

Distinguished
Jan 4, 2009
23
0
18,510
Just downloaded Two Worlds 2 cause it was on sale, but I noticed while running that my bottom card is about 20c than my top card. Doesn't seem to be this way with other games. Is there something wrong with this game? My set up? Please let me know.
 

jb6684

Distinguished
I have found that when running SLI or CrossFire that the two (2) cards Always run at different temperatures underload. This difference in temperature is generally 10C to 15C.

This is true in each of these example system I own:
- Corsair 800D case, dual Sapphire Vapor-X 5870's
- Antec P183 case, dual Sapphire Vapor-X 5770's
- Lain Li case, dual Sapphire Vapor-X 5770's (just tried this out, then moved 'em back to the Antec case)
- HAF-X case, side and two top 230mm fans, one front 200mm fan, MSI Talon Attack GTX460's


The ONLY solution I found that reduces this temperature difference between cards it to have a motherboard which allows the two cards to be separated by three (3) open slots. In this example I have one card in the Blue PCI-E closest to the CPU. And the second GPU in the White PCI-E slot, 3 open slots away....

http://www.newegg.com/Product/Product.aspx?Item=N82E16813131614

Using both Dark Blue PCI-E's, with single open slot between them one card runs Hotter.

I also have this MB with two (2) open slots between cards, same thing one card runs Hotter. And its in the HAF-X case so the ventilation is as good as it could possibly get....

http://www.newegg.com/Product/Product.aspx?Item=N82E16813131403R


So, yes, in my experience a temperature difference between cards in SLI or Crossfire under load is NORMAL....
 

My GTX560's only have a 1c difference between the top and bottom cards and the 8800GT's that were in the same rig beforehand only had a 3c difference between them.
 

jb6684

Distinguished
Ah, are the two 560's in the Black PCI-E slots? (that would make the spacing three (3) empty slots apart, the point at which I found that cards will run basically at the same temperatures....top & bottom)

http://www.newegg.com/Product/Product.aspx?Item=N82E16813127020

If the cards are in Black closest to CPU and the Blue PCI-E slot that would be only one (1) empty slot apart. In that situation I'd wonder why the temps were not more different top vs bottom.

The GPUs are very similar in cooler design to my MSI cards

http://www.newegg.com/Product/Product.aspx?Item=N82E16814261099

VS

http://www.newegg.com/Product/Product.aspx?Item=N82E16814127534


 


I still wonder why you chose Palit when you consider their quality and lack of cooling for the power vrm and vram. Last night I spent a few hours making a mod for my gtx 460. A backplate of sorts but need to get better thermal pads though :(
 

I know of several users of Palit cards and they have had no issues with them what so ever, I'm sorry if you have had a bad experience with them but I'm not going to base my buying decision on that.
 


Some of their cards have been good in the past but the gtx460/560 are different. Most manufactures have cut corners and compromised quality leaving it up to the consumer to pick up the costs later on. Without power vrm cooling and vram cooling these cards will typically have a much shorter life span than what would be normally expected. 160-200w without cooling for the power vrm is basically suicide. Prior to the Fermi era such cards that had any ware near that level of consumption was equipped with cooling in that era. The cores may be cool but the power vrm on gtx 460 and uncooled gtx 560 have been documented to reach up to 153c with 100c being the norm. Mosfets don't last for ever at those temps and is one killer or lcd monitors that have uncooled power supplies. You will likely get two to three thousand hours of use or more before problems begin to surface unless you know what they are and act before things get to be serious. The cards with the mid plate that have the thermal pads properly applied will fair better despite the higher core temps the who card overall will fair better in the long term. Without power vrm cooling and vram cooling much more air has to be pushed through the block to the pcb than normally would have been needed.
 

The 560's I went for are a non reference PCB design and as well as being shorter are almost entirely covered by the HSF, yes they may overheat and fail but there has been no sign of that so far.
 


That is how the gtx 460 is, it may be covered but check for your self during load by either feeling with your hand or using an infra red thermometer. The vrm is ware the hot spot is going to be at it's worst. Keep it cool and they will hold up for a year or two or perhaps longer but not guaranteed that they will last. At least you didn't buy Gigabyte as I have done. :lol:
 

I didn't buy the Gigabyte ones as they were more expensive and also seem to feature quite a lot in various forum posts about a 'Black screen' issue.
 


I have already ran across that one as well, its the unstable stock oc. Users have been able to stabilize their cards by either higher volts or the usual downclocking.
 

Yes but I don't see the point in buying an OC'd card that I might have to downclock to have it run stable, easier to just buy something else.
 


That is why I am not buying any more from that brand. Personally if I do get a gtx 560 ti it will be a full reference card with the mid plate. I did a lot of research on this card. Most top out around 930 to 950 with a few going to 1ghz but no higher than 1050 on the core and that is with max volts. Power consumption at those clocks and volts often leave the card drawing a few watts less than a stock gtx 570.