big temp difference between identical cards?

[trouserless]

I picked up a couple of used 5970's recently and have noticed that one card runs (on average) 10 degrees Celsius hotter than the other (58 vs 68).  I've had both cards in one open miner, and currently have them in two separate open miners sitting side by side in my basement and no difference.  I've kept the fanspeed the same and the clocks are identical.

I know when fabricating chips, parts are sorted in "bins" depicting the best speed that they will run at, but is it normal for a 10 degree difference from otherwise identical parts? (they even hash at the same rate).  Is this variance common?

My 5870's are the same temp, have the same settings and hash at the same rate...
  

[Photon939]

It's probably just a minor difference in thermal paste, a small gap in the paste or slightly off heatsink can significantly affect temperatures.

If the thermal paste is still stock you could probably lower them a bit if you replaced the paste with some decent stuff.

[cuz0882]

I agree, its probley the thermal paste. I always replace that stuff myself now to make sure. It does not take much to increase the temp by 10 degrees though. I'm assuming one is not blowing air towards the other. I have 15x 5970s running now and they all pretty much run the same except having them sharing a motherboard is obviously going to cause one to get a lot hotter, taking in hot air from the other. I would say a 5 degree difference between actual cards is to be expected, but even that difference is probably linked to the paste to some degree.

[trouserless]

I agree, its probley the thermal paste. I always replace that stuff myself now to make sure. It does not take much to increase the temp by 10 degrees though. I'm assuming one is not blowing air towards the other. I have 15x 5970s running now and they all pretty much run the same except having them sharing a motherboard is obviously going to cause one to get a lot hotter, taking in hot air from the other. I would say a 5 degree difference between actual cards is to be expected, but even that difference is probably linked to the paste to some degree.

You are correct, they are both side by side in separate rigs, drawing cool air from the back, blowing hot air into the center of the basement, no appreciable difference in their configuration/orientation.  I looked around this forum for tips on replacing the thermal grease/paste and it looks like it's a matter of disassembling the heatsinks, remove old paste (scrape clean, swab with cleaner) and apply a small blob of expensive toothpaste then re-assemble.  I could only find howto links for a CPU job:

http://www.hardwaresecrets.com/article/How-To-Correctly-Apply-Thermal-Grease/274/1

Any pics, tricks to keep in mind?  I'm good with hardware, but taking apart an otherwise functioning $300+ piece of hardware gives me pause.

[cuz0882]

It's pretty simple but there are videos of it being done on youtube. Also I do get a large difference in temps depending on what case I'm using. HafX case gives me some terrible temps despite its name. I have replaced the thermal paste on over a dozen cards and never had any trouble damaging the cards. So long as your careful, just don't over apply the paste. I usually open the card back up and make sure the full gpu is being covered and that no extra paste is getting on anything.