Hey. I have two GTX 480 hydro coppers running in SLI on an E770 board. Current loop is :
655 pump -> XSPC RX360 rad -> EK Supreme HF CU block -> GTX 480 #1 -> GTX 480 #2 -> reservior
One of the 480's (the first one in the loop) also has a backplate attached.
 
What is weird is that under load, the second 480 runs much hotter than the first. Is this normal?
 
I realize that it should run maybe a little hotter as it's getting water that has already been heated passing through the first 480, and on idle it's about 4°C hotter usually, but under full load that card #2 can run up to 20-25°C hotter than card #1, which just seems way too high.
 
As soon as the load is removed, both cards drop back down to near idle temps in a matter of seconds (idle temps are in the low 30°C range, load will show mid 40°C on card #1, while card #2 can get up to mid 60°C).
 
I am using 1/2" ID, 3/4" OD tubing on the loop, and a DangerDen SLI fitting between the cards, which has a smaller ID than the tubing.. could that be the issue?
 
As mentioned, the card #1 also has a backplate attached, but surely a backplate wouldn't make that much difference!
 
I have not tried swapping the cards around yet, which I guess would be the first thing I should try, but thought I'd ask first before draining the loop etc..
 
Appreciate any comments.

Well that is interesting. I have AIDA64 (used to be EVEREST).. the readings on there are ..unusual.
 
On idle, GPU 1 & 2 both show 0.963V
Likewise, GPU Vcc and GPU VRM voltage are about the same, however...
 
GPU VRM Current is 10A for card#1, 14A for card#2
GPU VRM Power is 9.5W for card#1, 13.25W for card#2
 
Is this what would be causing the difference?
How can I even these out... not sure if this is card BIOS or something to do with the PCIe power or what..
 

Nereus

 
Just installed EVGA precision... the GPU BIOS versions are different..

card #1: 70.00.21.00.82
card #2: 70.00.35.00.80

Using driver 266.58 btw.

That could be a good start point. My system is down right now so I'm going off of memory, but if you download EVGA OC Scanner I'm pretty sure you can manage the voltages with that. The temps on your second card seem WAY high. I'm not going to be much help as far as the BIOS goes because I have never messed with that on my cards.
Did you buy the cards new or used?
 

I would definitely give Jacob a PM about the BIOS. I would be willing to bet that is causing the difference in temps.
I say the second card is running WAY to high because my cards never go over 50c. Just seems to be something wrong there. I know ambient temps and loop etc. play a role but 20c + is a big variance as you  stated.
Download OC scanner tool and see what that gives you for voltages.
http://www.evga.com/articles/00530/

Yup, definitely different:
 
card #1: 1.04375 V
card #2: 1.0375 V
 
wait, but that says the 2nd card is lower Voltage?
 
unless it has them the other way around..
 
I'll PM Jacob tomorrow. Need some sleep now. Thanks heaps for your help!
 

Thanks, I'll mention that to Jacob too.

 
Just spent the last few hours draining the loop and taking the cards out and redoing the tim. I found from the previous (and original) owner that the tim was all original on both cards, as was the BIOS versions.
 
Turns out the card that was running hot had absolutely no tim or thermal padding on the mosfets whatsoever. I put MX4 on the GPUs, 1.0mm padding on the RAM, and 0.5mm on the mosfet.
 
On reinstalling the cards, I also swapped them around, put the cooling in parallel instead of series, and used compression fittings and 1/2" ID tubing between the two cards, instead of the single narrow in-series DangerDen SLI kit I had on there previously. 
 
The flow has definitely increased, judging by the turbulence going on in the reservoir. The temperature on the 2nd card (now first on the loop) is still higher though. It's about 3°C higher on idle (although it theoretically should be maybe a couple of °C cooler), and on load it's about 10°C hotter than the other card, which is still an improvement over the 20-25°C hotter it was previously, but still does not make much sense when it is first in the loop.
 
The GPU VRM current (amps) and power (watts) are still higher on that hotter card though, which I am hoping has something to do with the different BIOS versions.
 
At least there is progress :) 
 

No problem BRC - I'm waiting on Jacob to reply to another IM, and if I can flash to same BIOS, I'll see if that changes anything. Only one card was missing anything (the hotter one), and it was only missing tim/pads on the mosfets on the far right (or left, depending on your angle).. the GPU and main RAM modules around the GPU had tim/padding ok on both. That amp & watts variance is weird though, but I'm guessing it's the BIOS.. at least I'm hoping that's what it is, lol.
 
I'll get the serial numbers tomorrow when there's better light to see without pulling the cards, but hoping an RMA isn't required. Oh, removed the backplate on the other card - hasn't made any significant difference to the temp though.
 
Running EVGA OC Scanner at present.. the (now) first card in the loop is sitting on 55°C, the other at 46°C. Still way better than it was (had it hitting 70°C earlier before I redid the tim).
 
Thanks for keeping an eye on this :)

OK got latest BIOS from Jacob and flashed both cards. Both cards are now running BIOS version 70.00.35.00.80
 
Have swapped cards around so the hotter card is now first in loop.
 
Both cards set at 1.0375 VGPU in ELEET.
 
STILL on idle, using AIDA64 (ex EVEREST), GPU 1 & 2 both show similar Voltage, likewise, GPU Vcc and GPU VRM voltage are about the same, however, again the GPU VRM Current is 14A for the hotter card, 10A for the cooler card, and GPU VRM Power is 13.25W for the hotter card, 9.5W for the cooler card.
 
So this is not a mobo (PCIE) issue, nor is it a PSU issue as I tried swapping plugs. This only leaves the issue with the card itself.
 
I redid TIM on both cards which helped a little but not much. Tried water loop in series and parallel. Didn't make a huge difference, although getting rid of the DangerDen SLI fitting and reverting to just tubing and compression fittings improved flowrate.
 
One thing I noticed is that going from parallel to series decreased flowrate, so possibly that hotter card has a bit of gunk inside it? That could explain some things, but not the difference in amps and watts.
 
Idle now is still about 4-5°C difference, and running OC scanner for a while, the 'good' card levels out around 44°C, which is reasonable, but the other card, despite being first in the loop, climbs up to nearly 60°C within a few minutes and then levels out.. that's still a 10-15°C difference between the 2 cards on load, which is just not really acceptable for a water-cooled card on stock settings.
 
The only other thing I can think of now is to open up the block itself and see if there's some gunk inside perhaps causing a problem, but not sure if this would void warranty.
 
Noticed something else.. the PCB board on the edge on the hotter card is a black color, while the PCB board on the cooler card is a red-orange color.. so they're made from different material?
 
Would appreciate advice from EVGA staff here.. should I open up the hotter card, or RMA it at this point? Is the different color PCB a factor here? And what's with the amp and wattage being 40% higher on the hotter card on idle?..or are temp differences like this on a single loop within tolerance and I should just live with it? Both cards are on stock FTW settings btw.
 
  Just played MoH on high, x4 aa.. sitting 70°C on the hot card, the other was around 46°C.

Thanks. Have EK Supreme HF CU and 655 pump too. Using RX360 rad. Have a full board block, but not installed yet - want to get cards sorted out first. Just played MoH and was hitting 70°C on the hot card (first in loop) and about 46-47°C on other card. Cards at stock FTW clocks, MoH set on high with 4xaa only.

So I had kind of the same issue after RMA'ing one of my GTX480's as the one I got back as the RMA had a default voltage of 1.075v vs. 1.036. So the RMA card runs hotter for me as well. Installed AIDA64 and compared readings to yours and getting:
 
Idle:
RMA card (1.075v default) = 11.50-12.0A - 10.75-11.25w
Card 2 (1.036v default) = 11.0A - 10.50w
 
Idle temps are 33 and 34*C right now for me. Under load the hotter card will be about 4*C higher the card 2, ~45-50*C. This is after changing to parallel instead of series loop set-up. Had bigger differences in series, closer to 7-8*C.
 
I would think about RMA'ing card due to heat concerns. 70*C with your set-up is way to high compared to 45-46*C of your second card.... Just my two cents though...

mine are always within 3 or 4 deg. of each other....they do run warmer than most ....but thats because of a terrible rad choice that i have been meaning to replace....

Nereus

Well that is interesting. I have AIDA64 (used to be EVEREST).. the readings on there are ..unusual.

On idle, GPU 1 & 2 both show 0.963V
Likewise, GPU Vcc and GPU VRM voltage are about the same, however...

GPU VRM Current is 10A for card#1, 14A for card#2
GPU VRM Power is 9.5W for card#1, 13.25W for card#2

Is this what would be causing the difference?
How can I even these out... not sure if this is card BIOS or something to do with the PCIe power or what..

 
Your temps for the "hot" card seem way to high, assuming you are not clocking at 950-1GHz, esp on water.
 
The power levels don't seem out of place for power at IDLE. What are these power levels under load, where you really see the high temps? 10-15W cannot be power under load--should be more like 200W+.
 
You may have bypassing (or air?) in one waterblock. Since you have already done TIM, cannot be that.
 
One thing you could try is run the hot card by itself as a single in you loop and see what it runs at. Or, maybe swap waterblocks and see if the problem shifts with the waterblock or stays with the card.
 

 
Nods, that's the one thing I have not tried (one card only at a time) besides swapping physical blocks between the cards. I've messaged Jacob about 3 times asking if I can open the waterblock up and check for blockage without voiding the warranty, and he keeps asking me if I tried X or Y or Z which I've already stated I've tried.. kinda getting a little frustrated, but I guess he has a million IM's to clear every day, so probably doesn't read all of the content in them.
 
If I haven't heard anything by the weekend, I'll try running one block at a time and see if the flowrate changes significantly between the two.. I don't see what else it can be other than faulty, but the person I got it from (on this forum) is reliable and didn't have an issue to his knowledge, so blockage of some sort seems likely.
 
If any EVGA staff reading this can advise whether opening the block will void the warranty or not, it would be appreciated. Saves the cost of an RMA for me and for EVGA....
 
Thanks.

Any news on this? & sorry for jumping on this thread
I'm having the same problem with a 470 now.
 
Everything was running fine then when folding the system froze/rebooted & 1 card has not been the same since.
As you can see i'm running EK blocks.
Underload first gpu now reaches 100 on any game or folding within minutes.
Things i've tried:
unclocked
Clean drivers & Clean installed W7
Swapped card pci slots - always same card that gets hot
Re TIM card
cleaned on loop (run a single for all)
 
driving me mad now.. lol
any ideas anyone?
 
Thanks