SLI 1080ti Temp Difference

[liljamerz]

Here are the specs of my setup:

ASUS ROG MAXIMUS IX HERO
i7 7700k @ 5.0ghz (1.31v)
32gb Corsair Vengeance LED
EVGA GTX 1080ti FTW3 x 2 (ROG HB SLI Bridge)
960 EVO m.2 (500gb)
960 PRO m.2 (512gb)
860 EVO SSD (256gb)

The CPU and both GPUs are in a water loop, all parts are EKWB. Top radiator is 360mm and front is 280mm.

https://imgur.com/a/tvzlLmv (Picture of setup)

My question is about differences in temps between the video cards. My bottom card runs 4-6C hotter than the top card under heavy loads, even though the OC settings are synced with Afterburner. The temps never exceed 50-55C, so I cant complain there, but when my bottom card is 50C and the top is floating around 43-44C, I get a little concerned.

They are hooked together with the Terminal X parallel block, both have the same gpu blocks, both are same model of cards, both have same OC settings. Even when I put them on factory clock speeds, they will separate by 3-4C.

I removed the bottom card last night, took off the block, cleaned the paste, reseated the block with new compound (ekwb ectotherm), and got the same results. (so much work for no results lol)

Is this normal? I watched a video by Jayztwocents and he had 3 cards in parallel and all within 1-2C of each other. Maybe my top card just won the silicon lottery and the bottom one didn't? I know this isn't a huge gap by any means, but this is the first parallel GPU loop I've done, and want to make sure this isn't an issue.

Thank you for your help!

 

[jay32267]

I wouldn't think it is an issue that is going to cause you problems down the road...as it's only a few degrees.

What you have to consider....is that I don't think it would be uncommon to find that there is a 1 or 2 degree difference in the sensors themselves. Those sensors are cheap and aren't that precise.

So lets say you have two degrees or error here. Then one card runs two degrees hotter....so now you have 4 degrees.

Then you have error in actually reading the voltage from the sensor. This can add to the 4 degrees.

So right there....it's pretty easy to see how you can be off.

 

[Eximo]

Sensor difference could explain it.

Also just your typical silicon differences. Might just have a warmer than normal chip.

Anything below 60C is going to be just fine for Pascal under water.

I think if I saturate my loop I can only get the GPU up to 55C. And it looks like you have a lot more radiator, which makes sense with SLI.

 

[Killer01ws6]

My two 1080TI's show about 4 deg difference also. I am running
Asus - GeForce GTX 1080 Ti 11GB STRIX GAMING OC and they are 2.5 slot width cards... I thought that might be some of it, but I think Eximo might be on track with the sensor difference also.

 

[sla70r]

Have you tried switching the placement of the two cards ( swapping them) and then seeing if the pattern holds? Just curious really...

 

[Killer01ws6]

"Have you tried switching the placement of the two cards ( swapping them) and then seeing if the pattern holds? Just curious really..."
I started to ask him that also sla70r, but I know with his WC loop that is a much harder affair than us air-cooled GPU guys.