Any PC Builders Out There?

Rumors of Nvidia's next gen Ampere GPUs releasing on Sept 17

"Prices of current-gen enthusiast graphics cards are only going to rise as stock gets thin towards the launch of the Nvidia Ampere RTX 3000-series cards, and you're going to feel real foolish spending over the odds if the September 17 Ampere release date is confirmed next month."

https://www.pcgamer.com/nvidia-enthusiast-turing-graphics-card-discontinued/

 

They've created a unified portfolio around Threadripper and Epyc now that doesn't have any significant holes against Intels current offerings. They have a a targeted and well thought out solution for the HEDT, Workstation, and Server market at every level. It's quite a change from only three or so years ago.

 

So I hadn't much noticed what my temps were on my 1080's. Zero problems in any game.
But I happen to notice both were running about 50 deg at idle.

I was like...woah, what's running in the background that's using my gpus? Couldn't find anything.
But I do remember upping my voltage during some OC'ng after I got my new h100 fans in.

Looked at my Scanner X or whatever the OC GeForce prog was and it showed no voltage boost.

Anyway, watched a YT vid and it was super easy to take apart for replacing TIM.

No worries, both apart in just a few mins, cleaned and new tp applied to both.

One card(upper card that display port is connected to) now reads 31 at idle and the lower card reads 50 still.

The only thing I did was swap cards bringing the bottom card up top and the top card is now bottom.
The reason I did this because I figured the top card was the one being utilized the most and the bottom was barely being used in SLI.

So what's up?
I applied the TP in the pea fashion and spread evenly and thin with a plastic blade. Almost identically.

Should I take them out and redo both again?

What about "lapping"? Is that still a thing?
The surfaces on the heatsinks are smooth, but they look like they can be polished out quite a bit.

 

So SLI doesn't work with Windows OS. That is to say whatever Windows can use GPU hardware acceleration on will only utilize one card. So it's completely normal for whichever card is the "master" card to run a bit hotter than the "slave" card when doing things like watching YouTube or even browsing around in Chrome. I'm guessing the 50c card is the master, swapping card slots won't matter because once the card is the master the BIOS will always pick it by ID as the master, and the 31c card is just sitting around at idle. A good test would be to load a game or something that uses SLI and monitor the temperatures with HWMonitor. If the cards both level off at the same temperature, or with whichever card is on top being slightly warmer, it's a pretty good bet that nothing is wrong.

I don't recommend the pea and spread method anymore. It's not bad but funny enough you often don't end up with great even coverage after pressure is applied. Just leave a Pea or X and squish it with the heatsink next time. Wiggle it a bit and you'd be surprised how even the application will end up.

Lapping is still at thing but only for very special purposes. Like if you were going to do LN2 overclocking or you wanted to try to run a 10900K at 5.4 Ghz daily then you would look at lapping because you need every little advantage. For normal use lapping is completely pointless. If you're really paranoid about coverage and even contact you could use a liquid metal thermal paste but that is completely irreversible and can easily destroy your GPU if applied incorrectly.

 

Cool.
I do use HWM. And I think "Asseto Corsa" uses SLI.
I'll do some tests this weekend.

I do distinctly remember when I 1st installed these that one card always ran like 5 degrees higher.
They were like 32 and 36,37 when I 1st bought them.

After checking a few forums, 50 at idle is high.
I'm searching some old OC and build pics. Hopefully I have some screenshots somewhere.

 

Did you clean the radiator fins?

 

I did, yes.
Did a full clean on everything.

 

Can you look at something like Precision X or MSI Afterburner to see if one card is clocked higher when idling on the desktop? On my setup (RTX2060) I have a 144Hz 1440p primary display and a 60Hz 1080p secondary display. When I'm sitting at my desktop my card idles at 1350Mhz and runs close to 35C. If I Win+P and disable my 60Hz monitor, clocks drop to 400Mhz and the card cools down to 28C. Supposedly it's a bug with different refresh rate monitors. May be the same issue you're having?

 

Yea. I use precision X.
I have also read something to effect of what you're saying with multiple monitors raising the temps.
Also even just running 144hz on a 1440 display.
(Which is what I run)

I did drop the hz back to 60 and saw no change.

What I plan to do this weekend is remove both cards and install a single card in the upper slot. (Asus Z87 sabertooth board) (as I think the bottom slot closest to the ps is why one card is running hotter.)

Then remove that card and install the other card in the same spot and record temps.

Then I plan to do the same thing with both cards on the lower slot.

I also just noticed that I only have one fan pulling air in. The one in the lower front.
The large fan on the side, the one fan on the rear and both fans on the top on the rad are all pushing air from the inside, to the outside.

Which direction is the very large fan on the side supposed to go?

Again, I have no issues with any super high readings, artifacting, tearing or anything.
It's just an observation of the large difference in the temps. (20 degs).

 

I'll have to check and see if I can bring up 2 instances of precision X at the same time.
You can click on each vid card and monitor temps.

The evga scanner x prog, you can pull up 2 gpu monitors to monitor both cards.

It looks like in precision X, Card 2 (bottom card?)is not throttling down and that's what's causing the high temp??

 

How close is that card to the PSU? There should be at least a card and a half worth of gap (1.5 slots) between cards or else the fans will start to recycle hot air.

If it's a side fan like I'm thinking about it's supposed to push fresh air into the cards. That being said I know folks reverse them sometimes since most side fans don't have a filter and end up causing dust issues.

 

Look in Windows Task manager and see what's using GPU resources. That card should idle during everything except gaming and media tasks. Hopefully you don't have some bitcoin miner Malware or something.

 

Interesting. Although this guys temps aren't near as dramatic as mine, it's being said that precision X is reading the cards backwards and that the top card is actually the one running hotter.

Says something about switching the display port to the bottom card. Also turning the fan on one card to see which card is which to see if precision X ever fixed the issue.

Also, I'm going to turn the huge side case fan (idk the MM but it's yuge? 480?) To blow air from the outside, into the case.

 

I did check that the other day and saw no resources using the gpu. Also double checked and the voltage reads a bit higher on that card for some reason.
650 compared to the colder card of 625.
But they both are set for default settings.

 

This might be a PITA, but some forums suggest uninstalling GeForce Experience, all of your monitoring software, and reinstalling the card drivers. Then restart and reinstall whatever you need again.

 

I'll try that this weekend after trying each individual cards in both slots.

Ok,
It does turn out Precision X is wrong.
I clicked the chain to separate the cards, Turned up the fan on card #2 and its the top card thats actually running full blast.
(Stuck my fingers in between the 2 cards)

While the card on the bottom closest to the PS is barely running.

So card #2 is the top card.
Card #1 is the bottom card.

 

Also, the large fan is already blowing in as shown by this piece of cotton.

And there's plenty of room between the PSU and the bottom card #1. Which is running much cooler than the top card.

 

Lastly, with all fans on max, the case itself has more than adequate cooling. Ambient temp in the house is currently 74 degs.

Anyway, looking forward to messing with pulling the cards and testing this weekend.
It looks like it's isolated to card #1 Not idling correctly. Something is causing stress on it.

Thanks everyone!

 

is that hot lol? my 2070 idles at like 40 all day

 

I mean 50 is from what I've read.
There should really only be a 5 deg discrepancy between both cards.

The issue is, is why is one card not idling, causing the 20 deg difference.