Gtx 1080ti Crashing in Games at Stock Clocks

[Lord_Karango17]

Hi guys,

 

TL;DR: Recently bought a lightly used 1080ti. Put it in my pc after doing a clean driver install only to have it CTD in games at stock settings. Lowering the power limit to ~85% seems to make it far more stable in games however it still behaves a bit odd (occasionally hangs for a frame or 2).

 

Its been a couple of years since I posted here, mostly because my PC build had stayed pretty much the same from when I built it in 2013 and I didn't have many issues with it. That was till last summer when I replaced the 3rd gen i7 with a ryzen 2700 and all the motherboard and ram upgrades that went with it. I didn't have the money for the gpu upgrade then but I finally got around to it last week when I bought a used 1080ti from a guy that bought it in 2018 for some gaming and video rendering. He showed me it playing games on his PC (Wolfenstien Young blood, and the gpuz render test) and it ran without any problems or hiccups, other than some pretty noticeable coilwhine.

 

I used DDU to wipe the drivers and did a clean install after putting in the card and it seemed perfectly fine at the desktop and I was also able to enable the variable refresh feature of my freesync monitor which my old card could not due to not having the required display port version. However when I got around to playing some games, they all crashed to the desktop or occasionally locked the PC up shortly after starting the game. I found that lowering the power limit to 85% makes it more stable however this means it only goes to like 1600mhz as opposed to the 1900+ it would if everything was left at default.

 

Could it be my psu? its only 650W but its a gold rated unit (corsair RM650i) so I would expect it to be able to handle a single card. My old 780ti, if gpuz is to be believed drew around 200W in typical gaming, however the 1080ti goes over 250w at stock settings so maybe this increase is causing the issue, and lowering the power limit is getting it back to 780ti levels. I would think that not enough power from the psu would cause hard resets rather than just the games crashing tho...

 

The previous owner of the card has responded to my recent messages and said that he had registered the card on EVGA's site and since he didn't buy the card that long ago the option of a RMA could still exist.

 

Any input would be welcomed.

[Levent]

16 minutes ago, Lord_Karango17 said:

Hi guys,

 

TL;DR: Recently bought a lightly used 1080ti. Put it in my pc after doing a clean driver install only to have it CTD in games at stock settings. Lowering the power limit to ~85% seems to make it far more stable in games however it still behaves a bit odd (occasionally hangs for a frame or 2).

 

it seems you just had just the opposite effect you wanted, just leave it at %100 and see if that helps.

[Lord_Karango17]

1 minute ago, Levent said:

it seems you just had just the opposite effect you wanted, just leave it at %100 and see if that helps.

It was at 100% by default when the card was having all the crashing issues. I first tried underclocking it by 100mhz to see if that would help, but when that didnt help, I fiddled with the power limit and while increasing it had no effect, lowering it seemed to improve stability.

 

Seems very much like a power usage problem where either the power delivery on the card itself cant handle the full tdp (which would be odd because it worked fine in the seller's PC), or my psu is causing issues with the higher power usage of the 1080ti as compared to my old 780ti

 

[AndrewB121]

2 minutes ago, Lord_Karango17 said:

It was at 100% by default when the card was having all the crashing issues. I first tried underclocking it by 100mhz to see if that would help, but when that didnt help, I fiddled with the power limit and while increasing it had no effect, lowering it seemed to improve stability.

 

Seems very much like a power usage problem where either the power delivery on the card itself cant handle the full tdp (which would be odd because it worked fine in the seller's PC), or my psu is causing issues with the higher power usage of the 1080ti as compared to my old 780ti

 

Your PSU should be completely adequate though. 

 

What about cooling, what kind of temperatures are you getting under load?

[Lord_Karango17]

Just now, AndrewB121 said:

Your PSU should be completely adequate though. 

That's what I'm thinking. There is another PC in the house which I'll try to put the card into to see if the crashing persists.

[Levent]

Just now, Lord_Karango17 said:

It was at 100% by default when the card was having all the crashing issues. I first tried underclocking it by 100mhz to see if that would help, but when that didnt help, I fiddled with the power limit and while increasing it had no effect, lowering it seemed to improve stability.

 

Seems very much like a power usage problem where either the power delivery on the card itself cant handle the full tdp (which would be odd because it worked fine in the seller's PC), or my psu is causing issues with the higher power usage of the 1080ti as compared to my old 780ti

 

Interesting, lowering the power limit might indicate that it is overheating (or has issues with its VRM), also what kind of crashes you get? like weird colors on screen (like this), or just straight up app crashes (and event viewer indicates GPU related issue) or driver crashes?

[Lord_Karango17]

39 minutes ago, Levent said:

Interesting, lowering the power limit might indicate that it is overheating (or has issues with its VRM), also what kind of crashes you get? like weird colors on screen (like this), or just straight up app crashes (and event viewer indicates GPU related issue) or driver crashes?

Core temperatures have not gone past low 60C before the crash occues. Unfortunately this isnt one of the EVGA cards that has temp monitors on all the components so I can't tell if the VRM is overheating, but they are rated for such high temps that I dont think heat is an issue. The VRM components themselves may have failed in a manner that only 1 or 2 of the phases are defective so it does not become problamatic until power usage nears 250W.

 

Typically they're soft crashes where the game just hangs and I get kicked back to the desktop with an error message from the game. However one of the times while I was running the pendulum demo to test the freesync, it did have a crash with artificats like the photo you sent.

[Levent]

3 minutes ago, Lord_Karango17 said:

Core temperatures have not gone past low 60C before the crash occues. Unfortunately this isnt one of the EVGA cards that has temp monitors on all the components so I can't tell if the VRM is overheating, but they are rated for such high temps that I dont think heat is an issue. The VRM components themselves may have failed in a manner that only 1 or 2 of the phases are defective so it does not become problamatic until power usage nears 250W.

 

Typically they're soft crashes where the game just hangs and I get kicked back to the desktop with an error message from the game. However one of the times while I was running the pendulum demo to test the freesync, it did have a crash with artificats like the photo you sent.

 

That picture definitely shows some sort of VRAM issue. Underclock your VRAM (GPU memory) and see if that helps.

[MrBrightSyde]

30 minutes ago, Lord_Karango17 said:

 

Typically they're soft crashes where the game just hangs and I get kicked back to the desktop with an error message from the game. However one of the times while I was running the pendulum demo to test the freesync, it did have a crash with artificats like the photo you sent.

 

Looks like an RMA request, if possible.

[Lord_Karango17]

15 minutes ago, MrBrightSyde said:

Looks like an RMA request, if possible.

Yeah, tried to exhaust all my options before going that route. Odd that the previous owner didnt have any of these issues. Granted he was running it on a 1080p 60hz TV and vsync was likely enabled so the gpu usage may never have gotten high enough to causes these problems

[Mark Kaine]

1 hour ago, Lord_Karango17 said:

It was at 100% by default when the card was having all the crashing issues. I first tried underclocking it by 100mhz to see if that would help, but when that didnt help, I fiddled with the power limit and while increasing it had no effect, lowering it seemed to improve stability.

 

Seems very much like a power usage problem where either the power delivery on the card itself cant handle the full tdp (which would be odd because it worked fine in the seller's PC), or my psu is causing issues with the higher power usage of the 1080ti as compared to my old 780ti

 

Have you tried without the freesync shenanigans at all?  Just because it works doesn't mean it's "stable". 

 

If you didn't I'd try that,  set everything to default, don't use free sync and test everything again. 

 

 

It could be PSU, but I have a hunch it's the free sync thing tbh. 

 

 

[Lord_Karango17]

1 hour ago, Mark Kaine said:

Have you tried without the freesync shenanigans at all?  Just because it works doesn't mean it's "stable". 

 

If you didn't I'd try that,  set everything to default, don't use free sync and test everything again. 

 

 

It could be PSU, but I have a hunch it's the free sync thing tbh. 

 

 

Interesting thought. Tried setting everything to default and turned off fresync in the drivers and on the monitor, and same thing happened. a crash to desktop a few seconds after getting into the game menu.

 

I got the card used so the RMA process will be a bit tricky, but EVGA has a transferable warranty and the seller has been responsive to my messages so hopefully that route ends up being possible if nothing else works...

[Roland9385]

I have had a similar issue with my aorus 1080ti and I'm running an 850w gold psu so the psu is probably not the issue. Also I'm running it on a non variable refresh 4k tv. When I bought my gpu new I had to send it back to gigabyte for replacement and apparently the issue was with the vrms. The replacement works well with the right drivers but some drivers make it crash out of all ubisoft games I own which are mostly assassins creed and also random crashes on forza which might be unrelated to your issue but in my experience more information is usually better. Hope this helps.

Edit. I had forgotten to mention that the only stable clock is the gaming setting in aorus graphics engine for some weird reason. Gpu clock 1683mhz mem clock 11010mhz but it boosts up to 2050mhz gpu so my solution was to keep that setting and find a stable driver and not update  it anymore. 

 

Edited December 9, 2019 by Roland9385

[Lord_Karango17]

7 hours ago, Roland9385 said:

I have had a similar issue with my aorus 1080ti and I'm running an 850w gold psu so the psu is probably not the issue. Also I'm running it on a non variable refresh 4k tv. When I bought my gpu new I had to send it back to gigabyte for replacement and apparently the issue was with the vrms. The replacement works well with the right drivers but some drivers make it crash out of all ubisoft games I own which are mostly assassins creed and also random crashes on forza which might be unrelated to your issue but in my experience more information is usually better. Hope this helps.

Edit. I had forgotten to mention that the only stable clock is the gaming setting in aorus graphics engine for some weird reason. Gpu clock 1683mhz mem clock 11010mhz but it boosts up to 2050mhz gpu so my solution was to keep that setting and find a stable driver and not update  it anymore. 

 

Ok thats good to know. About to try the gpu in a seperate PC to see if the problem persists but if it does then it seems like an RMA is the only option.

 

One thing that EVGA support recommended is to use 2 separate cables for the 6 pin and 8 pin power. I've attached a photo of the setup I'm using currently where one cable from the psu splits into the 6 pin and 8 pin, which according to the support guy can cause the issues I'm seeing.

[Lord_Karango17]

Update: Tried the card in my room-mates pc which also has a founders edition 1080ti and while it lasted longer in unigen heaven, it still got unstable and crashed right after it hit 60C.

 

His psu had 2 individual cables for the 6 and 8 pin ports on the gpu so that rules out the power cables being the issue. Talking with EVGA support on possible solutions or an RMA

[Lord_Karango17]

Update 2: Talked with EVGA support and they defaulted to the response of "send it in for an RMA". I know their RMA cards are refurb units so hopefully the one I get from them does not have the same issue...

 

I will update when I get the card back just in case someone else has this issue.

[Lord_Karango17]

Update (hopefully the last): It seems like EVGA did not have any similar 1080ti cards on hand to send as a replacement so I got a 2080 xc ultra as a replacement. All is well in game and it works just like it should.

 

They did not clarify what exactlly was the issue with my original card and I find it strange how it worked in the pc of the guy I bought it from but just not on mine. Seems like a hardware related problem since I tested it in multiple pc's with similar results. For anyone that has a similar issue of the 1080ti crashing in games and generally being unstable until you lower the power limit and the clocks, it likely is a hardware issue with the vrm's and should be sent in for an RMA. Anyone getting those cards at this point will be buying them used as I did, so sticking with brands like EVGA that have robust RMA policies would be your best bet.

 

Just installed a water block onto the 2080 and am leak testing it now. With a mild OC it boosts to just shy of 2100mhz and settls in at 2045mhz once the loop is up to operating temps.

[jones177]

14 minutes ago, Lord_Karango17 said:

Update (hopefully the last): It seems like EVGA did not have any similar 1080ti cards on hand to send as a replacement so I got a 2080 xc ultra as a replacement. All is well in game and it works just like it should.

 

They did not clarify what exactlly was the issue with my original card and I find it strange how it worked in the pc of the guy I bought it from but just not on mine. Seems like a hardware related problem since I tested it in multiple pc's with similar results. For anyone that has a similar issue of the 1080ti crashing in games and generally being unstable until you lower the power limit and the clocks, it likely is a hardware issue with the vrm's and should be sent in for an RMA. Anyone getting those cards at this point will be buying them used as I did, so sticking with brands like EVGA that have robust RMA policies would be your best bet.

 

Just installed a water block onto the 2080 and am leak testing it now. With a mild OC it boosts to just shy of 2100mhz and settls in at 2045mhz once the loop is up to operating temps.

 

Thanks for the info.

I always wondered what EVGA did if the had no replacement cards.

I have 3 GTX 1080 tis under warranty. Now I hope one brakes.

[Kurupt]

Glad you got it sorted out.  I just rma my 1080 ti evga sc2 and I'm hoping to receive a 2080.  Fingered crossed.

[Zodwraith]

In the years since they surfaced I've yet to have a bad experience with evga. I have no idea how they are behind the scenes but I've had enough great cards from them that I automatically go to them for every GPU. This just makes me trust them a little more. Some companies wouldn't give an upgrade replacement to a 2nd owner after mounting a water block. Did they give you any grief over that or did they swap it no questions?

[Lord_Karango17]

On 1/6/2020 at 2:36 PM, Zodwraith said:

In the years since they surfaced I've yet to have a bad experience with evga. I have no idea how they are behind the scenes but I've had enough great cards from them that I automatically go to them for every GPU. This just makes me trust them a little more. Some companies wouldn't give an upgrade replacement to a 2nd owner after mounting a water block. Did they give you any grief over that or did they swap it no questions?

I may have worded it badly, but I never put a water block onto the 1080ti. My previous 780ti had a water block on it and I was going to get one for 1080ti if it ended up working properly, but as it did not, I sent it in for a replacement, and got a water block for the 2080.

 

But yes, I was very surprised at the RMA policies EVGA stood behind and how even 2nd hand purchasers are covered. The 1080ti I got was sold before their current return policy came into effect so I didn't even have to show a proof of purchase from the original owner.

[Lord_Karango17]

On 1/6/2020 at 11:41 AM, jones177 said:

Thanks for the info.

I always wondered what EVGA did if the had no replacement cards.

I have 3 GTX 1080 tis under warranty. Now I hope one brakes.

 

On 1/6/2020 at 1:59 PM, lbmoney33 said:

Glad you got it sorted out.  I just rma my 1080 ti evga sc2 and I'm hoping to receive a 2080.  Fingered crossed.

Hahah, not sure the 2080 is any faster than a 1080ti, but I guess RTX is nice to have...