Suddenly can't put any load on week old 2080 Ti without crashing

[TonyTurbo78]

It is just working now though
Corsair HX850 wasn't up to it. 1KW superflower has the juice
bf1 @ 144Hz 1440p maxed, roll on bfV!

Ah sweet, good news.

I'll be honest, I didn't read further down, good news PSU was at fault, bit of a relief that huh.

And so it should, I can do pretty much that on a 450 quid poo AMD card

 

[spikerules]

That’s odd. My 850w is fine?

 

[HazardO]

That’s odd. My 850w is fine?

Well my rig below with the hx850 and a pair of 1070s seemed fine. System hard crashed under load when I replaced the 1070s with a 2080Ti. So far so good with the superflower. What CPU are you running? Watercooling as well?

 

[spikerules]

Well my rig below with the hx850 and a pair of 1070s seemed fine. System hard crashed under load when I replaced the 1070s with a 2080Ti. So far so good with the superflower. What CPU are you running? Watercooling as well?

I’m running a 8086k @5.1GHz at the moment. Both GPU and CPU on air.

 

[HazardO]

I’m running a 8086k @5.1GHz at the moment. Both GPU and CPU on air.

I expect an 8 core Broadwell E + water cooling draws a fair bit more power.

 

[Lokken86]

The psu must have been faulty. I have the rm850x Corsair or whatever it’s called. It manages the rig in sig with 7 10tb ironwolf drives and 10 other external drives plus the 1950x @4ghz. I even had a 480 in the system along with the vega for a while and the system was still stable.

 

[Paladine]

I am running two systems with HX850i: one with 6700k (o/c 4.6GHz) other 8700k stock both with H100i AIO and a couple of drives + 2080Ti FE o/c to around +170/800 and the 12v rail is sitting at 12.20v stable in Corsair Link (I switched them to single rail mode).

Even with the O/C Corsair Link is showing max wattage (during gaming) as 548W from the wall so far. Keep in mind that that was running the 2080Ti at max (83C) and CPU at reasonable load as well.

I don't know how accurate Corsair Link is with regards to "PowerIn (W)" from the HX850i but I am guessing it is going to be reasonably close...

 

[HazardO]

The psu must have been faulty. I have the rm850x Corsair or whatever it’s called. It manages the rig in sig with 7 10tb ironwolf drives and 10 other external drives plus the 1950x @4ghz. I even had a 480 in the system along with the vega for a while and the system was still stable.

It was purchased back in July 2013, it's been paired with an x79 board with an AIO cooled overclocked 3930k + 2x 970s, then an x99 board with a watercooled overclocked 6900k and 2x 1070s, now 1070s swapped for a 2080Ti. Power supplies do degrade over time. I'm sure it would be capable of a few more years service in a less stressful environment.

 

[Questloved]

So I've had this new PSU in my PC for 4 days now and I've yet to experience a single crash even though I've been overclocked at 5ghz and 1.275v on my 8700k the whole time. Before I would get occasional crashes even at 4.8. Is it possible my old PSU was causing instability there as well?

Also, DPD are picking up my GPU tomorrow so hopefully I'll know the outcome of this RMA by the end of the week. I must say OCUK customer support has been fantastic with me so far.

 

[ltron]

My 2080Ti FE has been causing complete sudden system halts under load (First instance running FFIX bench, after testing BF1 @144Hz ULTRA settings, then again just in bf1 on an empty server). I'm not sure whether it's a faulty GPU overheating or my Corsiar HX850 (mk 2 IIRC) failing to supply the juice. It's been fine with my overclocked 6900k, watercooling and a pair of 1070s, but maybe the 2080Ti has pushed it over the edge (Although it is supposed to be a single 12V rail with 70A available). I've got my 1070s back in at the moment and it's running flawlessly. On paper as the 1070s are OC models they draw ~170W each while the 2080Ti is supposed to be less than 300W, but I guess if the boost clocks are much higher than the stated 1650MHz then the power rating is probably bogus. I estimate my CPU is drawing ~250W, not sure about the 2 d5 pumps in the watercooling.

Anyway a 1000W Superflower leadex 2 80+GOLD is arriving this morning, with the extra 12A on the 12v rail that should determine whether it's the GPU or not (It was getting very toasty though). Fingers crossed here as I expect an RMA to Nvidia to be a long and painful process if it is the GPU.

I'm having random TDRs in games with my new 2080Ti FE, I'm not happy. Going to try putting my GTX 1080 back in and testing the 2080 Ti in an old PC. I'm glad you and the OP not getting crashes with the new PSU.

 

[HazardO]

I'm having random TDRs in games with my new 2080Ti FE, I'm not happy. Going to try putting my GTX 1080 back in and testing the 2080 Ti in an old PC. I'm glad you and the OP not getting crashes with the new PSU.

Which drivers have you tried? Also did you do a clean driver installation or use DDU? It sounds like a driver issue but could be a faulty card. Doesn't sound like a power issue to me.

 

[DaveGW]

My 2080Ti FE has been causing complete sudden system halts under load (First instance running FFIX bench, after testing BF1 @144Hz ULTRA settings, then again just in bf1 on an empty server). I'm not sure whether it's a faulty GPU overheating or my Corsiar HX850 (mk 2 IIRC) failing to supply the juice. It's been fine with my overclocked 6900k, watercooling and a pair of 1070s, but maybe the 2080Ti has pushed it over the edge (Although it is supposed to be a single 12V rail with 70A available). I've got my 1070s back in at the moment and it's running flawlessly. On paper as the 1070s are OC models they draw ~170W each while the 2080Ti is supposed to be less than 300W, but I guess if the boost clocks are much higher than the stated 1650MHz then the power rating is probably bogus. I estimate my CPU is drawing ~250W, not sure about the 2 d5 pumps in the watercooling.

Anyway a 1000W Superflower leadex 2 80+GOLD is arriving this morning, with the extra 12A on the 12v rail that should determine whether it's the GPU or not (It was getting very toasty though). Fingers crossed here as I expect an RMA to Nvidia to be a long and painful process if it is the GPU.

When do you stop throwing money at these cards, do you ever stop and just play games?. No wonder interest in the 20xx cards has run stale.

 

[HazardO]

When do you stop throwing money at these cards, do you ever stop and just play games?. No wonder interest in the 20xx cards has run stale.

Yawn. My card is fine, my PSU needed replacing, and my system sees plenty of gaming action. If you've no interest in the 20xx series cards maybe you should refrain from posting in this thread.

 

[ArmitageShanks]

I'm having random TDRs in games with my new 2080Ti FE, I'm not happy. Going to try putting my GTX 1080 back in and testing the 2080 Ti in an old PC. I'm glad you and the OP not getting crashes with the new PSU.

I was having some sporadic random crashes when playing Project Cars 2 in VR after installing my 2080Ti. Drivers cleanly re-installed, still random crashes and sometimes even a BSOD. Everything works fine on benchmarks though, but benchmarks usually test the GPU and CPU in isolation so the overall load on the PSU is rarely a problem.

Switching to separate PCI-E cables for the 8-pin connectors seems to have resolved it (touch wood). My Corsair RM1000i runs as 12V multirail by default, but I'm too lazy to get to the USB connector to reconfigure it as a 12V single rail. If I ever went SLI again by adding another 2080Ti, I'd certainly have to though.

 

[ltron]

I was having some sporadic random crashes when playing Project Cars 2 in VR after installing my 2080Ti. Drivers cleanly re-installed, still random crashes and sometimes even a BSOD. Everything works fine on benchmarks though, but benchmarks usually test the GPU and CPU in isolation so the overall load on the PSU is rarely a problem.

Switching to separate PCI-E cables for the 8-pin connectors seems to have resolved it (touch wood). My Corsair RM1000i runs as 12V multirail by default, but I'm too lazy to get to the USB connector to reconfigure it as a 12V single rail. If I ever went SLI again by adding another 2080Ti, I'd certainly have to though.

Thanks, but I'm already using separate PCI-E cables for the 8 pin connectors, I also tried switching the PSU to single rail mode but none of it made any difference. I enabled debug mode in the Nvidia control panel which removes the factory overclock but it just crashed now after 4 hours of 3D Mark Timespy graphics test 2 in a loop( my favourite GPU stress test).

 

[ltron]

Which drivers have you tried? Also did you do a clean driver installation or use DDU? It sounds like a driver issue but could be a faulty card. Doesn't sound like a power issue to me.

I've just used the latest 416.34, tried DDU and clean install but didn't help. I don't think it's a power issue either as voltages are not dropping when it happened.

My faulty Asus Sabertooth X99 would sometimes cause GPU driver crashes, but it would also cause crashes that would hard lock the system. Since getting my Gigabyte X99 SOC Champion I've not had any display driver crashes and certainly not in games (there was a bug with Firefox that crashed the display driver but that was fixed in a driver update).

Will try my GTX 1080 again, thankfully haven't sold it and will try the 2080 Ti in an old PC.

 

[HazardO]

I've just used the latest 416.34, tried DDU and clean install but didn't help. I don't think it's a power issue either as voltages are not dropping when it happened.

My faulty Asus Sabertooth X99 would sometimes cause GPU driver crashes, but it would also cause crashes that would hard lock the system. Since getting my Gigabyte X99 SOC Champion I've not had any display driver crashes and certainly not in games (there was a bug with Firefox that crashed the display driver but that was fixed in a driver update).

Will try my GTX 1080 again, thankfully haven't sold it and will try the 2080 Ti in an old PC.

Have you tried logging the GPU temps to see what it gets to?

 

[ltron]

Have you tried logging the GPU temps to see what it gets to?

Yes, it only gets to 76/77 C after hours of stress testing so I'm happy with those temps, I'm not having the problems some on reddit were.

 

[ArmitageShanks]

Thanks, but I'm already using separate PCI-E cables for the 8 pin connectors, I also tried switching the PSU to single rail mode but none of it made any difference. I enabled debug mode in the Nvidia control panel which removes the factory overclock but it just crashed now after 4 hours of 3D Mark Timespy graphics test 2 in a loop( my favourite GPU stress test).

Yeah, I had a few crashes again myself yesterday (seemingly every second race in pCars 2 would crash), so that's at least ruled out a power issue.

I'm starting to wonder if it's a buggy nVidia driver problem. Currently running 416.34.

 

[ltron]

Yeah, I had a few crashes again myself yesterday (seemingly every second race in pCars 2 would crash), so that's at least ruled out a power issue.

I'm starting to wonder if it's a buggy nVidia driver problem. Currently running 416.34.

Have you got Nvidia driver stopped responding and has recovered messages in the event viewer when this happens, like me?