new card, things never work out...

[robAP]

2070 super. weirdness. I can fold it 100% no hiccups. I can run the pc folding 100% gpu and occt stress test on the cpu, no hiccups, psu strong as ever in this regard. However it is 10+ years old now, It was the 2009 xmas prize here. a Corsair CMPSU-850HX....

I can only assume it's a bit tired, it's been more on than off 24/7 running idle or gaming or whatever, but its rerely been off. I know things degrade inside over time, and it was FREE so I might just pony up for a good solid 650w unit.

Anyway, In playing games, light games, nothing even demanding, I'm getting system reboots. no bsod's or artifacts, just a blank screen and reset of the box.

I hate to think it's the new card. MY cpu idle states are 10% and 100% max in the windows profile. Seeing how this never happened with my old 1070 I have to assume its the new card....but the new card consumes up to 230+ watts. Through two cables.....perhaps a cable is bad or the socket i plugged the new one into is bad....But...its happening during light games too....

I will try swapping out gpu power cables and try different sockets in the psu, just to make sure i'm not seeing any weird bits geting old, or inability of some outputs inable to handle power load spikes? I don't know, I'm just theorizing now. However I'm not immune to things failing at the same time as other things getting replaced. Happens in cars to me...

Anyone with thoughts that have been through similar symptoms? That's what I'm looking for more now is other similar symptoms.

 

Edited April 16, 2020 by robAP

[ir_cow]

From my experience without diagnosing the problem more when I just get a random black screen it was been when the video card drivers crash.

This often for me is when I overclocked the video card, memory or CPU and it passes a stress test but still fails in a game. 

However it is strange that you can fold and not crash. If I had to guess without any more information would be the power supply and or CPU overclock. Folding doesnt use the CPU much at all. If you play a game the CPU is put until many types of load and so is the video card. Amps go up and down constantly.

You can do the following to check the cpu and video card.

CPU: Run Prime95. This will stress all cores at the same time. Much heavier load then the cpu will ever see normally. Make sure to watch the temps!

Video Card: Run Msi Kombustor in stress test. This will put the card under heavy load. More then any game would normally do.

Memory: Memtest86. Checks the memory to make sure its working properly at the desired frequency.

My guess is Prime95 will crash your computer.

 

[robAP]

1 hour ago, ir_cow said:

From my experience without diagnosing the problem more when I just get a random black screen it was been when the video card drivers crash.

This often for me is when I overclocked the video card, memory or CPU and it passes a stress test but still fails in a game. 

However it is strange that you can fold and not crash. If I had to guess without any more information would be the power supply and or CPU overclock. Folding doesnt use the CPU much at all. If you play a game the CPU is put until many types of load and so is the video card. Amps go up and down constantly.

You can do the following to check the cpu and video card.

CPU: Run Prime95. This will stress all cores at the same time. Much heavier load then the cpu will ever see normally. Make sure to watch the temps!

Video Card: Run Msi Kombustor in stress test. This will put the card under heavy load. More then any game would normally do.

Memory: Memtest86. Checks the memory to make sure its working properly at the desired frequency.

My guess is Prime95 will crash your computer.

 

Thanks for the reply Mr. Cow,

The key is that everything was just fine before the new GPU. CPU was fine and run stock. Memory fine and run stock. Temps fine. Adding this card brings a higher draw on the PSU, now using two cables. I'm replacing them with the unused spares and well see if anything happens there. That would only clear up a shorting of a cable or psu individual socket instability but how that is only in games is not likely. I will start there. My PSU rails seem normal, voltage isnt dropping on load.

Ill change out gpu cables, and then probably order up a new psu today (rm650x) as mine is out of its 7 year warranty (+4) so that end would be protection for the future of any build for another decade. And If it's not fixing anything, im returning the card.

Edited April 16, 2020 by robAP

[ir_cow]

I reason why I was suspecting the CPU was because it didn't crash during folding. But since you said this only started after the new video card then it could be to combo of both CPU+GPU drawing power. My Seasonic 1000 Watt Platinum stop holding a load over 600 watts after 7 years and was considered one of the best PSUs (at the time by JonnyGuru). So nothing is immune to age.

I am would agree it seems like a power supply issue, but there is only limited ways to find out. By stressing the CPU only and the GPU only, you can eliminate either being the problem. Which than leaves some windows software problem (drivers, etc) or power supply.

If you can run MSI Kombustor Stress Test for 10 minutes I can guarantee it is not bad video card.

 

[robAP]

Swapped cables. Swapped sockets, same. Psu ordered. May go return the card and swap it for a replacement. Will advise.

[robAP]

same as it ever was. ordered new PSU just to keep things under warranty and current and safe.

I just folded all night on both cpu and gpu. CPu was running right close to 80. card was 76. This issue only happens in a game, black screen reboot. My mosfets and chipset temps are 55-65 through all of this, and I just got a reboot. When I got the reboot, I had removed the case panels after folding on teh cpu was done, and continued folding on the card. This was for max cooling to see a difference. The card sat at 75c and folding 100%, I was playing a youtube video, and hunting for a possible replacement 2070s on newegg with quiet-ish fans. CPU unloaded. BOOT. no bsod or memory or graphical artifacts.

Its the card or the psu, nothing else as the only thning different is the card and its power draw (+70w over what I was pulling befor with my 1070sc). I'm not clocking anything up, running stock. I would find it hard to believe the 850 PSU was just fine at 380 watts but has degraded so much it cant handle 450, if that.

I think the plan will be to return the card, and when the new psu comes in, get the replacement so i'm not battling return windows vs. delivery of parts.

 

[robAP]

A FIX!

I believe I fixed the problem.Replaced the 11 year old HX850w Corsair with a new EVGA 850gq (Last one in a 25 mile radius of Best Buys) 

It is no longer crashing during the games. I will monitor behavior over the next 2 days and that gives me time in the GPU return window to evaluate. It's a MSI 2070 super ventus oc

JohnnyGURU reviewed this psu and was all kinds of glowing about it. I doubt I will ever use more than half the rated power, but then, that would put me square in the most efficient range, as well as silent fan mode. I have some replacement botom intake fans coming, and I found a way to mod the 280x case for two 80mm exhaust fans....where it doesnt have the option at all. I will then have abotu the best possible configuration for flow and we will see what temps I maintain while folding. (I'll even regrease the gpu if it remains as solid as things appear.

Thanks for the guidance, suggestions, etc.

I'm offering the former GPU over to Gamers Nexus for a possible old GPU test, to see what degrades and to show that not even the best PSU is immune to age. It was free to me courtesy of OCC and Corsair, so I figure I shoudl pass it along to do some good.

[ir_cow]

Glad it was just a power supply issue! 

From what I've read and testing myself over the years is that the Silicon will long outlast anything else on the PCB. 24/7 load (like from mining) will degrade the GPU and the overclocking stability lowers with age. Not because of the GPU is under load, its the applied voltage is never tested with 10 years lifespan in mine. I remember watching a interview with a NVIDIA tech a few years ago and the guy said the GPU is designed for 3 years 24/7 load operations. Obliviously  lower temps will make a difference. This is always why Quadro cards have lower clock speeds because the voltage is lower too. Its meant to last a lot longer. 

So in short, keep the video card cool as possible and don't overclock it if you want it to last 10 years. But good luck playing any 2019/2020 on a GTX 480. I usually suggest not to worry about it because you will replace it long before it dies from actual silicon desegregation.