After mobo change, screen goes blank on heavy 3D load

[RejZoR]

Like I've said, after preliminary (short) test, problems were gone after removal of software. I had to go to work, but I will test more today as I now finally have exact things to test. Well, at least it looks this way. When I pinpoint it exact, I'll have to check memory dumps and submit them to devs. But I don't want to throw accusations around until I confirm the thing. Have done that mistake in the past...

 

[RejZoR]

Of course it's now not bloody reproducible. Every freaking time something is broken and you kinda figure it out...

 

[DeathtoGnomes]

Hey no spamming posts!

Ofcourse you cant reproduce it, you didnt shake a stick at it.

 

[Derek12]

problems were gone after removal of software

because after removing that, no crash for 2x Killing Floor 2 games where before, it just went south every single time

After removal of particular software, 2x sessions of Killing Floor 2 and no crash

What software did you remove?

 

[RejZoR]

Done a total clean reinstall of windows just to be sure some stupid driver thing left behind even after refresh isn't messing things up. And installed back the alleged offending software. Lets see...

 

[Aquinus]

By chance, do you have you VGA powered via a single cable or two separate cables? If it is one, then try 2. Might help. I know it sounds odd, but I have some VGAs that do not like sharing power cables over both plugs.

besides, I had it connected this way before and it worked fine

I'll second this one, mainly because your PSU, @RejZoR, has +12v split between 4 different rails and I'm not entirely sure how it uses them. It may have worked okay for a long time but, PSUs degrade over time and it's entirely possible that the PSU was just getting by with the way it was wired up and is starting to struggle. Just for shits and giggles (since you still haven't been able to solve the problem,) to try splitting the GPU power between two different cables. Is it possible you used the same cable but, switched which port on the PSU it was plugged into? The specs for that PSU say that two rails are 12v @ 25A and the other two are 12v @ 30A. I guess it's possible that it could be stable with a single cable on a 30A but, not a 25A. Either way, I would try different configurations to see if it makes a difference but, it wouldn't be the first time that I've seen a machine be unstable in one wiring configuration and stable in another.

 

[RejZoR]

"PSU's degrade over time." It's not even 2 years old...

 

[Aquinus]

"PSU's degrade over time." It's not even 2 years old...

2 years can be a lot depending on the amount of load and temperatures it has been subjected to. Either way, I would still try using two cables instead of just one.

 

[DeathtoGnomes]

I prefer modular PSUs, and all the ones I have owned have been marked which plugs/sockets are for which and manuals often tell which rails are with which plugs/sockets. That should make is easy so you can chart what is plugged where and do some creative cable dancing as @Aquinus suggested.

I do have a hard time believing any PSU in the last 5 years will just degrade, I would like assume they would just flat out fail instead.

 

[MrGenius]

Try running the GPU on 2 rails/2 cables. Or, if you haven't already, switch the PSU to single-rail +12V mode. Or do a combo of both, 2 cables & single-rail +12V mode.

 

[RejZoR]

Switched to single rail mode and it froze anyway. I dont get it, if it's single rail mode, it shouldn't matter what cables are used and on what connectors.

 

[Aquinus]

if it's single rail mode, it shouldn't matter what cables are used and on what connectors.

It doesn't matter if a particular rail is to blame, it does if the cable itself is drawing too much power for a single cable and is causing a voltage drop that is sometimes acceptable and sometimes not. Regardless of what mode it's running in, try using two separate cables, ideally the shortest ones you got. That way you're minimizing voltage drop if the cables themselves are to blame.

 

[RejZoR]

Then I'd ask, why on earth would I want to own a top of the line 750W PSU that fails to power a god damn GTX 980 somehow. Or a better question how on earth it managed to degrade and fail in less than 2 years. What's even more bizarre is that sometimes fails 5 times in a row in same game and then out of the blue it decides not to fail for 1 whole day. It's why it's such a bitch to test because when I think I figured it out something, I actually haven't figured out anything when it just goes off randomly after a day.

I've now turned it over so it sucks the air from the outside and connectors are plugged into all other ports, but I'm still being a bit worried it was the PSU that fried the last mobo and now I have everything new hooked to the same potentially sketchy PSU... So annoying. I guess I'll stick that LC Power Arkangel 850W in my system again. A 100€ PSU vs. a 250€ one that I have now. Makes you question everything about high end components if it turns out to be a dud for real...

 

[Aquinus]

I'm not necessarily convinced it's the PSU, I'm just trying to rule out power delivery which could be something as mundane as a bad cable. Bending them a lot or constantly redoing the wiring could have exacerbated the issue which is why I mentioned it. A handful of reviews mention that the cables don't necessarily plugin tightly into the PSU. Perhaps a contact isn't as good as it could be?

 

[RejZoR]

It's now turned over, getting pure cold air, using the opposite modular connectors and now it's the waiting game again. If it goes, Arkangel goes into the box. Then the waiting game again...

 

[Aquinus]

I take it that you don't have another GPU you can test with that would be equal or greater with respect to power draw? If it continues to happen, I would want to rule out the GPU next since it seems like you've already done enough testing to make sure that the CPU and memory are stable.

 

[RejZoR]

I'm not necessarily convinced it's the PSU, I'm just trying to rule out power delivery which could be something as mundane as a bad cable. Bending them a lot or constantly redoing the wiring could have exacerbated the issue which is why I mentioned it. A handful of reviews mention that the cables don't necessarily plugin tightly into the PSU. Perhaps a contact isn't as good as it could be?

After thinking about it, I think you might be on to something. When I switched it to a single rail mode and problem remained (even though it shouldn't since single rail has like 70A), only thing left can really be cables. Haven't had a problem so far, but all cables are twisted since I didn't remove zip ties. Which also explains why one day works and not the other (after I've fiddled with it a bit and moved the cables). Not sure how I can fix it other than use another pair of PCIe cables. Luckily I have plenty since PSU is quad SLi/CrossfireX ready

 

[DeathtoGnomes]

Since you have a spare PSU, does this mean you also have spare cables to use? Use the external PSU (with a jumper) and hook it to your card to power it that way.

After thinking about it, I think you might be on to something. When I switched it to a single rail mode and problem remained (even though it shouldn't since single rail has like 70A), only thing left can really be cables. Haven't had a problem so far, but all cables are twisted since I didn't remove zip ties. Which also explains why one day works and not the other (after I've fiddled with it a bit and moved the cables). Not sure how I can fix it other than use another pair of PCIe cables. Luckily I have plenty since PSU is quad SLi/CrossfireX ready

Wait, you used zip ties? If you over tighten them, and power draw thru any wire was enough to heat up the wire, you could have melted the wire coating enough to cause a bare spot in the wires tied up causing arcing and other bad things. I've seen it happen in cars/trucks, several times. Not something you can see immediately but only after you tear things apart.

Velcro straps are safer to use, just saying...

 

[RejZoR]

Melted PSU wires? From 25-30 amps? Cars may do that because 70A go through single wire. And even those are usually very thick...

 

[DeathtoGnomes]

Melted PSU wires? From 25-30 amps? Cars may do that because 70A go through single wire. And even those are usually very thick...

Sure, wire size matters, as does quality. A spark doesnt need a lot space to jump over, voltage and amps determine how much space is needed before it jumps. The number of braided wire strands per foot (I think was the standard) counted towards its amperage rating. Some of the old power supplies, if you remember, had some very stiff harnesses, it was hard to bend them. It was a pain in the butt to get the 20/24 pin connector harness to take the right shape. That was because there were a low number of strands in each wire, and each stand was larger to compensate for the same amperage. Now more strands=more amps can be carried, more flexible wire harnesses. All it takes to start breaking down a "higher" strand count wire is one strand to break and start arcing. Something like repeatedly folding a piece of metal in half, eventually it breaks.

Also, wire insulation is not perfect, it can dry and crack too depending on quality, its kind of rare. Cheaper wire might also have thin spots or even a broken or bent strand sticking out, all caused during manufacturing and poor QC.

IF you can reproduce your issue by wiggling wires in your case....

 

[RejZoR]

And it happened again. I haven't touched the inside of the case at all. I never had such infuriatingly annoying problem ever. And I've seen a lot of dumb shit over the years.

Gonna change cables now and if that won't help, LC Power Arkangel is going in. It doesn't look bad at all actually. Internals seem incredibly well organized and cables, while not sleeved, they are flat and stealthy.

 

[RejZoR]

Sticked Arkangel PSU in and exactly same thing happened after 15 minutes of gameplay. Meaning PSU is fine. Apparently it's the graphic card spazzing out on its own. In totally random and illogical way. Only thing I have to test is GeForce 8400GS which is unlikely to be able to run Prey 2017 so I can test what's what. And I think I bought it from same retailer as the mobo. Which went bust like a year ago. Awesome.

 

[DeathtoGnomes]

time to examine the card and MB with a magnifying glass

 

[RejZoR]

Mobo is fine, I'm suspecting Sabertooth dying did something to graphic card as well. Which explains why it correlates with a mobo change and not before. Can't inspect it because of stupid warranty sticker if I think of sending it off to ASUS.

Have no idea how to check what is wrong with graphic card. It doesn't do this all the time, it can go for 2 days without freeze and then it'll repeatedly freeze every single time one day in a matter of 10-15 minutes in a game. It just makes no sense at all.

 

[DeathtoGnomes]

if its still under warranty do the RMA.