Some NVIDIA GPU Users Are Complaining of BSODs with Ryzen 3000 Chips

I've tried that. Just tried again actually.

When running Furmark it'll artifact the screen after like 3-5 minutes, then locks up and have to hard power if off. Thinking there might be card instability I set the card to silent mode which underclocks it. No dice. Thing barely breaks 43c during furmark.

I'm at a loss. Only other thing I can try is putting the card directly in a slot instead of using the riser cable for vertical mount.
 
The other option is to take the system down to minimal functional hardware outside of the box. Test and and see if it works. This eliminates the chances of a motherboard stand shorting something or anything like that.

Test with 1 stick of ram, the storage, and a video card. Prove that it works. Move the ram to different slots to see if that makes a difference. And yes get rid of the riser for the test. If you manage to find a stable setup THEN add pieces until you trigger the failure conditions again and boom you've found your culprit.

I've had to tell a few SE's to do just that and it has only failed to provide a solution 1-2 times out of the dozen or so times.. so around a 10% chance of just being flaked out motherboard.

(To build outside of the box use cardboard with a gap under the CPU tends to be the easiest way.)
 
I'm not tearing it out just yet. Even putting the GPU back in a slot is going to require some replumbing.
 
Can you test the riser card in another system without replumbing?
 
Grimlakin said:
The other option is to take the system down to minimal functional hardware outside of the box. Test and and see if it works. This eliminates the chances of a motherboard stand shorting something or anything like that.

Test with 1 stick of ram, the storage, and a video card. Prove that it works. Move the ram to different slots to see if that makes a difference. And yes get rid of the riser for the test. If you manage to find a stable setup THEN add pieces until you trigger the failure conditions again and boom you've found your culprit.

I've had to tell a few SE's to do just that and it has only failed to provide a solution 1-2 times out of the dozen or so times.. so around a 10% chance of just being flaked out motherboard.

(To build outside of the box use cardboard with a gap under the CPU tends to be the easiest way.)
Click to expand...

It's not a motherboard short. That problem yields a very different result. It doesn't BSOD, it simply makes a noise and then shuts down. I've actually seen and caused short circuits before, and it doesn't result in an error code in the event log, (except maybe kernel power errors) blue screens or anything like that.

This is not likely a RAM issue either. I believe Riccochet said he tired reducing the speed of it in another thread etc. If it was a compatibility issue, then that problem would likely present differently as well. Now, the theory about the riser has merit to it. I'd definitely investigate that, but tearing everything out and building it outside of the case will not yield an answer here. This isn't a matter of something shorting out.
 
yeah sometime this week I'll take it off the riser cable. When I have time to drain, bend new pipe and refill the loop.
 
The thing is, you might be able to force PCIe 3.0 and it may resolve your issue. That riser is probably not rated for PCIe 4.0, and I'm wondering if something's not behaving right.
 
Dan_D said:
The thing is, you might be able to force PCIe 3.0 and it may resolve your issue. That riser is probably not rated for PCIe 4.0, and I'm wondering if something's not behaving right.
Click to expand...

I've set the PCIe mode to 3.0 in the BIOS with no change. Either this cable is bad, my card is somehow screwed up from the waterblock install, or there is a BIOS/Driver issue.

Going to drain the loop today and start moving the card. Not much else I can do at this point. Maybe I'll pull the waterblock and reseat that as well. Make sure everything that's supposed to be touched is actually being touched.
 
Moved the card back to a slot and getting the same result. Did a DDU and reinstalled driver as well.
 
You are at the point where it's time to put the video card in another machine, swap RAM, and that sort of thing. Memtest is great, but it isn't definitive in my experience. I've seen modules pass that still didn't work right in a particular board. It's not that the RAM is bad, its that you have to dial it in manually, and often times you have to get into sub-timings to make it work and that's a lot of time, trial and error to make that work. RAM is cheap enough, that I just swap it and get something else. Will that work? Not necessarily.

I would try setting the RAM to JEDEC specs (although, I think you did if I remember right), up the memory voltage, load the full XMP tables if that board lets you. You may not be able to since it's the AM4 / X570 platform. Also, in your case AGESA code matters. So you might need to try a bunch of different BIOS revisions etc.

I tend to validate everything on an AMD build and make sure it all works on stock / air cooling and then convert to water cooling later for this very reason. I would probably chance it on Intel setups, but it would still be good practice there. Although, I tend to validate all my stuff ahead of time as all my boards are boards I've reviewed in the past at some point.
 
well, I feel stupid and slightly relieved. I pulled the card and took the waterblock off. Interesting find. I didn't have good contact between the block and die. In fact, one of the stand off's on the block that surround the die was un screwed a little bit. Made sure everything looked good this time, applied mo'paste, added some thermal pads where they were on the OEM cooler and not on the block/backplate (why not, right). Hitting max of 41c now under Furmark 2k 8xMSAA load. Before it was getting up near 60.

So, something wasn't being cooled properly. Seems stable now. Played about 10 minutes of PUBG and have WoW running in the background. Hopefully it stays stable.
 
Been there done that. Back in the day when 3.5 inch floppy disks were sill a thing I was building a PC with a buddy. And no matter what we couldn't get the thing to recognize the **** hard disk and boot. IT kept throwing drive errors. We tried different drives and everything.

Then after we walked away for a bit (after a couple of hours of WTFing.) We turned the computer on and watched the beige tower case as it tried to boot. But for some reason the floppy drive light stayed on.

Sighing we opened the case flipped the floppy drive ribbon cable to the right way, turned it on and everything worked.
 
Yeah it's definitely good now. Got in about 45 minutes of WoW with no issue.

At least this loop is easy to drain and fill.
 
Nice. I'm glad it seems sorted now.
 
You must log in or register to reply here.
Become a Patron!
Back
Top