Inconsistent crashes on Vega

[Kevi983]

Greetings, fellow gAmErS.

 

So I've got a weird issue with my Vega 56. I bought the card (which is the MSI blower style version) about 5 months ago and it has served me well, at first. But now it sometimes crashes at stock settings. I've got a 600 Watt Pure Power 11 as my PSU so I don't think power is the issue here.

 

There are 2 different types of crashes. When under gaming load (Forza Horizon 4 for example) it crashes by giving me a blackscreen (most of the time , but I've seen it freeze once or twice)and the gpu fan goes up to 100%, I hear sound for a while and then I'm forced to restart the PC. If it's not under load (when browsing chrome for example) it crashes by freezing and thats it, no blue screen, nothing. When either of those crashes occur I have a message pop up, the next time I boot the pc, saying that there's been an issue and radeon settings were reset.

 

My temps are fine, even though the HBM is much hotter than the GPU and I've tried setting the fanspeed higher, to no avail. I also have a weird flickering issue on my primary monitor only, which is connected to the GPU by DP. It's only in applications like Chrome and the event manager, windows settings, etc. I've never had the flickering when gaming and I can't shake the feeling, that it may have to do with the crashes.

 

I had a Vega 64 in my system too, for a while, which had the same flickering issue, but didn't crash. I've also tried stresstesting the card with FurMark and left it running for a whole 4 hours, before turning it off. It didn't crash. So yeah, I guess you could say I'm confused. For anyone wondering whether it may be the Ram or Cpu, I've also tried stresstesting these and they had no issue. My drivers are always up to date and I've got the most current BIOS for my Mainboard aswell. Which is a Gigabyte GA-AB350-Gaming and my CPU is a Ryzen 7 1800x. Leaving RAM and CPU clocks at stock also didn't change anything. These crashes happen like once or twice a day, when I'm using the PC.

 

I'm looking forward to your ideas, since all of mine just ran out. Good day.

 

Edit: For anyone interested, the best way I could discribe the flickering is that they kind of look like the artifacts you get when your OC isn't stable. The screen doesn't turn on and off or anything like that.

[Tristerin]

Do you suffer from GPU sag?  The power delivery pins are on PCI slot may not be contacting well if so.

 

I have had this issue before that was fixed with propping a GPU up.

[Kevi983]

18 minutes ago, Tristerin said:

Do you suffer from GPU sag?  The power delivery pins are on PCI slot may not be contacting well if so.

 

I have had this issue before that was fixed with propping a GPU up.

The card doesnt sag at all. Doesnt even budge when I try to shake it. But I've taken it out and put it back in the slowest and most precise manner I could, just be to sure.

[Kevi983]

21 minutes ago, Tristerin said:

Do you suffer from GPU sag?  The power delivery pins are on PCI slot may not be contacting well if so.

 

I have had this issue before that was fixed with propping a GPU up.

What issues did you have exactly? The flickering or the crashes?

[Tristerin]

Just now, Kevi983 said:

The card doesnt sag at all. Doesnt even budge when I try to shake it. But I've taken it out and put it back in the slowest and most precise manner I could, just be to sure.

Do you run MSI Afterburner?  If so remove the program and RIVA from your PC and see if that helps.  Deep bowels of Reddit have lead me there as well with issues on GPUs like this.  

[Tristerin]

Just now, Kevi983 said:

What issues did you have exactly? The flickering or the crashes?

Black screen crashes during load or not, random, blamed the PSU because I was having PSU coil whine.  I removed MSI Afterburner and Riva and it helped for like 1 week and then I got a black screen.  Researched the pins on the PCI slot (because I did have sag) and propped my card up, no problem.

 

My conclusion, while not scientific or potentially could be wrong, is that the legs of the PCI slot were getting bad contact and would random power crash (and get the radeon wattman reset at startup) with black screen and that their was bad contact with other pins as well that was communicating with the card and afterburner.

[Kevi983]

2 minutes ago, Tristerin said:

Do you run MSI Afterburner?  If so remove the program and RIVA from your PC and see if that helps.  Deep bowels of Reddit have lead me there as well with issues on GPUs like this.  

I've uninstalled every single on of these OC programs just to be sure. Also I've used DDU and all that jazz. Didn't make a difference. 

[Leanora]

before the issue started, was it then too on stock settings? 

Try undervolting the card by 5 or 10%.  

Try overvolting your card by 5 or 10%. 

[Tristerin]

1 minute ago, Kevi983 said:

I've uninstalled every single on of these OC programs just to be sure. Also I've used DDU and all that jazz. Didn't make a difference. 

 

It sounds like you have done a lot of pregaming on fixing this issue - have you tried a fresh windows install?

[Kevi983]

7 minutes ago, Leanora said:

before the issue started, was it then too on stock settings? 

Try undervolting the card by 5 or 10%.  

Try overvolting your card by 5 or 10%. 

It actually had it undervolted and the HMB oc'd back when it still worked fine. Then the crashes started occuring which led me to progressively lowering the clocks, so I decided to just go full stock and see if that even works.

[Kevi983]

4 minutes ago, Tristerin said:

 

It sounds like you have done a lot of pregaming on fixing this issue - have you tried a fresh windows install?

That will be my last resort. 

[Tristerin]

3 minutes ago, Kevi983 said:

It actually had it undervolted and the HMB oc'd back when it still worked fine. Then the crashes started occuring which led me to progressively lower the clocks, so I decided to just go full stock and see if that even works.

Could it be your HBM maybe overheating?  Im guessing here (and going to actually be messing with 1 of my R9 Fury this weekend replacing TIM and thermal pads on the HBM) because 1 of my 2 run much hotter than the other.

 

44 minutes ago, Kevi983 said:

My temps are fine, even though the HBM is much hotter than the GPU and I've tried setting the fanspeed higher, to no avail. I also have a weird flickering issue on 

Im not sure what is Fine for HBM (Im replacing with 1.5mm thermal pads)

[Kevi983]

Just now, Tristerin said:

Could it be your HBM maybe overheating?  Im guessing here (and going to actually be messing with 1 of my R9 Fury this weekend replacing TIM and thermal pads on the HBM) because 1 of my 2 run much hotter than the other.

 

Im not sure what is Fine for HBM (Im replacing with 1.5mm thermal pads)

I've checked GPU Z a couple of times, because wattman doesnt let you check HMB temps and the HBM is running like 7-10 degrees hotter than the core. It is said online that HBM should be fine at up to 90 degrees and I've adjusted the curve to severely go up at like 65-70 degrees core temp.

[Kevi983]

1 minute ago, Kevi983 said:

I've checked GPU Z a couple of times, because wattman doesnt let you check HMB temps and the HBM is running like 7-10 degrees hotter than the core. It is said online that HBM should be fine at up to 90 degrees and I've adjusted the curve to severely go up at like 65-70 degrees core temp.

In the stresstest, which was with the stock fan profile btw, The HBM went up to 83 degrees at most. and that went on for 4 hours, so I think that's fine

[Tristerin]

1 minute ago, Kevi983 said:

I've checked GPU Z a couple of times, because wattman doesnt let you check HMB temps and the HBM is running like 7-10 degrees hotter than the core. It is said online that HBM should be fine at up to 90 degrees and I've adjusted the curve to severely go up at like 65-70 degrees core temp.

So two diff GPUs doing the same thing essentially.  Id start to rule out the GPUs and move to the next culprits.  Have you run memtest?  Prime95 on the CPU?  OCCT on the PSU?

[Kevi983]

Just now, Tristerin said:

So two diff GPUs doing the same thing essentially.  Id start to rule out the GPUs and move to the next culprits.  Have you run memtest?  Prime95 on the CPU?  OCCT on the PSU?

I've stressedtested the CPU and RAM with prime 95. That was with an overclock. The CPU was at 3,8 GHZ and the Ram was at 2800mhz. I turned it off after 1 hour and like 10 minutes. Now I've even reset my bios to stock so my ram runs at 2133 MHZ and the CPU at 3.6 GHZ and it's still crashing from time to time.

[Tristerin]

1 minute ago, Kevi983 said:

I've stressedtested the CPU and RAM with prime 95. That was with an overclock. The CPU was at 3,8 GHZ and the Ram was at 2800mhz. I turned it off after 1 hour and like 10 minutes. Now I've even reset my bios to stock so my ram runs at 2133 MHZ and the CPU at 3.6 GHZ and it's still crashing from time to time.

OCCT on PSU?  Sorry just suggesting things atm that I would try.  Id hate to see two bad cards like that myself!!!

[Tristerin]

TBH I went through all of this...and propped my GPU up...and it all stopped.  It wasn't terrible sag either just slight sag on the card lol.

[Kevi983]

2 minutes ago, Tristerin said:

Id hate to see two bad cards like that myself!!!

It's only one Vega 56.

[Kevi983]

2 minutes ago, Tristerin said:

OCCT

I'm gonna try that today.

[Tristerin]

1 minute ago, Kevi983 said:

It's only one Vega 56.

Thought you had a 64 that was flickering, just re-read but it wasn't crashing I see.

 

Prop your card up, use a dowel rod or a pen or something.. I suggest you just try it

[Kevi983]

1 minute ago, Tristerin said:

Thought you had a 64 that was flickering, just re-read but it wasn't crashing I see.

 

Prop your card up, use a dowel rod or a pen or something.. I suggest you just try it

It's not sagging at all, I would have to kind of push it up. Would you still say I should try it?

[Tristerin]

29 minutes ago, Kevi983 said:

It's not sagging at all, I would have to kind of push it up. Would you still say I should try it?

Id be surprised if its not slightly sagging, that board isn't reinforced and that card isn't short and the power connectors may be pulling that end down as well depending on how you have it plugged in.  And yes I would recommend a slight elevation (it wont break the slot, don't jam it into the sky either lol).  Not sure what case but take a pen apart and cut down the center piece, or use a piece of wood even after measuring and cutting.  My Fury on my ASRock B450 board has a metal dowel rod slightly holding the far right closest to the panel end of the card up - solved all my black screen issues with wattman resets.

[Tristerin]

This isn't my pic but anything works for GPU sag lol - I use a metal dowel rod I cut down and painted black to match

[Kevi983]

5 minutes ago, Tristerin said:

Id be surprised if its not slightly sagging, that board isn't reinforced and that card isn't short and the power connectors may be pulling that end down as well depending on how you have it plugged in.  And yes I would recommend a slight elevation (it wont break the slot, don't jam it into the sky either lol).  Not sure what case but take a pen apart and cut down the center piece, or use a piece of wood even after measuring and cutting.  My Fury on my ASRock B450 board has a metal dowel rod slightly holding the far right closest to the panel end of the card up - solved all my black screen issues with wattman resets.

I sincerely thank you for all the input. I'll try that and report back to you if it makes a difference.