Vega 64 Crashing

[Clcare556]

I am a computer noob that has basic technical skills. I don’t know code or complicated computer processes but I can figure things out and understand them if explained or shown. 
 

I built a pc that consists of: 

msi b450 tomohawk

ryzen 5 2600x 

msi Vega 64 reference card 8gb

corsair rmx750w full modular psu

 

I let windows update and it installed some and software on the computer. I didn’t think much of if then I installed drivers for MOBO and for GPU. Running the latest software with Wattman. 
 

at first I tried to undervolt it while running heaven and while seemingly stable in heaven it crashed in apex legends. I reset to stock settings and it ran fine. 
 

fast forward to today I loaded rainbow six siege running it on high settings and it crashes as soon as I die. I was monitoring GPU heat and it was staying low 60 range. I also had fans adjusted up. 
 

after doing some research I concluded that I needed to delete any drivers that Microsoft installed and do a clean install. I do that and it runs for a couple of rounds and crashes. 
 

only other thing is I am not currently running two separate cables for both power slots on card. What are your thoughts? How do I make sure there are no more Microsoft installed drivers. I appreciate any insight or help I can get. Thanks!

[Streetguru]

Just now, c00face said:

I don't know dude. This whole section is filled with "AMD GPU is crashing, I just bought new AMD card and PC is crashing, etc..." Some of you who still think AMD is quality product, maybe should start rethinking.

Never had an issue myself even buying cards that were mined on...I tend to blame windows 10 more often these days, I can load the same GPU drivers under windows 7 and be fine.

 

 

15 minutes ago, Clcare556 said:

I am a computer noob that has basic technical skills. I don’t know code or complicated computer processes but I can f

Check your windows event log, might have some info on the errors causing crashing.

You could try setting the power limit to 50% to see if the issue is the PSU

Is your RAM running at default speeds?

[Deli]

I hope people aren't running Wattman and MSI Afterburner at the same time.

[Clcare556]

I am only running Wattman. I actually was just reading about the whole overclocked ram thing. I did adjust that. I will set to stock and try that. 
 

i would’ve gone the Nvidia route but I bought this card lightly used from a neighbor who builds PCs. 
 

on a side note I still may get separate power cables just to be sure. 

[Streetguru]

6 minutes ago, Clcare556 said:

I am only running Wattman. I actually was just reading about the whole overclocked ram thing. I did adjust that. I will set to

Having 2 separate cables really shouldn't matter unless you're doing a very heavy overclock.

Might just be system RAM instability, you can up it's voltage to 1.4V, or just drop it down one tier in mhz.

[Clcare556]

7 minutes ago, Streetguru said:

Having 2 separate cables really shouldn't matter unless you're doing a very heavy overclock.

Might just be system RAM instability, you can up it's voltage to 1.4V, or just drop it down one tier in mhz.

I’m running Corsair vengeance 3.2Ghz and I set it to 3.2Ghz when setting the computer up. I will check that tomorrow and see if that helps the issue.
 

I will also try to monitor how much power the gpu is pulling before it goes out. One other thing I noticed is the card gets up to p7 and crashes. 

[Streetguru]

Just now, Clcare556 said:

I’m running Corsair vengeance 3.2Ghz and I set it to 3.2Ghz when setting the computer up. I will check that tomorrow and see if that helps the issue.
 

I will also try to monitor how much power the gpu is pulling before it goes out. One other thing I noticed is the card gets up to p7 and crashes. 

Could just set the max to p6 as well then, might be really bad silicon, or your PSU is for some reason unable to power it.

Don't try the 500W OC for sure though.
 

 

[Clcare556]

12 minutes ago, Streetguru said:

Could just set the max to p6 as well then, might be really bad silicon, or your PSU is for some reason unable to power it.

Don't try the 500W OC for sure though.
 

 

I won’t be doing that lol. On another thread someone suggests use vga to connect monitor to MOBO and basically wipe gpu and install drivers in proper order I don’t know if that would help as well. 

[Streetguru]

1 minute ago, Clcare556 said:

I won’t be doing that lol. On another thread someone suggests use vga to connect monitor to MOBO and basically wipe gpu and install drivers in proper order I don’t know if that would help as well. 

I think every vega GPU has dual bios? You could try your other bios setting too.

 

but you don't need a 2nd GPU to reinstall windows, or even a new GPU bios the bios towards the I/O/back of the PC is the unlocked one.

You don't have onboard graphics to use the motherboard for displays though anyways.

[Falkentyne]

37 minutes ago, Clcare556 said:

I am only running Wattman. I actually was just reading about the whole overclocked ram thing. I did adjust that. I will set to stock and try that. 
 

i would’ve gone the Nvidia route but I bought this card lightly used from a neighbor who builds PCs. 
 

on a side note I still may get separate power cables just to be sure. 

What kind of crash is it?

Black screen?  Or do you get a "GPU crashed and has recovered?"

 

I can tell you first and foremost that if you are NOT using separate power cables (One PCIE 8 pin per power bank plug on the PSU), that is a huge mistake (e.g. if you are using a 2x8 pin connector that hooks up to a single power connector).  I believe the issue isn't the PSU itself, but rather how much current one of those cables can draw since they're split and you're combining two 8 pin inputs into one PSU Port.

.

I tried doing the two 8 pins into one power cable bank on my Seasonic when I had a r9 290X, which supported GPU+12v monitoring, and after some time, it went from 11.75-11.63v under load, down to 11.38v or sometimes as low as 11.25v !  It would not crash stuff like 3dmark or Heaven/Valley, etc, but running light load games (especially League of Legends) or just using the desktop while alt tabbed, would just black screen and hard lock the system randomly (no chance to recover).

 

I had three 2x8 pin cables, so I threw that one away and hooked up one of the two unused ones, and bam: 11.75v idle and 11.63v load, with no dipping to 11.50v.

Then I remembered what someone said about using two separate power connectors for these cards and not using a shared power 2x8 pin for both 8 pins.

So I hooked up a second 8 pin from a solo connector, so now both 8 pins were fed from separate lines and I never had a problem ever again.

 

I suspect you're having the same problem.  And it's worse for you since Vega 64 doesn't support GPU +12v monitoring (that's what told me something was acting wrong on the R9 290X).

[Clcare556]

12 minutes ago, Streetguru said:

I think every vega GPU has dual bios? You could try your other bios setting too.

 

but you don't need a 2nd GPU to reinstall windows, or even a new GPU bios the bios towards the I/O/back of the PC is the unlocked one.

You don't have onboard graphics to use the motherboard for displays though anyways.

That is true I haven’t flipped the switch though. What does that do exactly? I mean I know it swaps bios but how do you determine what that does to help?

[Clcare556]

3 minutes ago, Falkentyne said:

What kind of crash is it?

Black screen?  Or do you get a "GPU crashed and has recovered?"

 

I can tell you first and foremost that if you are NOT using separate power cables (One PCIE 8 pin per power bank plug on the PSU), that is a huge mistake (e.g. if you are using a 2x8 pin connector that hooks up to a single power connector).  I believe the issue isn't the PSU itself, but rather how much current one of those cables can draw since they're split and you're combining two 8 pin inputs into one PSU Port.

.

I tried doing the two 8 pins into one power cable bank on my Seasonic when I had a r9 290X, which supported GPU+12v monitoring, and after some time, it went from 11.75-11.63v under load, down to 11.38v or sometimes as low as 11.25v !  It would not crash stuff like 3dmark or Heaven/Valley, etc, but running light load games (especially League of Legends) or just using the desktop while alt tabbed, would just black screen and hard lock the system randomly (no chance to recover).

 

I had three 2x8 pin cables, so I threw that one away and hooked up one of the two unused ones, and bam: 11.75v idle and 11.63v load, with no dipping to 11.50v.

Then I remembered what someone said about using two separate power connectors for these cards and not using a shared power 2x8 pin for both 8 pins.

So I hooked up a second 8 pin from a solo connector, so now both 8 pins were fed from separate lines and I never had a problem ever again.

 

I suspect you're having the same problem.  And it's worse for you since Vega 64 doesn't support GPU +12v monitoring (that's what told me something was acting wrong on the R9 290X).

I think that may be part of the issue. I’m getting black screen. 

[Clcare556]

Just now, Clcare556 said:

I think that may be part of the issue. I’m getting black screen. 

One thing to note the fan doesn’t go to 100% like some report and it makes a steady faint beeping

[Streetguru]

1 minute ago, Clcare556 said:

That is true I haven’t flipped the switch though. What does that do exactly? I mean I know it swaps bios but how do you determine what that does to help?

The one away from the I/O on the card should be the "quiet" mode which will limit power to save on noise, it's also write protected so you always have a way to recover from bios mods.

[Clcare556]

Ok so I believe my psu came with separate pcie cables. I will install those tomorrow and try to troubleshoot some more. Unfortunately I will only be able to mess with it late night for now. Thank you for all your help and I will try to report back here. 

[Clcare556]

This morning I put 2 pcie cables in the system and checked the ram and while I set one profile to 3200 the other was on 3600

[Clcare556]

Gaming mode was aslo on. Now it’s off. 

[Clcare556]

Ok so I just finished a whole casual match in rainbow without crashing. One thing I did notice however is the gpu would be in state 2 or 3 in game and then all of the sudden fan speeds would go up and it went to state 7 where it would normally crash. 
 

I could not figure out how to manually adjust ram voltage. It’s set to auto currently. I may look into that. 

[Clcare556]

Ok so I do believe that it may be my power supply. I experience crashes when I mess with undervolting. I am getting a power kernel warning in the logs. I’m trying power save settings on gpu right now and made sure that windows power settings were on max performance. Any other thoughts?

[Clcare556]

The power save setting work fine and it has not crashed however temps were around 80c. I have two separate power cables at this time but my psu has only one rail. Theses are my Wattman readings while gaming.