5700XT possibly crashing system.

[idiocracy]

Have a red devil 5700xt, and have had some stability issues. I'm trying to figure out what exactly is going on, atm i'm not even sure it's the gpu. I am using the silent 180w bios.

Other than that i have a 2600x and a rm750. The gpu is connected with one cable from each rail.

 

What happens is that randomly my computer will crash, this can be under any load. Most times i'll get a black screen, audio goess brrr wrrrr urrrr and then it shuts down. It doesn't reboot, it shuts down while still being on. Screen goes to sleep but fans are still spinning. On occasion it'll recover to desktop and the application just crashes.

 

I had gpuz record a log.

Had a massive lag, gpu drops to 805/806mhz, memory to 200mhz, load to 0%, memory controller to 1, chip power draw to 9. Every other metric looks normal. The "lag" was present for 4.5 secs before it continued, was playing a game that utilises ~70/80% at ~1300mhz, ~65/90w of power draw. So the system was by no means under stress during this. It happens what appears randomly, can play under full load, half load for hours, doesn't appear to matter. Drivers doesn't report they've been reset after a crash.

Hwinfo reports 12.091v on +12v, and this drops to 11.933v during light gaming, i do not know if this is significant. +5v is always slightly above 5.

 

Windows logs noticed nothing during this, while a full crash only get's a "system reboot was unexpected".

 

Any suggestions?

[Gohardgrandpa]

I’m sorry for you. I had the same issues with a sapphire pulse 5700xt. I did a rma on the card and the new one did the same thing. 
I put my 2070 super back in the pc and had no issues. Put the 5700xt in and the same issues you have. I swapped ram from 2x16gb to 2x8gb, swapped psu from Corsair rm850x to a evga supernova 800 watt and same issues. 
 

I tried every setting I could think of in the amd software package. What worked best for me was installing the gpu drivers without installing the software package and even then I still had issues. 
 

I tried the gpu in my friends rig and it worked great. Literally the same components as mine. I sold him the card for $300 and put a 2080 super in my pc and no issues. Sold the 2080 super and put a evga 3080 in that pc and haven’t had an issue either. 
 

I used DDU numerous times to make sure it was a clean install. Just nothing but trouble with that model for me 

[idiocracy]

One thing i havn't tried is running only the drivers without the software package: https://community.amd.com/thread/246601

I'm trying that now and will give it a few days as it's too soon to say anything after just 5 hours.

[Gohardgrandpa]

I really hope that works for you. When the card works it’s great performance wise. Unfortunately I had nothing but issues and ended up selling the card for $100 less then I paid for it 

[BTGbullseye]

It sounds like you're either running very old drivers, (or didn't DDU before updating to one of the newer drivers) an unstable overclock, or your PSU is failing.

[idiocracy]

I tried without the software package and that didn't seem to help. Just had a hard crash where pc just shut down without shutting off. Earlier today i had a thing where the gpu went -1 in task manager, desktop was somewhat responsive, a video was playing which started lagging. Couldn't do anything besides reboot.

 

I also opened my pc and checked the cables was properly plugged in, i even switched them around to different ports just to see if that would do anything. Also checked inside my psu just to see if there was any dust. The fan only spins during high loads and the psu basically looks new even after 6 years, which means it's not really being stressed to a point where it even needs to turn on it's fan.

 

Power colours service is a customer nightmare, so i'm hoping the issue isn't the card. But not really sure what else to do.

[Mark Kaine]

On 10/4/2020 at 4:56 PM, idiocracy said:

Any suggestions?

List full system specs for a start. 

[-rascal-]

2 hours ago, idiocracy said:

I tried without the software package and that didn't seem to help. Just had a hard crash where pc just shut down without shutting off. Earlier today i had a thing where the gpu went -1 in task manager, desktop was somewhat responsive, a video was playing which started lagging. Couldn't do anything besides reboot.

 

I also opened my pc and checked the cables was properly plugged in, i even switched them around to different ports just to see if that would do anything. Also checked inside my psu just to see if there was any dust. The fan only spins during high loads and the psu basically looks new even after 6 years, which means it's not really being stressed to a point where it even needs to turn on it's fan.

 

Power colours service is a customer nightmare, so i'm hoping the issue isn't the card. But not really sure what else to do.

 

What version of the drivers did you download / install?

As of current, there is a WHQL certified version, and a beta/hotfix version.

 

At least for my 5700 XT, I was getting game and driver crashes every so often.

I was on a driver release from April/May 2020.

Just installed the WHQL September 2020 driver last night...seems to be better.

 

Anyways, once you get the full suite installed, go into Radeon software --> Tuning menu, and check what the P3 state voltage is.

That is the boost / target boost state. From what I know, the default uses max voltage (e.g. 1200 mV). Tuning the voltage down a little bit *might* help.

Say down from 1200 mV to 1175 or 1150 mV. Plus, that will help bring temperatures down. Again, this will depend on the card.

If you can upload a screenshot, that would be best.

 

This is the default / stock for my card ... 2064 MHz, 1200 mV for the P3 state.

 

Also, what programs / games are you running that will trigger the problem?

[idiocracy]

50 minutes ago, -rascal- said:

 

What version of the drivers did you download / install?

As of current, there is a WHQL certified version, and a beta/hotfix version.

 

At least for my 5700 XT, I was getting game and driver crashes every so often.

I was on a driver release from April/May 2020.

Just installed the WHQL September 2020 driver last night...seems to be better.

 

Anyways, once you get the full suite installed, go into Radeon software --> Tuning menu, and check what the P3 state voltage is.

That is the boost / target boost state. From what I know, the default uses max voltage (e.g. 1200 mV). Tuning the voltage down a little bit *might* help.

Say down from 1200 mV to 1175 or 1150 mV. Plus, that will help bring temperatures down. Again, this will depend on the card.

If you can upload a screenshot, that would be best.

 

This is the default / stock for my card ... 2064 MHz, 1200 mV for the P3 state.

 

Also, what programs / games are you running that will trigger the problem?

 

Running driver 20.9.2. Can try different version, but this is not a new issue.

The default P3 voltage is 1.2v. Changing volt has been tried earlier, now everything is stock, except for 2150mhz boost limited to 2100mhz. Did that because i thought that might've been involved. But as mentioned it also happens when light gaming around 900/1300mhz.

I've tried undervolt but had nothing but issues with that, so just assumed i lost the lottery.

 

Before 5700xt i had a 1070 and had no issues with that.

[-rascal-]

9 minutes ago, idiocracy said:

 

Running driver 20.9.2. Can try different version, but this is not a new issue.

The default P3 voltage is 1.2v. Changing volt has been tried earlier, now everything is stock, except for 2150mhz boost limited to 2100mhz. Did that because i thought that might've been involved.

I've tried undervolt but had nothing but issues with that, so just assumed i lost the lottery.

 

 

Okay 20.9.1 is the WHQL driver, while 20.9.2 is the optional/beta driver.

Your card's vBIOS boost target is a bit more aggressive than my Sapphire NITRO+ 5700 XT.

You might be stuck with the ~1.200V.

 

What games / applications is triggering the crash?

 

Quote

What happens is that randomly my computer will crash, this can be under any load. Most times i'll get a black screen, audio goess brrr wrrrr urrrr and then it shuts down. It doesn't reboot, it shuts down while still being on. Screen goes to sleep but fans are still spinning. On occasion it'll recover to desktop and the application just crashes.

 

On the old April/May 2020 drivers, I was getting this exact symptom on OverWatch.

Black screen (indicating driver crash), sound / game is still running, game than crashes afterwards, drivers crash and may/may not recover.

If not it just stays at the black screen with sound, and only manually rebooting fixes the problem.

So far, it is only OverWatch that triggers this.

 

Anyways, is this the same system in your profile?

When did you switch from a GTX 1070 to 5700 XT?

During the switch-over, did you run DDU to remove any left over junk from the nVidia drivers?

https://www.guru3d.com/files-details/display-driver-uninstaller-download.html 

[idiocracy]

47 minutes ago, -rascal- said:

 

Okay 20.9.1 is the WHQL driver, while 20.9.2 is the optional/beta driver.

Your card's vBIOS boost target is a bit more aggressive than my Sapphire NITRO+ 5700 XT.

You might be stuck with the ~1.200V.

 

What games / applications is triggering the crash?

 

 

On the old April/May 2020 drivers, I was getting this exact symptom on OverWatch.

Black screen (indicating driver crash), sound / game is still running, game than crashes afterwards, drivers crash and may/may not recover.

If not it just stays at the black screen with sound, and only manually rebooting fixes the problem.

So far, it is only OverWatch that triggers this.

 

Anyways, is this the same system in your profile?

When did you switch from a GTX 1070 to 5700 XT?

During the switch-over, did you run DDU to remove any left over junk from the nVidia drivers?

https://www.guru3d.com/files-details/display-driver-uninstaller-download.html 

 

I switched like back in february, i ddu'd everything. vbios has two limits. Overdrive limit which is 2150mhz, and frequency max which is 2100mhz.

 

I get this in any game so far, world of tanks, horizon zero dawn, snowrunner, detroit become human. Are the games i've been playing lately. World of tanks doesn't clock higher than ~1500mhz and mostly stays around 900/1300mhz.

 

I don't get audio playing in the background, i think i might've experienced that once. My system hard crashes, i can hear that when audio goes beserk and screen goes black. Audio doesn't always go beserk, sometimes screen just goes black but no audio every plays in the background, if there is audio it's killed.

[idiocracy]

Just did 3 reboots into safe mode, 1. ddu nvidia, 2. ddu intel & 3. ddu amd. Then rebooted normally, manually updated drivers with 20.9.1, no software package. We'll see if that does anything.

 

Edit:

One weird that happened is that now my gpu fans aren't spinning, at all. Card got super toasty and found fans were 0 rpm. While vbios states that start temp is 60c, and target temp is 70c. At +80c fans were 0 rpm. That's odd. Fan wasn't detected in monitor apps as well, but as soon as i installed the software package, they spun. Weird. So it seems the software package is required.

 

[idiocracy]

Ok, so since i did all that it has been running. It ran when i just installed the drivers, however the fan didn't spin, but it ran. With the software package installed it has not crashed yet either. Today i've played ~6 hours of horizon zero dawn. Now i'm not getting my hopes up.

But maybe, just maybe, ddu'ing all 3 might've fixed the problem, but i'll wait at least another day before i'll let my self be convinced.

[idiocracy]

I'm wondering tho, what exactly does the soc voltage affect? On red devil it's 1050.