[porschedrifter]

So I've been experiencing this issue lately, and I'm pretty sure I narrowed it down to my GPU, where upon wake from sleep/hibernation or turning system on, at sign in screen or after logging in in about a minute or so, my system screen will go black, gpu fan will go to 100% and system will be unresponsive. If I hit the reset button, I'll hear the mobo post and beep once, but gpu fan will still be at 100% and displays will still be black. Until I turn off the pc with power button and power on again, it'll boot back up fine, system will otherwise operate as usual. No BSOD can be found in log.

I have also seen this same lock up when turning on or off HBCC memory option.
Anyone else experiencing this?
This happens on stock clock and stock voltages
17.11.4 driver version.
Gaming wise, temps are fine and no crashes or artifacts while gaming at all.

 

[porschedrifter]

Nada? No one?

 

[The EX1]

Vega has a sleep/wakeup bug where the driver fails to bring the GPU out of the ULPS mode. That usually leads to the driver crashing which explains your 100% fan speed and screen blackout issue. Disable sleep/hibernation mode or get in the habit of shutting down your PC until the issue is finally fixed.

 

[porschedrifter]

Quote:

Originally Posted by The EX1

Vega has a sleep/wakeup bug where the driver fails to bring the GPU out of the ULPS mode. That usually leads to the driver crashing which explains your 100% fan speed and screen blackout issue. Disable sleep/hibernation mode or get in the habit of shutting down your PC until the issue is finally fixed.

Great, thank you! What about disabling ULPS?

*Edit* Disabling ULPS completely breaks hibernation and will BSOD so that's a no go.
Currently investigating if MSI Afterburner or HWinfo is causing the freeze.

 

[porschedrifter]

*Edit* Issue may be solved:
Quote:

Originally Posted by Mumak

Don't know if this has been already noticed here, but it seems that AMD has finally unblocked I2C access to VRMs on Vega.



Not sure if it's since Crimson 17.11.3, but with 17.11.4 it's surely there.
So now you should be able to see all GPU VRM details in HWiNFO including temperatures for both GPU Core and Memory VRMs.
Unfortunately some users started to experience crashes here and it happens when HWiNFO tries to access those VRMs via I2C. Some GPUs seems to be affected, some not. Don't know yet why is this..
If you're experiencing such hard crash during HWiNFO startup, then disable the "GPU I2C Support" option.

 

[hyp36rmax]

I have this issue right now with my test bench VEGA 64 LE. Any solution?

AMD 1700X, ASUS Crosshair HERO VI, Gskill 16GB DDR4 3200MHz CL14, VEGA 64 LE, Seasonic 1200 Watt XP3 Platinum PSU Current 18.4.1 Adrenalin Drivers.