I just singed up here, because I got tired of some issue with my R9 Fury card:
The card keeps overheating in stock settings, resulting in black screen and driver recovery - even when playing the most mediocre games like Minecraft or Starmade. I used to workaround this issue by adjusting the fanspeeds in Wattman. After adjusting the fanspeeds, the card is stable - no more crashes due to overheating. But Wattman keeps reverting them to stock setting as soon as something unexpected is crashing the drivers - so I'll have to readjust Wattman every time - and most of these times, I forget to readjust. This is very annoying! Besides being annoying this isn't very healthy for the card on component level - because hot components will age faster, than slightly cooler components. IMHO the smelly stench of planned obsolescence is in the air.
Is there a way to permanently increase the fanspeeds?
- the card is clean, no dust is blocking the cooler.
- the tower is well ventilated, there is no heat build up inside. (4 Fans - 2 front, 2 backside)
Describe your system:
Clear that DxDiag and fill in INFORMATION REQUIRED WHEN POSTING A QUESTION | Community as well as a GPUz shot showing the temperature at the time of recovery. If it's not hitting 100°C, then it's not overheating, and your problem is caused by something else, be it another card defect, or a faulty power supply.
Added in that form you linked me. I'll try to force the error with some benchmark tool - because I'm short on time around Easter. I'll add these GPUz shots later on.
Target Temp seems to be around 80°C with 26% fan speed. I don't see, why target temp has to be so high in stock configuration - I don't mind the additional fan noise when the temps could be around 45°C at 100% fan speed on the same load level - without any issues.
Just to clear something up.
Is the actual AMD driver crashing or just Wattman crashing and then restarts with the default settings?
Don't know if you read this about AMD Wattman from AMD Support.: How to Tune GPU Performance Using Radeon Wattman and Radeon Chill
I went to Gigabyte Support for your Motherboard : GA-890FXA-UD5 (rev. 2.1) | Motherboard - GIGABYTE Global and the latest BIOS it shows from 2010 is Version F6.
It seems, that the driver is crashing - but without a "driver has been restored" message:
- The screen freezes
- After 4-5 seconds the screen goes black for around 10-15 seconds.
- Then the screen reappears.
- Sometimes the game keeps running, sometimes the game has crashed. Depends on the game - and how it handles the driver reset.
Something I've also noticed ist, that the displays won't turn back on instantly from screen saver (turn off) after such incident.
BIOS version isn't public. Got it from the support guys to fix some raid ROM issues. (Updated RAID ROM Version)
Try disabling Fast Start. I have heard it causes a lot of problems. This is usually enabled by default under Control Panel - Hardware and Sound - Power Options - System Settings.
Prevent your computer from going into sleep or Hybrid sleep and see if that helps. If not then enable it again.
To clear things up, because it seems that you understand me wrong:
The delay just affects the screen to reappear after the screensaver turned off the monitors and after the said issue happened - The PC keeps running in idle. No Energysave, S3 or else. I disabled it.
When I do a full restart, everything is working fine again.
Things cleared, Thanks.
In short, what is happening is that the GPU seems to overheat while under heavy load because Wattman doesn't keep the customized settings on the fan after the GPU driver crashes or restarts. If Wattman's customize fan settings are set and working, your GPU doesn't crash anymore?
The Driver crashes only during Game play or anytime the GPU is stressed? Do you have any 3rd party OC software installed like MSI Afterburner, Asus Tweak II, or Trixx?
Black Zion mentioned about temperature. What temperature does your GPU show when it crashes?
So the MAIN ISSUE is that WATTMAN doesn't stay with the Customized FAN SETTINGS after a reboot or restart for whatever reasons. According to your first post, your GPU card doesn't crash anymore with Wattman's customize fan settings.
If I am repeating myself, sorry, I am trying to get a clear picture of when your GPU crashes and what happens afterwards. Maybe someone else might offer better tips on troubleshooting your problem with a clearer understanding of the problem.
The driver will be restored to default setting after each crash or installing new driver. They give you the ability to save your custom settings to a profile. So you can just reload the profile you made instead of having to do all settings again. Hard to remember those settings I know. When you save the profile you want, you can save it again to your documents folder too. In case you do an uninstall it wipes out your profiles.
If it is returning to defaults every time you boot, make sure to disable Fast Startup (windows 10 only) as already said. This is a good idea anyway. A few extra seconds isn't worth loading in old information every time.
Make sure to set you power limit slider to the right as far as it will go. Maxed for your card. Should help avoid those crashes to begin with.
Good Luck & Happy Gaming!
Disable any form of link state between you card & in your bios setting can mess with card reading temps properly limiting power taxing it. Another idea is disable any form of power savings on your OS change all settings to High preformance disable any hybrid sleep state settings as can cause unstability issues to seen it before. Another idea could be bugged drivers which could be throwing a bug in they system throwing everything out. A suggestion would be roll back to the most stable whql driver which worked great run with that & do a stability test, leave if for a week on the most stable driver your system was running happily on then wait for the fresh whql drivers which be due soon. I suspect could be driver bug between OS & driver side. Another thing is pull your card give it a decent blow out even give it a blow out repaste. Run your build without card for a bit then put it back in & run on the most stable older drivers for a week till you know your build is running stable. If running any 3rd party software completely wipe it as will cause problems between AMD drivers see it alot. Your PSU to me doesn't sound suspect tho can get PSU wattage apps to monitor your voltage & wattage across your build which be prety accurate guide if your PSU is faulty. I suspect could be bugged driver from OS end more so going of your description. Bit more info when you get a chance be handy.