I have noticed my games crash after some time, the screens freeze and I get the message that drives timed out.
It started happening randomly while I was recording some game footage (Sea Of Thieves) on highest details and it crashed. I then played for some time with replay on and it also crashed after about 30minutes. Will try playing without the replay on but it's not solving the problem.
GPU: MSI RX 590 8GB OC
CPU: Ryzen 5 5600x
MOBO: ASUS TUF GAMING B550-PLUS
RAM: Crucial 16GB (2x8GB) 3600MHz CL16 Ballistix Black
CPU Cooling: SilentiumPC Fera 5 Dual Fan 2x120mm
System drive: Promocja: Samsung 500GB M.2 PCIe NVMe 980
Other drives are 1TB HDD and 240ssd
PSU: be quiet! Straight Power 11 650W 80 Plus Platinum
Case has a good ventilation with 6 fans in.
Temps in game are:
CPU +/- 63 celsius
GPU +/- 75 celsius
I have tried turning off image sharpening, that didn't do it
It's happening on both of the versions
One problem is that if you reinstall the Radeon driver after DDD using the network, the AMD Microsoft driver can automatically intervene via the device installation setting and Windows Update. Therefore deactivate everything. You can also deactivate the two AMD services with msconfig. The adrenaline software is too confusing, the Radeon driver probably tweaks the graphics settings of the card. That is why the crash occurs again after occasionally stable gaming, only changing parameters or a clean reinstallation helps. After a crash with restart, Radeon restores the default setting, but varies the core voltage.With my RX5700XT and currently Cyberpunk 2077 with the driver only 21.10.2, after massive problems I first maxed the graphics card to the limit, after 2 days then again ups, and now with a core clock instead of 2100 to 1350 and core voltage from 1200 to 1050 with game graphics Medium and 2xMSAA with 96 FPS tested. The higher the MSAA, the lower the FPS, at 0x FPS 170, 4x FPS 60, 8x FPS 30. The graphics machines of the games with the Radeon driver are often not compliant, even if the advertisement at the start of the game suggests this.
Try a clean install of 21.10.2, it has been mostly fixed in 21.10.2 as can be seen in the release notes. It is the combo of GCN cards and Ryzen CPUs / Motherboards; furthermore, older Radeon users who have been dropped from support in 21.5.1/21.5.2 are plagued with this problem permanently.
Hope it helps. It did for me.
I did, but not a clean install and it crashed one again. Did you turn off any settings in the radeon software?
Most likely will try to hold on to this card until black friday and buy an rtx ;c
But your description says you are using 21.10.1? I am referring to 21.10.2 which released not too long after.
No the settings that I use are globally: sharpening, virtual super resolution, freesync.
But, I usually turn sharpening off per game except if it has strong temporal aliasing in the game. Furthermore, I do not record often.
You can try to check if you are thermal throttling.
Thanks a lot, good point I was using different ones. Just updated, removed any dust from the card and run a 5min OCCT test and the max temp was 74 and it didn't crash so for so good. Will update you in a while with the results. c:
Still crashes with the recommended drivers. I have no clue that to do now
@Naniewiczit is going to crash if you leave OCCT running. Our Polaris cards get way to hot with an OCCT test, something I tried to explain on another post where OCCT was also used.
This years Radeon drivers for RX 400 - RX 500 cards get extremely hot when running OCCT, to the point where it thermal throttles and locks up the PC.
EDIT: I am assuming you left OCCT running for 3Hours? Not good for your card in my opinion. Thermal paste is going to dry up quicker and leave more parts of the GPU chip unprotected, which results in you seeing an average of say 78C, but likely with clock-speeds and power usage jumping all over the place and then finally a hard-lock. OCCT results in my RX 480 using 145-150W instead of the default 110W. I have tested with a lot of very demanding GPU games, and the highest my power usage went was 120W which is a realistic gaming scenario. OCCT pushes it way past that and results in unbalanced power delivery and the fans not being able to spin at proper speed.
EDIT 2: Ok but was this gaming shortly after OCCT? I remember benchmarking with OCCT a year ago, and then gaming 10min later and the computer still locked up since certain components are still too hot after an OCCT run. I don't stress test with it anymore, I use Heaven Valley or some of 3DMarks stress tests.
To clarify - it crashed during regular gaming
Sorry I edited to my comment, since one can only post every 10min. Like I said, I stress test with "3DMark" tools or "Heaven Valley". They only push loads to what max load gaming scenarios would and prevent the system from locking up / overheating, except if airflow or cooling is really bad; in this case, you know something is wrong.
I do not know if you have updated your motherboard BIOS already, since I did this and clean installed 21.10.2.
But I would suggest letting your PC cool down until the point the GPU fans hit 00%, then shutdown your PC for 20seconds and try gaming again without doing an OCCT stress test.
So I narrowed down the problem. Driver timeout happens when I'm either streamning the game to my friends or discord (didn't try OBS) or when I'm recording with the Radeon Software. While recording with xbox game bar / streamlabs everything seems fine. I was also having some PC shutdowns but it was related to bad headphones drivers (Hyperx cloud II)