whenever in modern games I encounter artifacts like this:
+ the worst is the constant crashing/freezing. it goes like this
game freezes with black screen while audio gets stuck in mini loop
->crash to destop
->the error line - Display driver stopped responding and has been recovered
->Default wattman settings restored due to unexpected system failure
->with radeon settings stuck
message: Host application has stopped working.
It happens from 30 seconds of game to 30min at max. on average 10 minutes.
In FFXV its rather random but in first city Lestallum its unbearable to point where I can play maybe for 3minutes max, with 30 seconds being shortest time. I know that game is riddled with problems but seems my case is extreme.
Game runs at rather modest settings on fps capped at 30 in 1080p. same with games below (except on those I cap at 60fps)
Happens really often in NIOH too and with less frequency in NFS Payback and Ori and Blind forest or other games.
Doesn't really happen in benchmarks but that maybe just because they ran short time.>
Now im running it on default with temps stable at 75C, I also tried more aggresive temp profile where temps were under 50C but it crashed same as on 75C.
Currently on 18.3.4 drivers, but tried 16.11.5,17.11.1,18.2.1,18.3.2 but it didnt made any big difference.
I Think I exhausted all the fixes I could find here and on the internet, did memtest, prime, tinkered with vBIOS switch, wattman fans, underclocking, voltage settings, tdr registry... only thing I cant do is try swap parts because I Dont have any.
So like the question stands can this be fixed with drivers or some sw change or is this HW fault that needs an RMA of card?
Download OCCT (Free diagnostic software) and run the GPU test with GPU ERRORS box marked. This will check your GPU under stress and also check to be sure your GPU isn't producing any errors while under stress. I have read that sometimes poor power to GPU card will produce artifacts. With OCCT you can check your PSU outputs to make sure they are within range. This should eliminate the GPU and PSU Hardware, at least, initially.
By the way, Sapphire has very poor warranty (two years) and very difficult to RMA.
How long should I run it for some proper testing?
Also not sure how warranty in regards with Sapphire works here (europe) since I rma it at shop that sold me card and they are dealing with it. luckily I still have 3/4 year of warrant left.
Bit sad that I only know of Sapphire things now, from what I read before buying it was all EVGA of AMD and other praise for them.
Generally you run OCCT GPU test with Error checking for about one Hour at FULLSCREEN.
Don't check your PSU with OCCT unless you have a high quality PSU. I found out that cheaply made low quality PSU can be damaged running OCCT PSU test. If you run it and as soon as it starts, the computer reboots, than you may want to take a closer look at your PSU. if the PSU test starts and runs, keep a very close eye on it to see it is not being stressed to the point where it might get damaged.
Otherwise, while running the GPU test, OCCT shows you all PSU Outputs during the testing. So, if you see anything strange in the PSU outputs, than you can try running OCCT PSU test. But be careful and watchful while doing so.
Someone else here in the Forum purchased a Sapphire GPU card in Europe, since he quoted in Euros rather than Dollars. Here is what he had to say about the Sapphire warranty after his card went bad 1.5 years after purchase:
I dont think it overheated, I was only browsing the Web... The cooler on that card is so massive that the Fans are stopped by Factory on low loads and Temperatures below 60° (or something around that area).
PSU could be another thing, but I dont have another PSU laying around with 2x 8-Pin PCI-E connectors
I already contacted Sapphire, they just got back to me. They said they wont do the RMA and want me to send the card to the Store where I bought it. The Store told me that they would send that card in for Repair, but if they cant repair it I would get the Time Value of the card... Which is only 150€, from that money I cant get anything close to a Fury performance wise
Here is the recent thread about the other Sapphire GPU card from above comment: Sudden death of Sapphire R9 Fury?
Have you tried increasing your power limit slider under Wattman to it's maximum? Move it all the way to the right likely either +25 or +50 for that card. Just max it.
The "game freezes with black screen while audio gets stuck in mini loop" is often fixed with this simple setting. My RX 580 did the same thing. Make sure your power settings for Windows are on High Performance too.
Did that for hour, 0 crashes 0 errors, though the graphs of +12 and others looks bit weird. not sure if its supposed to be like this with those dips, also the test was static
image not sure if its correct or not.
In the meantime I tried reflashing the vbios thats comes with card but its still same crashes in game.