Yesterday(13/06/2020) I was gaming while watching youtube on my second monitor when I suddenly had a BSOD with the Video TDR Failure -- amdkmpag.sys error. At one point during the BSOD cycling the error code changed to amdkmdag.sys with the same Video TDR Failure.
After this happened my windows 10 pro became unaccessible due the same BSOD crash as soon as my computer reached the windows pin entry screen. I have used safe mode to remove my graphics drivers(with DDU) which is what was suggested in the fixes online but as soon as I attempt to reinstall the drivers I experience the same crash again... I can load into normal windows as long as I don't install my graphics drivers which I need to get my 2nd monitor working and get performance out of the GPU obviously.
I have dowloaded the .sys files from the net and placed them into my win32 folder incase they were corrupted somehow but this hasn't helped at all.
I have attempted to use multiple driver versions from the driver page including; Adrenalin 2020 Edition 20.5.1 Optional, Adrenalin 2020 Edition 20.4.2 Recommended (WHQL) and I also gave the Radeon™ Pro Software for Enterprise a crack just incase. Prior to my current issue I was using the Adrenalin 2020 Edition 20.4.2 Recommended (WHQL) driver.
I have also tried SFC /Scannow in cmd but no errors are found.
I have made sure all power cables are seated correctly, checked the graphics card hasn't moved somehow etc
My computer is less than 2 weeks old, I had it running completely fine running a variety of applications/games before encountering this problem yesterday. I have not as yet overclocked anything. Very frustrating to say the least.
PSU: SilverStone ST75F-GS 750W Strider Gold Power Supply
System SSD: WD Green 480GB 3D NAND 2.5in SSD x2 in Raid 1
OS: Windows 10 Pro
Thanks for reading.
Further to the above I have also added the following attempts to resolve my issue;
I have completely formatted my system drives and am running a fresh windows 10 environment however the problem persists, my thinking was that there may have been a corrupted file that wasn't being removed by DDU but this appears not to have been the case.
I have reslotted my GPU into a different PCIEx16 slot on my MOBO.
I have made sure my PC is configured to use maximum power settings.
I have started my computer with only my main monitor attached.
I've been checking a lot of guides on youtube etc but thus far none have worked.
I'm running out of ideas here??? So much frustrated sadness.
EDIT: Just now tried to use the auto detect hardware driver installer and received an error 182 - Radeon Software installed detected AMD graphics hardware in your system configuration that is not supported with this software installation. This didn't happen the first time I tried the auto detect route.
My second reply to my original post;
I have tried using Ubuntu to diagnose through Linux however the GPU artifacts/green screens in Linux, which it doesn't do in an Windows environment interestingly.
I swapped in my old Strix GTX 960 4gb and my computer runs perfectly again. This combined with the above leads me to the point I'm sure the MSI RX 580 card is a dud and will be returning it to get a replacement.
Humorous reply this time, I called the Australian MSI repair center in order to start the warranty system for my GPU yesterday. The woman who answered the phone there didn't know the difference between a graphics card and a motherboard and got them mixed up three times during the ~4minute call. I lost faith and went back to the company I purchased the card from despite them saying multiple times it'd be faster dealing with MSI directly. Not sure how someone working for a computer repair center doesn't know the difference between a motherboard and GPU, like even as a receptionist you'd think they'd know better given the amount of calls they'd get. Anyway was a good laugh after the stress of it all.
I hope no-one else is suffering from the issues I outlined above, I've personally built 3 systems over the years for myself and 8 for friends. My first two were overclocked to the limit, I had my 2600k running stable at 4.8ghz(from 3.4ghz stock, I've seen people running them over 5ghz with water cooling) with a Noctua NH-D14 air cooler, shes a big girl but she puts the work in! I got nervous at 4.8ghz though and lowered it to 4.5ghz despite it being stable in prime95 after a 6 hour burn in.
Absolutely incredible CPU, which I had problems finding something better to replace it in my new build despite the 8 year gap from when I bought it to when I upgraded just recently.