Suspected AMD CPU failure...

[quysang219]

Hello all,

Sorry I put this in General Discussion because at this point I am at a total loss on what went wrong with my rig. I am bringing it to AMD Warranty tomorrow, but before that, I hope maybe some of you could diagnose what has been going on with this PC of mine. I thank you very much for your time to read this, and appreciate all advice or sharing of similar experience that you might have had.

 

So I bought this whole new rig in 4 months ago. Ryzen 5 3400G as CPU with Vega 11 integrated. Gigabyte B450 Aorus Pro as Motherboard. Two sticks of G.Skill Trident Z 3000MHz as RAM. An NVMe SSD 240GB from Lexar. And a 700W PSU from Hydro or Xigmatek X-Power 450W. No dedicated graphics card yet.

 

Everything worked great until the last 30 days when I started experiencing game crashes and blue screens indicating graphics driver issue (Error list below) every time I play 3-20 minutes of game. I was using AMD Adrenalin 20.4, so I followed instructions on their official website to remove the current version and upgrade to 20.7. Problem not solved. I downgraded to 20.2. Problem not solved. Then I thought maybe I should not hold the Windows 2004 upgrade, so I upgraded and tried Adrenalin 20.4 and 20.7. Problem not solved. Out of desperation, I clean-reinstall Windows and tried those Adrenalin version again. Problem not solved. Once during this period of time, I put in the 700W Hydro PSU and RX 570 and got the same problem. I removed the card afterwards.

 

At this point I was like "Ok I'm done with AMD bye." So I clean-reinstalled Windows again (using installation media. After reinstalling, I always check and update to the latest version), and then used Windows' default graphics driver, which also worked fine previously. To my surprise, I still got crashes and blue screens (Error list below). A friend told me maybe my SSD got an issue, so I removed it, got another SSD, installed Windows on it and ran it with my rig: Same issue persisted. Still blue screen.

 

I am so confused right now. Yes the notifications all pointed at graphics driver but after so much testing, I really do not think the problem is with the software, or Windows, or the SSD, or the power supply. My CPU must have got some issue with Cache or some pins that reads graphics driver (?). I am so unsure of this. Below is the list of blue screen errors I have got so far, and some screenshot from my game. By the way, by game I mean all my games: World of Tanks, The Long Dark, League of Legends, Unturned, etc. After 3-20 minutes they just crashed. Once I also cleaned and added thermal paste to my CPU again. It never overheated, so it was not about temperature as well...

 

I hope to receive your kind help and advice. Thank you very much! 

 

List of blue screen errors:

• attempted execute of unexecute memory
• irql not less or equal
• system thread exception not handled
• system service exception
• page fault in unpaged area
• driver overran stack buffer

 

 

[AbydosOne]

8 minutes ago, quysang219 said:

List of blue screen errors:

• attempted execute of unexecute memory
• irql not less or equal
• system thread exception not handled
• system service exception
• page fault in unpaged area
• driver overran stack buffer

Those all sound like BSoDs you get when your RAM isn't quite 100%. Run a memory test (MemTest86 is the one I use) to ensure your RAM is stable. If you have DOCP on, turn it off and try your games again.

[Nena Trinity]

27 minutes ago, AbydosOne said:

Those all sound like BSoDs you get when your RAM isn't quite 100%. Run a memory test (MemTest86 is the one I use) to ensure your RAM is stable. If you have DOCP on, turn it off and try your games again.

Even that isn't a 100% if the memory is simply incompatible with the board... (had that issue before)

[quysang219]

1 hour ago, AbydosOne said:

Those all sound like BSoDs you get when your RAM isn't quite 100%. Run a memory test (MemTest86 is the one I use) to ensure your RAM is stable. If you have DOCP on, turn it off and try your games again.

Crap I think I knew it... You were right. I put my RAM back to 2133 (default) and it ran fine...

But doesn't AMD advertise their CPU to take 2933MHz of system memory frequency? My RAM had the capability to go 3000MHz. Previously I did set it to 3000MHz but the system always reset to 2133MHz, so last month I set it at 2933MHz. It never reset and instead I got the whole mess above.

 

Thank you very much man. I think the problem is clear. I will try with 2666MHz. AMD is so unstable. They advertises 2933 and yet I got this...

[AbydosOne]

22 minutes ago, quysang219 said:

Crap I think I knew it... You were right. I put my RAM back to 2133 (default) and it ran fine...

But doesn't AMD advertise their CPU to take 2933MHz of system memory frequency? My RAM had the capability to go 3000MHz. Previously I did set it to 3000MHz but the system always reset to 2133MHz, so last month I set it at 2933MHz. It never reset and instead I got the whole mess above.

 

Thank you very much man. I think the problem is clear. I will try with 2666MHz. AMD is so unstable. They advertises 2933 and yet I got this...

Are you manually setting the RAM speed, or using DOCP? Use the DOCP setting first, then adjust the speed so that the timings and voltages are right.

Ryzen is picky about RAM and timings/frequencies, so you may need to invest some time into dialing it in.

[quysang219]

31 minutes ago, AbydosOne said:

Are you manually setting the RAM speed, or using DOCP? Use the DOCP setting first, then adjust the speed so that the timings and voltages are right.

Ryzen is picky about RAM and timings/frequencies, so you may need to invest some time into dialing it in.

I don't know DOCP until you mentioned. When my pc boot I press Delete, then i go in the bios (Gigabyte b450 aorus pro). There is an option to turn on profile 1, which brings the freq to 3000MHz but next time my pc boots it would go back to 2133. So next to that option there was a number. I just changed it by pressing +, so that i reaches 2933. was that docp or xmp?

Update: My games run stable at 2666 now. I never changed anything related to voltage so maybe I should now? :( 

[AbydosOne]

51 minutes ago, quysang219 said:

I don't know DOCP until you mentioned. When my pc boot I press Delete, then i go in the bios (Gigabyte b450 aorus pro). There is an option to turn on profile 1, which brings the freq to 3000MHz but next time my pc boots it would go back to 2133. So next to that option there was a number. I just changed it by pressing +, so that i reaches 2933. was that docp or xmp?

Update: My games run stable at 2666 now. I never changed anything related to voltage so maybe I should now?

You should (you can on Asus, so I assume this to be true) be able to set the DOCP setting to Profile 1, and then decrease the frequency until you get to stability. The DOCP profile has the voltage already in it, which is why it should be loaded first.