What the heck is wrong with my build?

[Curio]

Of all the machines I've built down the years, this one is by far the most unreliable and biggest pain in the caboose. It just constantly CTDs. Some days it doesn't do it at all. Other days it's every few mins.

Started life with a Radeon 570 which I figured was bad so got a 1650 Super. Still had issues. Since then it's had a new PSU (Seasonic 650w), new RAM and new mobo. Still happens.

Not overheating. Reinstalled Windows. Tried every video driver known to mankind. Chipset drivers. BIOS versions. Data cables. Power outlets. Power cables. Surge protectors. Every setting in Windows and video drivers I can find.

Worse Still, event logs show NOTHING in application or system and there are no errors (usually) when it crashes. If there is an error it's usually the video driver crashing.

The only clue is that it nearly always seems to crash when loading a game level or just after a game has loaded. It rarely crashes while actually playing. The only thing I can think of is a bad SSD. I have 2 SK Hynix Golds - one for OS and one for installs. They check out fine with SMART tools and various stress tests. But it's the only thing left I haven't swapped. Or the Ryzen CPU, which seems an unlikely suspect.

Anybody got anything I could try? Known issues I'm unaware of? Thanks!

 

[Chris Beard]

Of all the machines I've built down the years, this one is by far the most unreliable and biggest pain in the caboose. It just constantly CTDs. Some days it doesn't do it at all. Other days it's every few mins.

Started life with a Radeon 570 which I figured was bad so got a 1650 Super. Still had issues. Since then it's had a new PSU (Seasonic 650w), new RAM and new mobo. Still happens.

Not overheating. Reinstalled Windows. Tried every video driver known to mankind. Chipset drivers. BIOS versions. Data cables. Power outlets. Power cables. Surge protectors. Every setting in Windows and video drivers I can find.

Worse Still, event logs show NOTHING in application or system and there are no errors (usually) when it crashes. If there is an error it's usually the video driver crashing.

The only clue is that it nearly always seems to crash when loading a game level or just after a game has loaded. It rarely crashes while actually playing. The only thing I can think of is a bad SSD. I have 2 SK Hynix Golds - one for OS and one for installs. They check out fine with SMART tools and various stress tests. But it's the only thing left I haven't swapped. Or the Ryzen CPU, which seems an unlikely suspect.

Anybody got anything I could try? Known issues I'm unaware of? Thanks!

Sounds pretty thorough to me, very odd you're not getting anything in event viewer. Just to check - when changing graphics card drivers are you using DDU please?

 

[HoraceC]

Sounds like a possible compatibility issue. Have you checked installed CPU model against the motherboard approved list?

 

[Joxeon]

Try turning off the CPUs boost clock it's called "core performance boost" in bios to rule that out.

 

[Finners]

I'm also leaning towards the Ryzen CPU, it's behaving very much like mine does if I play around with PBO. Crash to desktop rather than a blue screen or anything else. If yours is the 1600AF like in your sig then that's a bit of an odd ball CPU wasn't it? Released when the 2000 series came out but given the 1000 series name.

Is it all games or just one?

CPU probably isn't faulty, just a compatibility thing with your mobo that it's boosting just a little to high for a split second and crashing. Or PBO is active by default and causing it.

What motherboard do you have?

 

[Tetras]

What stability testing have you done with the cpu and memory? Do you always run with XMP enabled?

What monitoring, or other software do you have?

Does this happen in all games, or are some games immune?

If it is an old game, have you updated your DirectX install?

 

[Curio]

Thanks for all the suggestions, chaps.

With the new mobo (ASUS B450) I did make sure both the Ryzen and the exact RAM model were on the QVL. I will try messing with some of the CPU settings in the BIOS though.

I did forget to mention that I don't overclock this rig at all. In fact, I've even underclocked the video card a little. The RAM has been tried at both default settings and XMP profiles.

It's a frustrating problem because sometimes it can be stable for a while so it looks like I found the issue. Grrr.

Worst game by far is SW Battlefront 2 - some days that crashes almost constantly when loading in a map. Doom Eternal I don't think has ever crashed that I can remember. Everything else crashes out at least once per session. But again, some days are better than others. I'm starting to think it depends which way the wind is blowing...

 

[cimurray]

Bit of a long shot as you seem to have done all the usual stuff. As your board is an Asus does it have a "PBO Fmax Enhancer" setting in the BIOS? If so try disabling it as it is nothing to do with AMD and it overrides AMD's defaults and sometimes pushes too much power into the CPU.

 

[Chris Beard]

I'm starting to think it depends which way the wind is blowing...

How good is the wiring in your building? If you really have tried everything else it could literally be which way the wind is blowing (or other environmental factors) affecting the power at the wall.

I don't propose taking this step until you really have tried everything else (and preferably have tested another known to be good PC from the same wall socket), but you could consider an Uninterruptible Power Supply (UPS).

 

[throwaway4372]

chkdsk

Reference article for the chkdsk command, which checks the file system and file system metadata of a volume for logical and physical errors.

docs.microsoft.com

 

[champ222]

have you tried building the system outside of the case?

 

[welshrat]

I would be tempted to give the memory and the SOC a little more voltage. Are you running XMP, try 1.37 V for memory and 1.1v - 1.12v for SOC and see if that helps. I had trouble running 32gb 4 sticks of B-Die on my old 3700X but those voltage values totally solved it. I have since moved to a 5900X but kept the memory voltage at 1.37v despite it probably being stable now as I cannot be arsed with crashes.