Freeze only when idle

The_Bullet

The_Bullet

The_Bullet

Associate
Joined
7 Mar 2003
Posts
193
Location
Close to the Edge
Hi all. Apologies for the long post, I’m trying to provide as much info as possible. I’m after some troubleshooting advice, or maybe some experiential knowledge. In late October I bought an Asus TUF RTX 3080, and for good measure a Corsair HX1000 PSU. I fitted these into my existing system (replacing a RTX 3060). This system prior to the upgrade was solid as a rock for 2 years. Things appeared fine. Benchmarks were as expected, and it ran games well. On around 13th November I got a deadlock freeze in windows. This has only happened when idle. No black screen or bsod. Only thing I could do was a hard reset. There were no error logs anywhere. The only logs for the time was relating to my hard reset. There are no whea logger events or the like. This happened again 9 days later (around 60 hours plus of run time). I took the decision to return the PSU and replaced it with a 1000w Asus Rog thor psu. I also disconnected a mechanical hard drive. Things looked good until yesterday when it happened again twice within a few hours of each other. This is about 180 hours of problem free run time since the last freeze.

I ran memtest overnight error free. OCCT tests run error free. All temps reported as fine. Fans do not ramp up when the freeze occurs. I’m struggling to see how I can troubleshoot this, with errors taking anywhere from a couple of hours to almost 2 months to appear. I have played games for hours on end with no issue. Anyone seen the like of this, or could point me in the right direction?. Can a fault with a graphics card manifest like this?. Many thanks.

System - Windows 10 pro. Asus x-570f gaming motherboard. Ryzen 3900x cpu. Crucial ballistix 32 gb ram at 2666mhz.Crucial mx500 1TB system ssd plus Crucial 2TB ssd. Asus tuf 3080 GC and Asus Rog Thor 1000w psu.No overclocking.
 
I'd run some scans on your SSDs, maybe replace the cables as a precaution and you could try bumping up some of your voltages, though 2666 is probably run at stock anyway? What's the power like in your house?
 
Last edited:
CrystalDisk report ssd's as good. I also used Seatools extended tests, all ok. Ram is stock, and Power to house is stable. No outages, or flickering lights etc. Nothing changed, other than the aforementioned GC and PSU swap.
eta .... I replaced sata cables after first instance in november. I've tried researching the windows 10 updates installed around that time, and I can find no mention by others of this type of issue.
 
Last edited:
The_Bullet said:
CrystalDisk report ssd's as good. I also used Seatools extended tests, all ok. Ram is stock, and Power to house is stable. No outages, or flickering lights etc. Nothing changed, other than the aforementioned GC and PSU swap.
eta .... I replaced sata cables after first instance in november. I've tried researching the windows 10 updates installed around that time, and I can find no mention by others of this type of issue.
Click to expand...

There have been reports of Ryzen CPUs crashing/freezing only at idle, but I think that's usually out of the box (not after 2 years, heh) and resolved with either bumping voltages, or an RMA.

Is there likely to be any link to sleep mode? Sleep is notoriously unreliable and drivers or devices being incompatible with sleep is quite common, which are sometimes messed with through automatic updates. Late HDDs and early SSDs could also have super aggressive power management features.

One way to test the above would be disable all sleep (mainly C) states in the BIOS and sleep & power management features in the OS.

I assume you pulled (did not re-use) all the modular cables when you replaced PSUs?

Can you share CrystalDiskInfo screenshots, with (if the values aren't readable) the raw data format changed?
 
Last edited:
Hmm, tough one there but I would focus your efforts on the other changed variable, the GPU.

Are you using different rails to power the GPU? Even though it's the same family of GPU I would clean out the GPU drivers, check for updated motherboard bios and AMD chipset drivers, update/install those and then do a fresh, clean install of the latest Nvidia drivers.

Failing that then it's the potentially arduous task of testing/using a clean new OS install which should ultimately show whether it's hardware (GPU) fault or something in the software.
 
Thanks for the replies. Similar to all but ruling out the psu, I have all but ruled out the gpu, as I have temporarily got hold of a 4070ti. I put that in earlier today, and around 6 hours in was another idle freeze. Looking at the event viewer again, there are still no error logs to help. However, something that I missed from before - even though the critical event kernel power 41 marks the time of the system reset, in the "error" dropdown list it has "event log error 6008 the previous shutdown at .....was unexpected", which has the time of the system freeze, not the time when I noticed and pressed reset. Maybe this is an artefact of the way that particular event being recorded, I don't know. So I am stumped. I could try the c-states setting, but why was the MB cpu and memory rock solid for 2 years with c-states at default? btw no over/underclocking volting on gpu or other components.
 
Well after all but ruling out the GPU you might want to try a clean fresh OS, maybe even Win 11? ;) . Do you have a spare SSD etc you could run a test OS on leaving your original one in tact?
 
It sounds similar to what I was having my 3080. Odd behaviour on the desktop but consistent load was generally fine. I solved it with a replacement psu, but I think it was more about how it handled transient spikes than actual wattage.
 
MartinPrince said:
Well after all but ruling out the GPU you might want to try a clean fresh OS, maybe even Win 11? ;) . Do you have a spare SSD etc you could run a test OS on leaving your original one in tact?
Click to expand...
That is what I am going to do, minus the Win 11 part. Also disconected a usb hub and trying a different mouse and KB. After that it will start to get expensive :eek:
 
You must log in or register to reply here.
Back
Top Bottom