Help, repeated freezing behaviour! Can I save this computer?

[guillow]

Hi all,

 

Hoping to get some ideas on if I can save this computer or not.

 

Made this computer following a deep learning guide 3 years ago and have had trouble with it ever since. The computer will sometimes freeze every few hours, to every few days. There seems to be no specific pattern to the freezing in terms of what can provoke it. I can be watching youtube, and it can freeze. I can be working in Excel, and it can freeze. I can be surfing, and it will freeze. When it does freeze, what happens is that any audio will go into a "brrrrrr" distorted recurrent sound (not sure how else to describe it). And I will have the screen freeze with no input from mouse or keyboard. The keyboard will not be able to change its caps lock function, so i'm pretty sure its really locked up. I have found, that sometimes, when Nvidia has new graphics drivers come out and I haven't gotten up to date. The freezes can be more frequent, and then after i update things are marginally better, but I'm not sure if that's the case...

 

What I have:

CPU: AMD Threadripper 1900X 3.8 GHz 8-Core Processor  ($99.99 @ Amazon) 
CPU Cooler: Fractal Design Celsius S36 87.6 CFM Liquid CPU Cooler  ($192.63 @ Amazon) 
Motherboard: Asus PRIME X399-A EATX sTR4 Motherboard 
Memory: Tforce Delta 16gx2
Storage: Samsung 970 Evo Plus 2 TB M.2-2280 PCIe 3.0 X4 NVME Solid State Drive  ($169.99 @ Adorama) 
Video Card: PNY Blower GeForce RTX 2060 SUPER 8 GB Video Card 
Case: Lian Li O11 Vision ATX Mid Tower Case  ($144.99 @ Adorama) 
Power Supply: EVGA SuperNOVA 1000 G5 1000 W 80+ Gold Certified Fully Modular ATX Power Supply  ($164.99 @ Amazon) 

 

 

What I have tried:

- i have run memtest overnight with dual check with no memory errors

- i have stress tested the GPU with furmark - no freezing during the runs

- i have stress tested the CPU with Prime95 x 1-2hours and no freezing

- brand new install of windows onto brand new m.2 drives

- new m.2 drives

- replaced a fractal design AIO with a Noctua 14U. This has helped in controlling CPU temps during prime95 runs, but the freezing continues

- i bought new memory that is validated as per the Asus memory validation list
Memory: G.Skill Trident Z RGB 16 GB (2 x 8 GB) DDR4-3200 CL16 Memory  ($51.99 @ Amazon) 
Memory: G.Skill Trident Z RGB 16 GB (2 x 8 GB) DDR4-3200 CL16 Memory  ($51.99 @ Amazon) 

- switch out the GPU with a 1050ti

- reset the bios with a pin reset

 

I feel like the only thing left is to try:

- new PSU

- new 1900x

- new motherboard

 

But at this point, it's getting to a point where, it wouldn't make sense to buy a new motherboard it feels like. And that it would be better to cut my losses and get a new computer altogether?

 

Any suggestions on how to troubleshoot? (i don't even get a BSOD...)

[jaslion]

Have you updated the bios? First gen threadripper is not a stable cpu or a good cpu. Weird guide to recommend it as it really is just not good at all. A at the time cheaper 3600x easily beat it.

 

If this is an older bios could be its simply being unstable as is common.

 

What speed are you running the ram at? Try 2133mhz is you havent already. Gen 1 threadripper did not have a good memory controller.

 

[guillow]

I think the intent of the guide was to provide something that had a ton of PCIe lanes to allow for multiple GPUs. Never managed to get there, and now as a daily driver, I think I'm suffering all the downsides without any of the upsides. 

 

After suffering through it for 1 year i caved and went on the hunt for QVL memory. I have it now in quad channel, but still having trouble. memory clock is set at 1067 which I think is 2133 after the doubling. 

 

I'm starting to think this is just related to the threadripper platform. But surprised that others would be able to tolerate this problem since it is fairly frequent. I was looking at getting a Gen2 or Gen3, but even now in 2024, they are still in the over $1000 range! crazy! Even the used ones on ebay are more than a new i7! 

 

I had updated the bios last year to latest, but just checked Asus' site and they just released a new version this month....crossing my fingers...

 

At my local computer shop, they suggested to check my nvme drive. I find that funny considering that I have 2 drives, one for OS one for files, and both are fairly new with the build, and this started with the build start as well... So will run a drive diagnostic tonight, while backing everything up with an attempt to change the NVME drive.

 

If that doesn't work. I bought a refurbished corsair Hx1500, to change out the PSU.

 

If it gets to the point, where I have to change out the MOBO or CPU, i will probably throw in the towel.

Any other ideas on how to troubleshooting random freezing? i wished there was a BSOD so I can at least figure out the problem!

[guillow]

updated to new bios. And got a nice freeze 15min in. sigh

will try to a nvme swap for os and see...

[guillow]

as I went to do a long diagnostic with the samsung magician tool, it turns out I never used a samsung NVME driver...

now that that's installed, will run long diagnostics, and see if that did the trick...

[Robchil]

2 minutes ago, guillow said:

as I went to do a long diagnostic with the samsung magician tool, it turns out I never used a samsung NVME driver...

now that that's installed, will run long diagnostics, and see if that did the trick...

do they have an nvme driver?.. usually it's the controller driver on the mainboard you have to install from AMD. 

 

[guillow]

i've definitely gone to Asus for the x399 board and done all kinds of driver updates. 

And windows and nvidia, and I think amd as well.

 

But yeah, samsung has it's own NVME driver...and installing it allows the samsung magician software to do a extended SMART test... Previously it was a generic controller driver name in device manager...

 

so far so good...

would be nice if the fix was this easy...missed driver...

[guillow]

full scan, os drive, all LBA sectors are good. 

will scan the other nvme, and play around... sometimes the freezes can take days to show up, which is annoying for troubleshooting. (although recently it has been at most 12hours...)

[guillow]

after a few hours of use. Freezing has returned. 

So it's not the NVME driver.

 

So far what I have tried:

 

Drivers:

- cleared CMOS by removing bios battery, shorting, and using default bios settings (plus TPM for windows 11 and full cpu fan speed)

- updated to latest bios

- updated NVME driver

- updated nvidia driver

 

Testing:

- prime95 x 1 hour

- furmark

- memtest

- samsung magician full scan of both nvme drives

 

Parts:

- replaced a fractal design AIO with a Noctua 14U. This has helped in controlling CPU temps during prime95 runs, but the freezing continues

- swapped out memory with CVL as per the Asus memory validation list
Memory: G.Skill Trident Z RGB 16 GB (2 x 8 GB) DDR4-3200 CL16 Memory  ($51.99 @ Amazon) 
Memory: G.Skill Trident Z RGB 16 GB (2 x 8 GB) DDR4-3200 CL16 Memory  ($51.99 @ Amazon) 

- switch out the GPU with a 1050ti

 

It feels like I've ruled out:

- GPU (gpu swap)

- Drives (full scan)

- CPU cooling (prime95)

- memory (memory swap)

- bios

 

So it feels like all I can do is:

- try another new install with all drivers uptodate (although this will be like the 3rd time over 2 years trying)

- try a different psu

 

Any other ideas? on how to troubleshoot this?

It feels like a driver issue with the older 1900x, but just don't know how to prove it, find it, or solve it...

 

[guillow]

did clean new install, deleted partitions, formatted, reinstalled

I'm not adding any drivers unless I get trouble.

So trying bone stock windows updated, and only critical apps...

 

Will see how this works out...

[guillow]

Update, been up running a few days with nothing at all...

I guess it was some kind of corrupted driver or app that was running.

 

Interestingly I have NOT updated drivers from:

- asus mobo

- nvidia

- samsung NVME

 

Really, all i did was use barebone windows install and updating. 

 

Have also avoided some older apps that I worried would cause issues:

acrobat 2017

synology drive

 

Have to add back in synology drive, hopefully that isn't the issue...