PC completely freezes under load

[IceChes]

Hello. I am coming here to report an extremely bizarre freezing issue that has been happening ever since I switched to an HRR display on my Arc B580. It is extremely similar to this post a few days ago. Actually, exactly the same symptoms. Nothing in event viewer or anything. The difference is that I am on the i5-14400F which is SUPPOSEDLY NOT AFFECTED by Intel's issues. Some have mentioned it could be power related, but I'm getting ahead of myself. Here are my system specs. MBO is motherboard by the way, just trying to keep the list readable.

CPU: Intel Core i5-14400F
GPU: Intel Arc B580 (ONIX AIB model)
RAM: 2x16 DDR5-5600 from Crucial
MBO: Asus Prime B760M-A-AX
SSD: WD Blue 256GB m.2
HDD: WD Blue 4TB
PSU: EVGA 450BR

Under load, primarily gaming load, the system will occasionally completely freeze. Both monitors freeze on the most recent frame, all USB devices will momentarily go dark and then turn back on a few seconds later, and occasionally there is a pop or crack from the speakers. Because this started after I got a high refresh rate monitor (180hz), I initially assumed it was the Intel graphics driver dying, but event viewer had no errors. Several people have mentioned it could be power, but I am running furmark and prime95 at the same time while typing this and the system is still stable. I plan to leave those programs running for a while throughout the evening (with the CPU locked to PL2) just to be safe, but I don't think it's power. Some people have mentioned memory, but I think memory issues usually cause a BSOD? I thought it may be oxidation, but the 14400F isn't officially affected. However I did buy this one used (dumb, yeah), so I'm not sure what kind of life it's lived. The motherboard BIOS is updated to the latest microcode. Someone else mentioned ASPM, which I have since disabled.

 

The only weird thing I have noticed so far is weirdly high virtual memory load that happens occasionally. When I was playing Final Fantasy XV yesterday it was at 97%. 

 

Games I've seen this in:

• Final Fantasy XV
• Splitgate 2

There well may be more, but I don't play many games very often. All games launched through Steam.

 

I have HWinfo64, prime95, cinebench, furmark, throttlestop, all the fancy stuff, so any information that could possibly be needed there's a good chance I have.

I will update this thread with new clues as they come up.

Updates List:

• Someone mentioned I should run a memory tester so the computer is now running memtest, furmark, and prime95. Will update with results.
• Nothing turned up, but I didn't run it for very long.
• Intel Processor Diagnostic Tool turned up nothing.
• Furmark and Memtest64 ran at the same time for 9 hours was stable.
• Prime95 in mixed mode ran for 9 hours was stable.
• Final Fantasy XV left to idle for 2 hours was stable.

[CatTNT]

the way to rule out Intel CPU slowly frying itself is to underclock the P cores by 300-500MHz from stock. My 14900K was fixed for a couple months by lowering the 6GHz cores down to 5.7, but recently i started to have issues again and have dropped all the P cores down to 5.6 GHz, and that fixed it. These CPUs are all going to die and fry themselves, in my experience with a 100% certainty. 

[StDragon]

23 minutes ago, CatTNT said:

These CPUs are all going to die and fry themselves, in my experience with a 100% certainty. 

If it's a new CPU that's installed with a BIOS updated within days or weeks, it's probably fine for the life of the PC. But if you've been running that CPU for months before updating the BIOS, chances are damage to the CPU die has already occured, sadly.

[CatTNT]

This is my 2nd 14900K i got, its only ever ran the "patched" microcode and still ended up killing itself. At this point its either Intel's fault or the motherboard. The CPU tried to go at 330W+ despite the fact it should be using the Intel Default 253W power plan. It just ignores whatever power limits i put on it, even when i manually put a 253W cap, it still went to 330-350W

6 minutes ago, StDragon said:

If it's a new CPU that's installed with a BIOS updated within days or weeks, it's probably fine for the life of the PC. But if you've been running that CPU for months before updating the BIOS, chances are damage to the CPU die has already occured, sadly.

 

[IceChes]

1 hour ago, CatTNT said:

the way to rule out Intel CPU slowly frying itself is to underclock the P cores by 300-500MHz from stock. My 14900K was fixed for a couple months by lowering the 6GHz cores down to 5.7, but recently i started to have issues again and have dropped all the P cores down to 5.6 GHz, and that fixed it. These CPUs are all going to die and fry themselves, in my experience with a 100% certainty. 

Well it's a locked chip, which I don't believe I can change the clock on. It runs at around 125 watts of the max 148 (Intel's TDP) if I let it burn at PL2 indefinitely. Is the system just randomly snapping out instantaneously a symptom of oxidation? I know it has suffered some sort of issue because I had a BSOD a few weeks ago after the CPU literally just stopped responding, but I was under the impression that the system didn't usually just instantly die like that and did some kind of BSOD first.

EDIT: How would I test to see if it was a CPU problem? Just run prime95 all night?

[CatTNT]

18 hours ago, IceChes said:

Well it's a locked chip, which I don't believe I can change the clock on. It runs at around 125 watts of the max 148 (Intel's TDP) if I let it burn at PL2 indefinitely. Is the system just randomly snapping out instantaneously a symptom of oxidation? I know it has suffered some sort of issue because I had a BSOD a few weeks ago after the CPU literally just stopped responding, but I was under the impression that the system didn't usually just instantly die like that and did some kind of BSOD first.

EDIT: How would I test to see if it was a CPU problem? Just run prime95 all night?

If you can't manually change the clock multiplier per core, then you can try changing the BCLK (base clock) down from 100Mhz to 96-95MHz. It just takes the BCLK x core multiplier = core frequency. So a default 14900K its 100MHz * 60 = 6Ghz core clock. For 14400F, if you can't change the clock frequency down from 47, you can try a BCLK of 95, so it'd be 95 * 47 = 4.465Ghz. Maybe even 100-200MHz lower just to be sure, but that should be enough. 

Or you can try to use Intel Extreme Tuning Utility, but if it's locked then I don't know how to get around that. Maybe try disabling hyperthreading and/or E cores and see what happens. 

[IceChes]

1 hour ago, CatTNT said:

If you can't manually change the clock multiplier per core, then you can try changing the BCLK (base clock) down from 100Mhz to 96-95MHz. It just takes the BCLK x core multiplier = core frequency. So a default 14900K its 100MHz * 60 = 6Ghz core clock. For 14400F, if you can't change the clock frequency down from 47, you can try a BCLK of 95, so it'd be 95 * 47 = 4.465Ghz. Maybe even 100-200MHz lower just to be sure, but that should be enough. 

Or you can try to use Intel Extreme Tuning Utility, but if it's locked then I don't know how to get around that. Maybe try disabling hyperthreading and/or E cores and see what happens. 

I mean I can try it. I just ran prime95 for about 9 hours and it went ok fwiw.

EDIT: Am I going to have to test this by just sitting AFK or self-botting in games?

EDIT: I tried changing the clock multiplier to 92% or something in ThrottleStop and my performance died. Went from 120 FPS to 30. Maybe I'm doing it wrong.

[CatTNT]

1 hour ago, IceChes said:

I mean I can try it. I just ran prime95 for about 9 hours and it went ok fwiw.

EDIT: Am I going to have to test this by just sitting AFK or self-botting in games?

EDIT: I tried changing the clock multiplier to 92% or something in ThrottleStop and my performance died. Went from 120 FPS to 30. Maybe I'm doing it wrong.

When my 14900K was dying it started with random reboots overnight every couple of days, and eventually BSODs within 30 minutes of booting in windows. 

I just reread your original post, and you said something about virtual memory. I just remembered I had a similar issues, same symptoms at least, because I had a game server automatically back up every 20-30 minutes of gameplay, and my computer would freeze and becomes totally unusable for about 30-90 seconds. I found out it was because whatever the backup was doing completely filled up my main drive's I/O bandwidth (Which shouldn't happen, because it was only 10-15GB and I have a 4TB NVMe drive with 4-5GB/s read/write). Is it possible that you have some sort of OneDrive, Google Drive, or any other software that is backing up, reading, or writing a lot of data? I noticed smaller, but similar symptoms when setting my Nvidia app clipping/recording to use really high quality settings. Maybe some Intel app/driver is automatically clipping recordings? 

[IceChes]

46 minutes ago, CatTNT said:

When my 14900K was dying it started with random reboots overnight every couple of days, and eventually BSODs within 30 minutes of booting in windows. 

I just reread your original post, and you said something about virtual memory. I just remembered I had a similar issues, same symptoms at least, because I had a game server automatically back up every 20-30 minutes of gameplay, and my computer would freeze and becomes totally unusable for about 30-90 seconds. I found out it was because whatever the backup was doing completely filled up my main drive's I/O bandwidth (Which shouldn't happen, because it was only 10-15GB and I have a 4TB NVMe drive with 4-5GB/s read/write). Is it possible that you have some sort of OneDrive, Google Drive, or any other software that is backing up, reading, or writing a lot of data? I noticed smaller, but similar symptoms when setting my Nvidia app clipping/recording to use really high quality settings. Maybe some Intel app/driver is automatically clipping recordings? 

I have Steam recording I guess. But that's not that much data. I also have OneDrive but I use it approximately never so it shouldn't be doing weird stuff.

 

My computer hasn't rebooted ever without me making it. The only "this is definitely the CPU" issue I've had was a single BSOD a few minutes after loading windows with a CPU watchdog error. I guess the CPU just gave up that time. Haven't seen it since. 

I also just left Final Fantasy XV idling and nothing happened for a few hours (and I haven't had issues idling in other games), so I'm thinking this is an issue with the PC needing to do something and then not being able to and something then panics and kills windows, whether that be the memory not being able to do something, or the CPU freaks out, or the graphics driver crashes and takes windows down, or something. Could also explain why benchmarks, stress tests, and performance tests aren't triggering it. They just aren't doing the right (wrong) thing.

So far I've only noticed this sort of snap-crash thing on Final Fantasy XV (both using DXVK for Intel Arc performance and native DX11) and Splitgate 2. That said, I haven't played other games extensively yet on this system other than Minecraft (which isn't very demanding. it's also not a Steam app so it doesn't have the clipping enabled, dunno if that's relevant). I'm going to add a list of games and programs the PC has crashed while running to the initial post.

[IceChes]

Just happened again, twice in a row, less than an hour between playing FFXV again. Was stable for several days before until now.

[IceChes]

I've managed to create some kind of problem running Ycruncher. Last night I left it running and came back to my PC on the sign in screen. At school I left it running and came back to a checksum error on core 10.

[IceChes]

Got another error in Ycruncher, some kind of logical error on core 10. If anyone knows more about Ycruncher please reply, this is the only thing I've tried so far that gets my PC to freak out.