So when CPUs overheat and shut down...

[tRidiot]

This degrades the silicon inside, and can lead to inability to OC anymore? Or significantly degrade it?

Seems like my 3950x has gone downhill quite a bit due to my recent mistake in applying thermal paste. It runs fine at stock, but I have little to no overclocking ability anymore.


When I got it, I could do 4.4, sometimes 4.5. Then I removed and reapplied my cooler, it was getting too hot (like 94 on P95), reapplied again with Kryonaut this time and it seems I might have spread it unevenly, it was shutting down instantly when trying to run stress tests. I pulled it off and saw this:






I wouldn't have thought that would have significantly degraded my chip. In fact, I didn't know why I was shutting down immediately except I would reboot into a "CPU overheat" error that required me to go into the BIOS. Like, it was happening so quick I never saw the temps rise in HWMonitor. I would click to start P95 or C20 and within 1-2 seconds would be crashed. Eventually I did see a it once turn red in HWMonitor and saw a 115 temp (yikes!).

Now, after reapplying again, in a much more even layer, my temps are 60s to low 70s at load at stock, but my overclockability has tanked.

I am running 3.9 right now, using the multiplier option in Asus AI Suite 3, but if I go into the BIOS and change my multiplier to 38, it crashes immediately like before when trying to run a stress test.


Although, now that I think about it, my RAM is new, too... but I thought it should be good. It's Samsung B-die 3600 14-15-15-35... should I downclock my RAM to try to see if my processor will clock higher again? This just occurred to me, so I'm kinda thinking it out as I type. Maybe I'll do that, go set my RAM to 3200 and see if I can get my OC back up. If it turns out it's the RAM, I'm sending this stuff back. I ain't paying that kind of money if it won't run at advertised speed and timings. Hmmm...

 

[RJARRRPCGP]

A shutdown even after redoing the TIM?! I would also start to put the motherboard on the hook, for a possible MOSFET failure or of course, a rare PSU issue.

 

[tRidiot]

Hmmm... I purposely chose the Crosshair VIII Hero for its supposedly heavily overbuilt VRMs, etc., and wouldn't think my 1000W PSU should give me trouble on only CPU testing - with heavy GPU testing I was drawing over 500 watts on my 3090.

I didn't consider those things. I tried downclocking my memory to 3200 and was still shutting down with P95 on 39.00x and lasted a few seconds longer on 38.50 but still shut down.

Wonder if I should put in my other Micron-based sticks and recheck.


So you don't think quick temp-based shutdowns would have 'degraded' my processor?

 

[RJARRRPCGP]

So you don't think quick temp-based shutdowns would have 'degraded' my processor?

Your symptoms are more serious, as it doesn't appear to be a crash and reboot. Instead, it appears that you lose power! It's possible that your 3rd gen Ryzen degraded, but I don't expect that unless you were doing a manual OC.

 

[tRidiot]

Your symptoms are more serious, as it doesn't appear to be a crash and reboot. Instead, it appears that you lose power! It's possible that your 3rd gen Ryzen degraded, but I don't expect that unless you were doing a manual OC.

You mean manual in the sense of changing the multiplier? I've done that in the BIOS as well as with Ryzen Master and Asus AI Suite 3.

Currently running AIDA64 stress test....

All cores have clocked to 4.1 from 3.5, loads at 100%, memory being stressed at 3600MHz also... CPU package temp at 77C with a low fan speed profile.


So it is running higher clock speeds on all cores clocking itself up than if I set the multiplier myself at 38 or 39. at which point it crashes on stress. I don't get it.

 

[RJARRRPCGP]

You mean manual in the sense of changing the multiplier?

Yes, especially with Vcore at more than around 1.35V.

 

[tRidiot]

I've tried manually in the BIOS and running my offset down a bit to 0.1. Didn't seem to help.

 

[RJARRRPCGP]

While over here, I'm focusing on good low-thread-count performance, because of Halo, mostly.

 

[Johan45]

You mean manual in the sense of changing the multiplier? I've done that in the BIOS as well as with Ryzen Master and Asus AI Suite 3.

Currently running AIDA64 stress test....

All cores have clocked to 4.1 from 3.5, loads at 100%, memory being stressed at 3600MHz also... CPU package temp at 77C with a low fan speed profile.


So it is running higher clock speeds on all cores clocking itself up than if I set the multiplier myself at 38 or 39. at which point it crashes on stress. I don't get it.

Reflash your BIOS but first, remove the AISuite and Ryzen Master ASUS ROG forum has a removal tool for AISuite that cleans things up. Both those programs make low-level settings and after crashing like you have can cause a whole world of hurt. Reflasing the BIOS will release any hold either of those have and starts you out fresh. I've been there with Ryzen Master while doing reviews, any change would crash like that. Reflashing fixed it for me.

 

[EarthDog]

Reflash your BIOS but first, remove the AISuite and Ryzen Master ASUS ROG forum has a removal tool for AISuite that cleans things up. Both those programs make low-level settings and after crashing like you have can cause a whole world of hurt. Reflasing the BIOS will release any hold either of those have and starts you out fresh. I've been there with Ryzen Master while doing reviews, any change would crash like that. Reflashing fixed it for me.

This.

The shutdown shouldn't have caused a problem....thats why it's there is to prevent damage.

 

[tRidiot]

Ok, good tips.

I am stuck at home iced in and had to close my office for a second straight day, so I can mess with this a little bit.

 

[tRidiot]

So I went ahead and flashed my BIOS and went with a complete format and Windows reinstall.

I got some software installed (benchmark and monitoring stuff, nothing from ASUS) and set multiplier to 40.00. Figured this was a nice basic easy OC - started P95 and immediately shut down, power button wouldn't shut it off even with long hold. Had to disengage the power cable.

Rebooted into BIOS, reloaded all optimized settings, rebooted. Loaded back into BIOS, set RAM at 3600 FCLK at 1800 with stock DOCP settings of 14-15-15-25, saved and rebooted to Windows. Everything was fine, ran P95 plus AIDA64 with stock CPU. Got this.



All cores at 100%, running 3.9/4.0 or so on all. Temps in the mid-70s. I figured, "I'm good."

So I went back into the BIOS, set multiplier to 38.00, back into Windows, started P95 and immediately crashed to black screen again. Again had to pull power to reboot into BIOS and go through the process again.


I don't understand... am I doing something wrong where I can't even OC to 3.8? I mean this CPU will boost cores to... 4.7?

- - - Auto-Merged Double Post - - -

And even BETTER news is... I got selected for a 5950x in the Newegg Shuffle. LAST NIGHT.

Apparently I didn't get the notification through the app like I thought it was supposed to do, and I didn't get push notifications of the emails like I thought I would... so I missed it.

 

[tRidiot]

I ran BoostTester and this is what I got... not sure if this tells me anything except that CCD1 is quite a bit behind CCD0.

 

[tRidiot]

Ok, so now I'm at stock everything and getting random reboots.

Twice in the last 10 minutes, just surfing the net, not more than 3 tabs open.

How do I narrow down what may be causing this? Bad MB? That could explain my problems with the inability to OC and now having random reboots, that would be my thought, but... I dunno. I'm open to ideas... I guess PSU could be a possibility, as mentioned above. I could pull an older Enermax 850 I have and use it to test. This Silverstone is very old, just never been used. Sat in a box since being tested by JohnnyGuru in 2012 I think. I guess being as old as it is, maybe that would be my most likely culprit.

 

[EarthDog]

You're at the point of testing individual components out, yeah. If you have a known good PSU, I'd try it first, but seeing as how you're having problems at idle (surfing the net) I'd lean more towards a bad mobo. Sucks.

 

[tRidiot]

You're at the point of testing individual components out, yeah. If you have a known good PSU, I'd try it first, but seeing as how you're having problems at idle (surfing the net) I'd lean more towards a bad mobo. Sucks.

Yeah I think I will try that, if I can find enough PCI-E connectors for it. If you have 3 connectors, will it run off two if you're not pulling alot of power? I don't know if there's a sensor in the socket to tell it if there is a plug in there or not. I could always pull my 3090 and run my GTX580 Matrix.

Man, I'm hoping it's just the PSU, that would make me feel a lot better that my CPU isn't the culprit. Of course, who knows, that could still be a problem, too, but all things considered, the simplest solution usually is the correct one. Having both the CPU AND the PSU or MB go bad would be a real bad lotto draw...

 

[Zerileous]

Maybe I'm missing something here, but the irregularity of it all, I would test out memory first. Could be something simple where XMP doesn't like your system. 3600C14 is not always easy to run. Since you have AIDA64 why don't you set to bios defaults (leave RAM at slow / stock speed) and run the memory/cache test for a few hours. You could even test drive it for a few days (the loss of performance won't kill you for a day or two). If all good, then go ahead and enable XMP and do the same.

 

[tRidiot]

Ok, that sounds like an easy plan. I can set that to run while I'm at work.

Set everything to auto in the BIOS and RAM is now at 2133 15-15-15-36.

I'll leave it like this for the day and see what it looks like when I get home. Obviously if it shuts down/reboots I would think it's not the RAM, as cutting the speeds didn't fix it. But if the system stays up and AIDA crashes, could be RAM. If everything stays up I would highly suspect the RAM... as cutting all the speeds seemed to solve the stability issue.

 

[freeagent]

My system just shuts off like that too if I have a bad mem OC. Mainly if things are just too tight.. or too fast I suppose.. I have to remind myself occasionally that this is not an Intel and it does not behave like one.. Is your ram on your QVL? I would assume it is being G.Skill.. for both my pairs of Tridents I checked the QVL on both sites and it was a go, and I haven't had any problems really other than what I have caused myself. But at first I thought maybe you cooked a CCD or X or whatever by not having TIM and 1.35v..

 

[RJARRRPCGP]

This Silverstone is very old, just never been used. Sat in a box since being tested by JohnnyGuru in 2012 I think. I guess being as old as it is, maybe that would be my most likely culprit.

Shutdown or crash? More likely invisibly failed caps, (without the tops becoming convex) which are reportedly common, when they are sitting! (not powered up)