Sudden brief Temperature Spikes - I7-8700k: Old Paste, Pump, or Something else

Aireoth

Supreme [H]ardness
Joined
Oct 12, 2005
Messages
6,043
I ran this processor overclocked to 5GHz for years, then suddenly about 3 months ago I started crashing due to overheat. I immediately rolled back the OC to stock and reapplied thermal paste (mine is AS-5 and old about 7 years so). Due to work volume I didn't have time to really fool with my settings until yesterday, and I noticed that even under stock settings it will spike to 100c with the system idling, no abnormal CPU usage reported by task manager.

I use a H150i Pro AIO water cooler that is 4 years old and manually set the pump to maximum which seems to have helped a bit, but cannot overclock at all due to the spikes without pushing 100c under load. Planning to pick up new paste, but what are the chances its a failing pump (corsairs program doesn't show any issue with pump rpm) or aging CPU/MB combo being to blame?

Really if it turns out to be more than paste it will hasten my replacement schedule as I want to switch to a new Ryzen system and would replace the cooler at that time as well.
 
I ran this processor overclocked to 5GHz for years, then suddenly about 3 months ago I started crashing due to overheat. I immediately rolled back the OC to stock and reapplied thermal paste (mine is AS-5 and old about 7 years so). Due to work volume I didn't have time to really fool with my settings until yesterday, and I noticed that even under stock settings it will spike to 100c with the system idling, no abnormal CPU usage reported by task manager.

I use a H150i Pro AIO water cooler that is 4 years old and manually set the pump to maximum which seems to have helped a bit, but cannot overclock at all due to the spikes without pushing 100c under load. Planning to pick up new paste, but what are the chances its a failing pump (corsairs program doesn't show any issue with pump rpm) or aging CPU/MB combo being to blame?

Really if it turns out to be more than paste it will hasten my replacement schedule as I want to switch to a new Ryzen system and would replace the cooler at that time as well.
Much more likely that the mounting pressure is off or the pump is failing than the paste going bad. I've had 10+ yo boards that have compound for the NB and/or SB that is dry as a bone but still did a decent job transferring heat to the heatsink.
 
I ran this processor overclocked to 5GHz for years, then suddenly about 3 months ago I started crashing due to overheat. I immediately rolled back the OC to stock and reapplied thermal paste (mine is AS-5 and old about 7 years so). Due to work volume I didn't have time to really fool with my settings until yesterday, and I noticed that even under stock settings it will spike to 100c with the system idling, no abnormal CPU usage reported by task manager.

I use a H150i Pro AIO water cooler that is 4 years old and manually set the pump to maximum which seems to have helped a bit, but cannot overclock at all due to the spikes without pushing 100c under load. Planning to pick up new paste, but what are the chances its a failing pump (corsairs program doesn't show any issue with pump rpm) or aging CPU/MB combo being to blame?

Really if it turns out to be more than paste it will hasten my replacement schedule as I want to switch to a new Ryzen system and would replace the cooler at that time as well.
yeah time for new paste. you should be able to feel/hear if the pump is running. if it is, feel the tubes, if one is really hot you might have a clogged barb in the rad. ive had one corsair do that to me after ~5 years on an oc'd fx system.
 
Much more likely that the mounting pressure is off or the pump is failing than the paste going bad. I've had 10+ yo boards that have compound for the NB and/or SB that is dry as a bone but still did a decent job transferring heat to the heatsink.
Its torqued down rather nicely and seems fairly level to me, how would I check mounting pressure?

yeah time for new paste. you should be able to feel/hear if the pump is running. if it is, feel the tubes, if one is really hot you might have a clogged barb in the rad. ive had one corsair do that to me after ~5 years on an oc'd fx system.

Pump does vibrate, corsairs link program reports an RPM of 2790-2820, line temp of 31.1 to 35 degrees. That is if the program is reliable.

tubes do not feel noticeably hot, maybe a little warm to the touch.

* I have some AS-MX5 on order, but I'm tempted to just head down to Memory Express and grab some MX-4 to try it.
 
Its torqued down rather nicely and seems fairly level to me, how would I check mounting pressure?



Pump does vibrate, corsairs link program reports an RPM of 2790-2820, line temp of 31.1 to 35 degrees. That is if the program is reliable.

tubes do not feel noticeably hot, maybe a little warm to the touch.

* I have some AS-MX5 on order, but I'm tempted to just head down to Memory Express and grab some MX-4 to try it.
Easiest way to check pressure is a small dot of compound on center of cpu and then torque it down, unscrew and see the spread pattern.
 
Repasting with MX5 helped, water temperature went up by 7 degrees and cpu cooled more successfully, still a initial spike in temp on a prime95 stress test hitting 94 degrees, but smoothed out to 80 afterwards and stayed there +/- 5 degrees for 15 minutes.

Didn't know my local computer store had MX-5 in stock as the website wasn't even showing MX-4 yesterday, but ever since they expanded to Ottawa their quality control has taken a hit (frankly it did with their initial out of province expansion as well). Sadly I don't think they'll retain the 'best store ever' ranking in my heart for much longer, but likely will be the best place around here for a long time.

Time to see if the OC will hold too!
 
Using my old profile its throttling at 4.7Ghz with core 3 hitting high 90s, more testing required and maybe a delidding. Also noticing a slight slowdown in desktop operations (right-click and close mainly), maybe the system is just getting to old.
 
Yeah, its pretty clean, little dusty but I dust my systems at least once a year, usually two or three times.
 
Please explain what "load" is causing 100c.
literally nothing running, pc booted for about 10 minutes, start hwinfo64 and start tracking, walk away, within a few minutes at least one core would hit 100c. Cpu load averaging 5-10% due to various background work programs (onedrive, teams, logmein) and Eset (idle) running.

*With new thermal paste its runs about 32 degrees and hits 59 on occasion instead of 100, pretty much same circumstances.

**log files don't show much for any sudden cpu usage spike, just fluxuating between 2 and 10% that roughly corresponds with Onedrive looking for file changes.
 
Last edited:
Back
Top