Weird temperature issues with cpu (water cooled)

[barnack]

Greetings,
my computer is 3 years old right now, and it was a really good one by 3 years ago standards.
Since the tower case is let's say "homemade" to not be too precise, the airflow kinda sucks; i had a lot of temperature issues with cpu and motherboard; then i got water cooling for my cpu with the radiator placed outside so it wouldn't depend on the internal case temperature; and thanks for the space saved compared to an internal huge air cooler i could add a fan that points directly to my motherboard heat sink.
Since i did that my computer works flawlessly with intense and demanding games *during winter*… and only during winter with open room windows.

During summer, it's still a miracle if i can run a game with minimum settings for 1 hour, and that's AFTER i added the water cooling.

The weird thing is, i'd expect my cpu to rise in temperature gradually, hit the 75° limit i've set and then shut down my computer as planned.

However, i tried playing with Open Hardware Monitor (from now OHM) on the side, checking constantly motherboard, cpu and gpu temperatures. Since i added that fan in front of motherboard's heat sink it stopped being a problem (before it was the mb that reached critical temperatures first), now it doesnt get over 60°. The gpu has 2 fans of which only one works automatically, and the other one is supposed to self-start when gpu gets too hot, but apparently it self starts too late, because many times i've seen gpu temp go up, second fan starting right when pc suddenly shut down. But that's easily solved manually setting gpu fans to 100%; doing that gpu doesn't get past 65-70°.

Now to the problem: the cpu. It seems to act weirdly; while mb and gpu gradually increase their temps and then stabilize, cpu suddenly jumps from 22-30° idle (even during summer which i think is awesome temperature, considered it's practically air temperature here) to 60+. Only a couple times i could see the reading in OHM get to 80+ one instant before the pc shut down. (OHM updates readings every second). And it jumped to 80+ totally suddenly from around 60°; which made me suppose all other times my pc shut down with no apparent over 75° reading it was because the cpu suddenly jumped to 80+ and the pc shut down before i could manage to see the reading.

But if i see the temperature reaching 73-74 and i instantly close the game/program which is draining resources in literally 1 or 2 seconds, no more, it's back again to 22-30°.

So it cools incredibly fast to incredibly low temps, but it utterly fails at keeping stable temperature on stress.

Running any stress tool my computer shuts down, no matter what. Even running memtest it shuts down, and it's not because of the memory blocks (i've checked them individually in someone else's computer).

Could be the psu? I highly doubt it, since it wouldn't explain why it doesn't have problem in winter at opened windows.

What can be the issue?

________________________________________
Motherboard: CROSSHAIR V FORMULA-Z
CPU: AMD FX(tm)-9590 Eight-Core Processor (8CPUs), ~4.7GHz
Memory: 4x 4GB Trident-X
GPU: AMD Radeon 7800 Series (too lazy to check the exact version, shouldnt matter though xD)
Water cooling:
CPU only,
Pump: Aqua Stream XT
Radiator: EK Water Blocks EK-CoolStream PE 360

 

[1LiquidPC]

What PSU are you using? Could be starting to fail. The 9590 is quite a power hog as well. Some folks had stability issues with those even running stock speeds as AMD really binned them at the limit. Regarding cooling, what has been done regarding maintenance with your watercooling setup? Fans on case and GPU cleaned as well?

 

[barnack]

I clean dust regularly once per month, two months at worst; change filters once per year (during summer because of the mentioned bad internal airflow i remove the side panel, making filters useless anyway; i just clear dust more frequently).
The gpu i'm pretty sure doesn't have tempreature problems, everytime i force both its fans to run it gets to stable temperatures and stop there.

What do you mean by "running stock speeds"? If you're supposing i'm overclocking, sorry to tell you i'm not. I never even remotely thought about overclocking considered i have all these problems without it.

I already swapped psu with another model after some suggestions from bleeping computer's forum last year.
Now i'm running a "power lc66006p2 v2.3"

 

[1LiquidPC]

The issue occurs with 2 different PSUs? Some quick searches generally put Power-LC at lower to mid-range quality depending on model. Also, what watercooling maintenance have you been able to do? Fresh coolant, block cleaning, ect?

 

[barnack]

Yeah, my previous psu was even lower range quality but to get my hardware i had to save money somewhere; still i noticed no difference in the frequency the shut downs occured when switching to this psu.
I've literally reset my water loop months ago. Dried the entire loop, made water flow with decent and constant pressure on both directions in the radiator and cpu block, dried again these parts, refilled with new liquid.
Then went winter, so i dont really know if its actually winter cold that makes all better, or summer is worse just because more time passed since last liquid-lopp clean up...

BUT, since luck hates me and unluck loves me apparently my pump died 1 hour ago. I got it out, opened it (thought there could be some dirty something around in the loop?) but it was surprisingly clean. For the pump i opened a topic in the device's company forum, dont want to switch focus here.
___________________________________
Rarely the shut downs occur during start up, before windows (when you can press a button to enter bios), if i try to restart my computer right after it shut down.

If i run any cpu intensive task, like editing and processing videos, i have to run BES Cpu Limiter and limit the cpu used by that software to 20%, else cpu will get hell hot hell quick. Or can it be not enough powerful PSU as you're suggesting? Can limiting a software's CPU usage reduce the overall power consumption?

 

[nerro120]

The 9590 was never a good choice, as mentioned before many people had issues running at stock settings, it is only a 8350 overclocked to the limit. Most people had to under lock it to keep it stable. It's a 220w CPU and alot of the am3 motherboards can not handle that. It could be the cpu overheating, could be the vrm overheating. Bad airflow is definitely not helping. It would make even more sense that it works better in the winter than in the summer for temperatures being the issue instead of the power supply

 

[1LiquidPC]

Possible the failing pump was causing occasional startup shutdowns. Yes, lowering max CPU speed will lower temps and power consumption. Spikes in temps are still normal with custom watercooling. Still, should be nowhere near shutoff at stock.

Have you tried a remount of CPU block. Thermal compound dries over time, and this C Pool U could do that rather quickly. Also, VRM temps could be on high side during summer from awful airflow. Any way to upload pics?

 

[barnack]

Changed Thermal paste 1 month ago. My spikes are not from time to time; shouldn't the cooler prevent the spikes from getting to 80+ from 30 in an instant? Because that's what makes it shutdown, reaching temps higher than the set limit.

As i said, the radiator is outside of the case, so it's totally independant from internal air tempratures. Room temperature air flows in it.

Possible the failing pump was causing occasional startup shutdowns.

Ibut doesnt explain why my problems have been continuous since the pc was new to today... well Yesterday until pump failed xD