4790k temperature problems after replacement cooler

Lyam_

Reputable
Nov 16, 2014
15
0
4,510
Hi all, first post on Tom's Hardware

Long story short, have a 4790k cooled by a h80i liquid cooler. After my PSU failed, it fried the cooler, so I got a replacement one and this time, got it fitted by a professional (first time was a DIY)

Only change is he made the fan intake from outside of the case, and exhaust inside (it's the only fan functioning in the case, other than the GPUs stock cooler)

Temperatures are now much higher than they were pre-replacement, idle and load.

-Idle, 32-39c
-Load (farcry 4, ULTRA, 1080p, for 3 hours) 62-69c
-Prime 95 (max temp stress test) 91-98c around 6 seconds into test

I don't care for benchmarking, so prime 95 doesn't bother me. I would just like to know what could be causing the spike in temperature. A list of things I've done since my concern are as follows

- used arctic silver 5 replacement thermal paste
- I actually lost one of the waterblock mounting screws, so theres 3 tightly secured instead of 4, but the temperatures remain the same even when I had 4 on the replacement cooler, it's currently the only variable
- I've gotten rid of boost, so it runs at 4.00Ghz load, rather than 4.4
- Set the coolers fans to 100% speed when temps hit 60c
- Ensured correct thermal application
- Maintained Cpu's stock voltage

is 69c steady playing a game like farcry 4 negligable? Or should I be considering remedies to that sort of temperature? Is the missing mounting screw affecting arctic silver 5's performance? Like I said, the temps were the same even before the missing screw.

Cheers!
 
Solution


Yep, you have the basic process. Re-enable turbo. Decrease the CPU voltage by small amounts (Try two steps at a time). You'll likely be using a "voltage offset" option in the BIOS rather than entering in the exact voltage, which preserves lower voltage at idle. Test prime95 for an hour or so (or if you are impatient like me, 15 minutes) if there are errors then up the voltage a few notches, do a thorough prime95 test (several hours at least), and you're done. It'll be a bit time consuming.

The reason for taking it...
While it is debatable, when only a single fan is available I much prefer to have it as an exhaust fan. I've found a single exhaust to be much more effective than a single intake, so first thing I would do is change that. (basically, I found that with a single intake more often than not the cooler air found quicker ways to leave the case and didn't push out much of the hot air, while a single exhaust at the back of the case immediately removes hot air from around the CPU/motherboard/videocard)

At the 100% fan speed, 4ghz stock voltage you are talking about that doesn't sound right. Do some runs with the case open to see if access to more air helps (in an attempt to find the exact issue) and see if you can fix that waterblock mounting if possible... even mounting pressure should be more effective, at least theoretically.

Haswell (and Ivy Bridge) are notoriously hot runners though, so perhaps this is normal... Frankly it's not acceptable in my opinion, I really wish intel had a better thermal conductivity with the heatspreader.
 



Thank you for the response. I'll have to see if there are any places about I can pick up screws, hopefully they aren't unique to corsair and I can pick up one from any hardware store.

I will be readjusting the fans when I'm home from work, and will update the thread when done so. I can confirm immediately that having an open case has no impact on temperatures, save for maybe 1-2c on GPU, but it's temperature doesn't bother me one bit.

Is the cpu temp is considered safe, I won't be to nitpicky about it, I just wish it would run as cool as it used to.
 


Those load temperatures for gaming are perfectly safe, so you are fine to keep it how it is honestly (I would consider under 80c to be safe, but lower is better). If you're interested in keeping temperatures down and not interested in overclocking, you could turn boost back on and leave default clocks and then reduce voltage until you can't pass prime95, then up it a notch or two so you keep stability. That will definitely reduce heat. It's like overclocking in reverse, and without the risk of frying your CPU.
 
Awesome, cheers. I really appreciate the advice

So to clarify, I should re-enable turbo mode, and then reduce voltage? Find a suitable level and then bump it up a little? What is the best way to get this done?
 


Yep, you have the basic process. Re-enable turbo. Decrease the CPU voltage by small amounts (Try two steps at a time). You'll likely be using a "voltage offset" option in the BIOS rather than entering in the exact voltage, which preserves lower voltage at idle. Test prime95 for an hour or so (or if you are impatient like me, 15 minutes) if there are errors then up the voltage a few notches, do a thorough prime95 test (several hours at least), and you're done. It'll be a bit time consuming.

The reason for taking it slowly is that if you go too low the PC won't boot, and you have to use the BIOS reset pins on the motherboard. It's quite easy, but a hassle, so it's best to avoid doing that. If you have a dual-bios motherboard (I believe many gigabyte ones do) this is a non-issue.

While it doesn't tell you how to do it, here is a bit of reading on voltage vs temps vs clockspeed. This article was for Ivybridge, but haswell behaves very similarly. It gives a good example of just what kind of differences it makes.
http://www.anandtech.com/show/5763/undervolting-and-overclocking-on-ivy-bridge
 
Solution



Worked a treat, down to around 57-62, will be installing fans today, thanks!