i5 2500k bad temperatures

[SentinelAeon]

Hello, i am having high temperatures, trying to figure out what is wrong. HWMonitor shows around 75W draw from i5 2500k under PRIME95 inplace test and i reach max of 65C.

CPU: i5 2500k default clock, undervoltage at 1.1V
COOLER: Xigmatek SDT1283

This Xigmatek cooler is old but if u check old tests, it has no problem keeping hugely overclocked c2q processors at bay. I even compared it to other coolers, cross reference them and came to a conclusion that default i5 2500k should NOT go over 50C in prime. What is more, i even undervolted the proc.

I checked everything so there is only 1 thing that remains. I sanded my i5 2500k a while ago. But, i got impatient and actualy used drill sander attachment to sand it. Now the surface is straight when i use a ruler. BUT, i am pretty sure the problem is that it is unevenly sanded. I think some sides of cpu are more sanded then others. Now when using ruler, it will still show straight surface. But when you put it into motherboard socker and cooler on it, i am pretty sure there is a gap on 1 side.

The problem is, i have no idea how to verify and test this. So i am here asking for some ideas on how to do that. Once again, cooler is old but a very capable coolers and it should have no trouble keeping this cpu under 50C. Btw not sure if its relevant but, all parts of cooler are pretty cold when i touch them even under load. While my graphic card when its at 65C, its heatpipes are hot so i can barely touch them. xigmatek heatpipes are cold to the touch. Even the plate covering part of heatpipes where they touch cpu is cold to the touch.

 

[Mandrake4565]

Not really sure where you're getting your info from but 65 c is not hot with a 2500k using Prime 95 as a stress test. Temps seem about right, without knowing any of your other hardware.

 

[trents]

You can check for unevenness of the cooler base by looking at the spread pattern of the thermal paste. If the coverage seems even you are making good contact. If not even, you are not making good contact.

Using a sanding disk on the end of a drill to "lapt" the cooler is not a preferred method.

But I agree with Mandrake. The temps you quote after running Prime95 seem actually pretty normal. Were they higher before sanding your cooler when you ran Prime95?

 

[Tech Tweaker]

Under 50°C while under load is unlikely.

My 2500K under full load running P95 blend or small FFT's was around 54-58°C with a 240MM AIO or Dual Tower heatsink if I remember correctly. Been a while since I tested it, so I can't remember what cooler or heatsink I was using. Stock cooler was 60-65°C+ for me.

Also, I think you mean the HDT-S1283.

That cooler I don't think natively supported the LGA1155 socket type, did you have to modify it to make it fit?

 

[SentinelAeon]

HDT S1283 yes. Well i still think its high, i was supposed to go 4500Mhz with cooler such as mine. It doesnt natively support LGA 1155, but if u take the pins out and replace them with screws, it fits.

Im gonna try today 1 last time and see if i can do anything bout the cpu. I will measure the sides to see if they are even, nothing more i can do. It really bothers me though that at 65C cpu temp, the heatsink is basicly cold. Like i said, if i touch my R9 390 at the same temperature, everything is hot, heatpipes, even the backplate, and i cant hold heatpipes for long without getting to hot.

 

[Zerileous]

Okay to address this issue with touching a CPU vs GPU cooler. The GPU is a 275W card and the CPU is 95W stock and under volted. Temperature is how hot a particular part is, but not how much energy is emitted from the part as heat. This can seem contradictory, but it depends on a lot of factors such as getting heat out of the part.

Bottom line, touching the cooler is not a good measurement of how well it's working. If you do want to do more sanding on the IHS of the CPU, I would recommend drawing an X with sharpie, or even covering the whole surface with sharpie. Sand on a hard flat surface like glass. The sharpie will show you if it's being sanded evenly. Be mindful not to over round the edges.

 

[EarthDog]

Im gonna try today 1 last time and see if i can do anything bout the cpu. I will measure the sides to see if they are even, nothing more i can do. It really bothers me though that at 65C cpu temp, the heatsink is basicly cold. Like i said, if i touch my R9 390 at the same temperature, everything is hot, heatpipes, even the backplate, and i cant hold heatpipes for long without getting to hot.

A lighter with a yellow flame and bonfire with yellow flames are both the same temperature...which do you think has more energy and difficult to cool??

Anyway... what is the in place test? Small fft? Blend? Large? I'm assuming one of the small fft?

 

[Mandrake4565]

HDT S1283 yes. Well i still think its high, i was supposed to go 4500Mhz with cooler such as mine.

What is making you think it's high? Give us more information and maybe we can give you more feedback. What are your ambient temperatures? What case do you have? How many fans etc.....

4.5 Ghz is achievable on a 2500k but there is no "supposed to" when overclocking. There are so many other variables to take into account.

 

[trents]

HDT S1283 yes. Well i still think its high, i was supposed to go 4500Mhz with cooler such as mine. It doesnt natively support LGA 1155, but if u take the pins out and replace them with screws, it fits.

Im gonna try today 1 last time and see if i can do anything bout the cpu. I will measure the sides to see if they are even, nothing more i can do. It really bothers me though that at 65C cpu temp, the heatsink is basicly cold. Like i said, if i touch my R9 390 at the same temperature, everything is hot, heatpipes, even the backplate, and i cant hold heatpipes for long without getting to hot.

This raises a red flag to me. It fits, but are the clamping force vectors evenly distributed?

 

[SentinelAeon]

Just wanted to say i was right, the cpu wasnt even - after i evenly sanded it, the max of temps dropped from 69 to 56C. Then i noticed that since i lapped the cpu a lot, the socket attachment system ware blocking the cpu cooler to make full contact. After i removed this from motherboard, max of temps dropped to 53C. To sum it up:

prime95 inplace, cpu vent set to 50%, system vents set to 50%
Before: max recorded temp 69C
After1: after lapping cpu max recorded temp 56C
after2: after removing the attaching socket system from MB cpu max recorded temp 53C

My job is done and i am happy. This will give me some headroom for overclocking for when i need it. But the day didnt go as smooth, i spent entire day so i will write here what happened.

What a day. Moved my 2 intake fans at the back of the case - made them outtake fans. Cleaned the dust, though it wasnt bad. Removed some plastic cover from PSU. Simple, fast, efficient. Then put it in and wouldnt give picture. Thought it was GPU, because it gave picture with other gpu. Then i tried x4 pcie instead of x16 and it gave picture. Then i thought it was beant pins on 1155 socket. Went around to straighten them and well, suddenly i had no picture anywhere. Spent 3 hours bending those pins around and as a stroke of luck, got it working again. Not only that, i found out that my main graphic card does indeed WORK in x16 pcie socket, IF i dont use the plug thingy on motherboard that holds it in place. Strange but thats how it is Anyway now everything seems to be working and i swear, there will have to be a god damn good reason for me to ever unscrew cpu cooler from motherboard again. Those cpu pins, aside from being a terrible engineering solution, are a huge stress. I had them all straight and it wouldnt work. I dont even know what i did to get them working again. Terrible engineering solution, just terrible.

 

[Zerileous]

Sounds more like mounting pressure than bent pins to me, especially with a substantially shortened IHS and DIY mount. Did any pins visibly appear bent? Glad you got it sorted and your attempts to straighten things out were not damaging.

Not sure what you mean by plug on the motherboard, are you referring to the lever at the back of the slot? Usually this engages automatically when the card is fully seated. Running the card not seated fully is risking a short. Maybe there is some dust shorting out the slot when you fully install the card.

 

[trents]

The only bad engineering was your first attempt at "sanding" the heatsink and then resanding it to make it level again. Removing that much material threw off all the spacing tolerances need for the cooler base to make contact with the CPU top while the retention cap was in place. So then you removed the retention cap to eliminate that problem which changed the spacing tolerances again and caused excessive mounting pressure to distort the socket pins which in turn caused the pins to either make contact with each other (shorts) or not make contact with the their mates on the underside of the CPU. And that is what gave you PCIe slot issues.

Moral of this story is you should have purchased a new cooler made to fit that socket in the first place instead of improvising. That would have saved you a lot of stress.