Page 1 of 1

High temps on 2600K (new build)

Posted: Wed May 04, 2011 8:29 am
by FakeAlGore
I received all of the parts for my new build (see my final post in the thread) yesterday, and I assembled it excitedly. It powered on to the UEFI BIOS on the first try.

But then I noticed the CPU temperature kept going up. It leveled off at 65°C at idle just in the BIOS. That just doesn't seem right. I see folks on other threads and forums talking about 40°C idle temps, and I fear I've done something wrong.

It's possible I don't have the HSF installed correctly. I used the stock cooler. When I pushed in on the pins, they would click into place, but almost every time another one would pop back out. I eventually got all four of them to stick in the down position (and verified the black bits were through on the backside), but I still think it's possible I messed this up. I used the tiger-striped pre-applied thermal paste, as well.

I think I'll run by my local computer parts retailer to pick up the highly-recommended Cooler Master Hyper 212+ on my way home. I have some unopened thermal paste I can use.

And a quick question: when I was lowering the handle to lock the CPU in place, it sounded like I was crunching something. I quintuple-checked everything, but no matter how many times I tried it there was always that sound. Is this common with LGA 1155? The last processor I put into a system was a Socket 478, so the pins were on the processor instead of the motherboard.

Re: High temps on 2600K (new build)

Posted: Wed May 04, 2011 8:41 am
by gbcrush
Granted, I haven't played with a stock cooler in ages, but I'd say 65 C sounds about right for the stock Intel job. Someone else was asking about high-temps, using the stock intel cooler and they were running at about 75.

OK, so my anti-stock-cooler fanboyism is showing a little here. All the same, take your time, be patient and put on your 3rd party heatsink correctly. I'd also suggest adding a quality (and quiet) 120mm fan, like the Nexus RealSilent 120.

...but however you do it, enjoy watching the idle temperature drop. :)


EDIT: Not sure about the crunching noise. I think what you'd have to worry about is cracking the heat-spreader plate ontop of the CPU. But using the lever shouldnt do that--this usually only occurs when one over-tightens a badly designed heatsink.

If you crunched the pins, I think you'd know. It be unlikely for your PC to work. :)


ALSO EDIT: If you're going to attach a new heatsink, I recommend cleaning off the CPU before re-applying thermal grease. Rubbing alcohol works well for that.

Re: High temps on 2600K (new build)

Posted: Wed May 04, 2011 8:48 am
by derFunkenstein
The current pushpin system is friggin awful. I never ever ever use them because it always feels like I'm bending the board unnaturally. I think if you get a bolt-though heatsink and apply thermal paste correctly you'll see idle temps in the high 30s or low 40s, and load temps lower than your current idle temp. Intel really screwed the pooch on its current locking mechanisms.

Re: High temps on 2600K (new build)

Posted: Wed May 04, 2011 8:49 am
by FakeAlGore
Thanks for the tips, gbcrush and derFunkenstein.

I was thinking along the same lines re: the chip not working if I had actually crushed something, and I'm glad to hear 65°C isn't an abnormal temperature for that setup. Either way, I'll still pick up the Hyper 212+ this afternoon and will post back with my temperatures.

One last question: I found this guide to BIOS settings on the ASUS P67 boards. Do you or does anyone else see anything wrong with it? If not, I'll likely go through it tonight once Windows is installed and working.

Re: High temps on 2600K (new build)

Posted: Wed May 04, 2011 10:25 am
by mmmmmdonuts21
I think the temps are very high idle. With a stock heatsink on my Xeon E3-1230 @ 100% load I am right around that temperature. I know your getting a new heatsink which is a great idea, but I think you just seated the chip incorrectly. I know with intel stupid push pin technology its tough to do correctly but the way I check is I try and wiggle the heatsink once I think its clamped down. I apply a fair amount of force and if it moves a little bit I know its not seated correctly.

Re: High temps on 2600K (new build)

Posted: Wed May 04, 2011 10:33 am
by DancinJack
Yeah, 65C is too high. The stock Intel cooler, while on my i7-860, kept it about 50C during the summer with ~75F ambient. That push-pin crap needs to go. Bolt-through needs to become standard for enthusiast processors.

Re: High temps on 2600K (new build)

Posted: Wed May 04, 2011 10:49 am
by CAIN
I thought the CPU was at 75% when in the BIOS, and not idle? Or has this changed now?

Re: High temps on 2600K (new build)

Posted: Wed May 04, 2011 11:05 am
by mmmmmdonuts21
The CPU is not at 75% in BIOS at all. It is usually idle.

Re: High temps on 2600K (new build)

Posted: Wed May 04, 2011 12:49 pm
by Scrotos
From my adventures in CPU temps: viewtopic.php?f=2&t=76425

Yeah, idle at 65C is high.

And for that guy wondering about 75% CPU use, well, snag a copy of Prime95, that'll get 'em workin' (not in BIOS though).

Re: High temps on 2600K (new build)

Posted: Wed May 04, 2011 12:55 pm
by ludi
Wait, I thought the advanced power savings features aren't enabled during the BIOS routine, meaning the CPU is not at the same idle value that it will experience when fully booted and running a modern OS?

Re: High temps on 2600K (new build)

Posted: Thu May 05, 2011 8:45 am
by FakeAlGore
I installed the new cooler last night (with quite a bit of swearing).

My idle temps in the UEFI BIOS are now at 44°C +/- 2°C. That's a significant improvement, and I'm going to say this problem is solved.

What's interesting to note is the idle temperature in Windows. It's at 40°C +/- 1°C. This leads me to believe CAIN might be right in regards to the UEFI BIOS imposing either a slight load on the CPU or not utilizing HLT instructions and SpeedStep.

Now that I have Windows and the AI Suite installed, I'll proceed to test it under load tonight (maybe after an hour or two of Just Cause 2). Once I feel confident that my temperatures are under control, the journey to 4.8 GHz will begin.

Re: High temps on 2600K (new build)

Posted: Thu May 05, 2011 8:50 am
by gbcrush
That's great news. And thanks for following up. I'm sure I'm not the only gerbil TR that likes to hear about things working out.

Congrats on the new rig, the lower temperatures, and the excuse to give it some "testing." :D

Re: High temps on 2600K (new build)

Posted: Thu May 05, 2011 8:54 am
by mmmmmdonuts21
Quick and dirty way to see how hot your cpu will run. Use either open hardware monitor http://openhardwaremonitor.org/ or CPUID Hardware Monitor http://www.cpuid.com/softwares/hwmonitor.html, run LinX http://www.xtremesystems.org/forums/showthread.php?t=201670 and put it on 4GB or 8GB of memory and look at the temperatures of the system when it is at load. It won't be there initially but after 15-20 seconds you will see your processor temperature skyrocket. If you can run LinX your gaming temps will be usually lower if you have enough cooling in your case. Just a suggestion.

Re: High temps on 2600K (new build)

Posted: Thu May 05, 2011 8:55 am
by DancinJack
FakeAlGore wrote:
he AI Suite installed...


Delete that bloat crap.

Overclocking should be done in the BIOS. Check temps with CPUID HWMonitor or similar. (speedfan or realtemp - can someone attest to the reliability of those please)

Re: High temps on 2600K (new build)

Posted: Thu May 05, 2011 9:14 am
by Scrotos
ludi wrote:
Wait, I thought the advanced power savings features aren't enabled during the BIOS routine, meaning the CPU is not at the same idle value that it will experience when fully booted and running a modern OS?


Ok, but what is the CPU doing? Is the BIOS a multithreaded application hitting more than one core and really getting some FP calculations pushed through? I think even a modern OS "idling" still has a ton of services in the background popping up from time to time, listening for events.

I could be wrong but logically it doesn't make sense to me that a BIOS (even EFI) would be hitting the CPU as hard as a modern OS. Sure it might not be throttling down the MHz or voltage, but still it's not exactly peggin' the CPU at 75% or even more than 5% or less.

Re: High temps on 2600K (new build)

Posted: Thu May 05, 2011 9:28 am
by FakeAlGore
DancinJack wrote:
FakeAlGore wrote:
he AI Suite installed...


Delete that bloat crap.


After using it for a few minutes, I'm inclined to agree with your assessment. I'll remove it tonight and use HWMonitor to check on my temperatures. Do you know if it can write to a log file at a predefined interval? I plan to run the test overnight, and I'd like to see if the temperatures are stable throughout testing instead of just knowing it isn't too hot when I wake up.

Scrotos wrote:
I could be wrong but logically it doesn't make sense to me that a BIOS (even EFI) would be hitting the CPU as hard as a modern OS. Sure it might not be throttling down the MHz or voltage, but still it's not exactly peggin' the CPU at 75% or even more than 5% or less.


It's my assumption that the UEFI BIOS isn't utilizing the HLT instruction to properly power down the CPU at idle because it is run as a single process without a separate kernel handling such things.

For a real-world example, the Windows 9x-based operating systems did not implement the HLT instruction because there was not a system idle process like in Windows NT-based operating systems. To properly power down the CPU during idle on such systems, you had to run a program such as Waterfall Pro, Rain, or CPUIdle that would do nothing but run in the background issuing HLT instructions at the lowest CPU priority. This lowered power usage and heat dissipation.

Re: High temps on 2600K (new build)

Posted: Thu May 05, 2011 9:38 am
by DancinJack
FakeAlGore wrote:
After using it for a few minutes, I'm inclined to agree with your assessment. I'll remove it tonight and use HWMonitor to check on my temperatures. Do you know if it can write to a log file at a predefined interval? I plan to run the test overnight, and I'd like to see if the temperatures are stable throughout testing instead of just knowing it isn't too hot when I wake up.


CoreTemp can.

Re: High temps on 2600K (new build)

Posted: Thu May 05, 2011 11:43 am
by FakeAlGore
DancinJack wrote:


I just tried this out on my work laptop, and it is absolutely perfect for what I'm trying to accomplish. Thank you!

I'll report back tomorrow with my temperatures under load.

Re: High temps on 2600K (new build)

Posted: Thu May 05, 2011 3:48 pm
by Scrotos
FakeAlGore wrote:

Scrotos wrote:
I could be wrong but logically it doesn't make sense to me that a BIOS (even EFI) would be hitting the CPU as hard as a modern OS. Sure it might not be throttling down the MHz or voltage, but still it's not exactly peggin' the CPU at 75% or even more than 5% or less.


It's my assumption that the UEFI BIOS isn't utilizing the HLT instruction to properly power down the CPU at idle because it is run as a single process without a separate kernel handling such things.

For a real-world example, the Windows 9x-based operating systems did not implement the HLT instruction because there was not a system idle process like in Windows NT-based operating systems. To properly power down the CPU during idle on such systems, you had to run a program such as Waterfall Pro, Rain, or CPUIdle that would do nothing but run in the background issuing HLT instructions at the lowest CPU priority. This lowered power usage and heat dissipation.


Yeah yeah yeah HLT, NOP, Win9x, etc. I remember/know that bit, but what I'm saying is that even if one of the cores is 100% going on NOP or something similar, it's not nearly the same as if it's encoding video and the cores have to really work at branch prediction, instruction decoding and recombining, and hitting the FPU and integer units hard.

I suppose my contention is that even at 100% single core NOP in BIOS it's not really going to test... anything. And might as well just ignore whatever "load" that is for the purposes of testing heat and all that.

Re: High temps on 2600K (new build)

Posted: Thu May 05, 2011 7:53 pm
by morphine
axeman wrote:
Most machines will have higher idle temperatures showing in the BIOS versus the OS, UEFI or not, because they're not bothering with HLT instrucations or whatnot. That's nothing curious. No one else ever sat in the Hardware Monitoring screen in the BIOS watching all the numbers? 8)

I did for a second, then a thought came to my head: "I actually have a life!" :P

Re: High temps on 2600K (new build)

Posted: Fri May 06, 2011 9:13 am
by FakeAlGore
Scrotos wrote:
I suppose my contention is that even at 100% single core NOP in BIOS it's not really going to test... anything. And might as well just ignore whatever "load" that is for the purposes of testing heat and all that.


I don't think anyone's arguing that point. I just pointed out that temperatures are higher in the UEFI BIOS than in Windows probably due to HLT not being used. I think we completely agree on the only point I made.

As for testing under load, I now have my first official gripe of the new build. The 200mm chassis fans both have bright-white LEDs installed on them that can be seen through the ventilation grills. Since I can't shut the door between my office and our bedroom, I had to turn my PC off last night to get any sleep. The lights were just too bright. I'll work on either disabling or removing the LEDs tonight.

Also, thanks to resetting the BIOS to "optimized" defaults and choosing the optimal performance option in the "EZ" mode, my temperature is now down to 29°C at idle in Windows. I can hardly wait to see what it's like under load in Prime95 and LINPACK. I was able to run Just Cause 2 at 2048x1152 with every setting at its maximum at 50 FPS. That's much better than my 30 FPS at 800x600 with every setting at its lowest that my previous PC did!