correct Temps I5 2500K overclock

Mister Gray

Distinguished
Nov 12, 2011
23
0
18,510
Hello Everyone,

I have an issue with my CPU giving up and down ratings on SpeedFan. I have researched the issue and found numerous similar issues but nothing with an answer that reoccurs

Specs:
Case: CM HAF 922
CPU: I5 2500k
GPU: EVGA GTX580 3GB
MOBO:ASUS P8P67 Pro (B3 revision if that matters)
RAM:Corsair Vengance 8 GB
SSD: Intel 120GB (scary I know)
PSU: 750 Watts

I normally never see the temp budge from -60 C but if it goes into sleep mode it woill come back with a temp of 124 C..........Shut it off cut back on and normal again. This was with the stock cpu before, took off and reapplied (did not put more paste on). Followed by last nights project of swapping in a CM Hyper 212 and applying well enough Artic Silver and getting the same -60

Is there anyway I can fix this to read my temps?? I wish to try an overclock now that I possess a decent enough cooler.

Thanks for any replies :D
 
Solution
The op has the i5 so you should give advice relative to his situation. I got my i5 to 4.7 before hitting a bit above 80C on the 212+ (antec 300 illusion stock fans on low, 1.35v). Temps are very dependent on roomtemp, in the summer I'll was idling in ~40c, now I'm ~30c in winter, roomtemp ~80F, ~70F respectively.

Mister Gray

Distinguished
Nov 12, 2011
23
0
18,510
Ok so with my MOBO setting on "performance" it's showing 43 C while at an idle with the CM Hyper 212 installed.

Does this appear average for this cooler? Room temps is about 72 F
 

diellur

Distinguished
Apr 7, 2011
1,345
0
19,460


In a word, SpeedStep.

To expand on that...the i5-2500K has SpeedStep technology built into it which will allow it to throttle to 1.6GHz when idling to lower its power consumption and ramp up to it's full speed when working (download CPU-Z and watch it in action). SpeedStep doesn't kick in until Windows boots, so when it's in BIOS the CPU is going full tilt...that's why it'll seem as if its hotter.

I used to use the Hyper-212, and I'd idle at between 30-32 deg C in a 700D chassis, with about 21-23 deg C ambient temps. Your reading of 28-30 deg C seems fine to me.
 

cbrunnem

Distinguished
just a related side not on how low end the hyper 212+ is, i idle at right around 29-31 with a noctua nh-d14 and an i7 2600k @4.5 ghz......

point being, if you want to to an overclock at 4.5 or above return that cpu cooler. it wont keep your temps in a comfortable zone.
 

diellur

Distinguished
Apr 7, 2011
1,345
0
19,460


I wouldn't consider the Hyper-212+ low end...it's actually one of the little gems out there that performs well above it's price point. I agree, there are better coolers for the higher OCs, but Asus also demonstrated that it was more than adequate for overclocking up to 4.6GHz with a single fan:

http://hardforum.com/showthread.php?t=1578110

 

cbrunnem

Distinguished

im just speaking from experience here. ive had both the hyper 212 and a nh-d14 that i have now on my rig.

at 4.5 and the 212 i was seeing 80*c
now the the nh-d14 i see69*c max and can run lower voltages.

yes the 212 cools well but not at 4.5.
 

diellur

Distinguished
Apr 7, 2011
1,345
0
19,460
I've got to be honest...69 deg C with the NH-D14? That seems high to me. What voltage are you running at?

I used to have the Hyper-212+ before I got the H100 (still got it on my shelf as a back-up, as it happens), and at 4.5GHz I was nowhere near that temp...high 60s, IIRC.

I'm not flaming here, but I wonder if you have some airflow issues in your case?
 

cbrunnem

Distinguished


you got to keep in mind that i am running an i7 with hyperthreading. hyperthreading itself needs more voltage so you have to things in addition that adds more heat.

so an i7 has more heat because of hyperthreading and because hyperthreading needs more voltage to be stable.

i bet if it turned hyperthreading off i would get lower temps and voltage.

its 4.5 at 1.340 volts.

i know the temps are higher then some people report but right now its 23*c in my house and my chip needs a tad more volts then some.
 
The op has the i5 so you should give advice relative to his situation. I got my i5 to 4.7 before hitting a bit above 80C on the 212+ (antec 300 illusion stock fans on low, 1.35v). Temps are very dependent on roomtemp, in the summer I'll was idling in ~40c, now I'm ~30c in winter, roomtemp ~80F, ~70F respectively.
 
Solution

cbrunnem

Distinguished


the temp differences should be very similar though.
 
Large ffts, I prefer prime and render for days on end. You will notice large ffts stress it more and create more heat than small fft; only to be compounded by longer test times. You only tested for an hour didn't you?

Small fft is just not enough stress for a good stress test, you need to be on large or blend which does both, it's similar with ibt, normal is not enough, it needs to be on extreme.
 

madchemist83

Distinguished
Ibt is overrated. All it does is heats up ur cpu to insane temps. Small ffts actually will heat up cpu faster and to a higher temps then large or blend. And it's very usefull to check ur cpu and cache stability. Usually if u borderline on voltages it will give u an error. If it passes 15-20 min it's usually good to go. Sometimes people say blend failed after 8 hours. My thought is ur power grid will have spikes ones in a while and that will cause error.
Nevertheless those temps look too high.
 

Mister Gray

Distinguished
Nov 12, 2011
23
0
18,510
Well Im assuming that an attempt at an overclock would be possible at this point yes? I have been hesitant to anything until I have read up enough. Any advice anyone could be willing to share. Thank you again for the replies. :D
 
I would agree prime is better and I hate any synthetics but prime is just easy to stress test with. Large and blend still use the cache and that sentence sounds like it doesn't test cpu. ("And it's very useful to check ur cpu and cache stability.") Wouldn't you say it's very useful to check stability of everything being oc? There's a reason why it lets you change between large, small and blend.

20min test is why your temps are lower than everyone else (and roomtemp). Of course the longer you test the higher chance you will fail. Gaming is no where near as intensive as rendering so I would say an hour is fine but 20min is not enough. And I would still suggest blend because you want to test everything. Your power grid theory makes no sense, I won't go into the reasoning as long as you have a decent psu.

If you really think those temps are not normal; compare.
http://www.overclock.net/t/968053/official-the-sandy-stable-club-guides-voltages-temps-bios-templates-inc-spreadsheet
http://www.google.com/search?ix=seb&sourceid=chrome&ie=UTF-8&q=2500k+temp+212%2B

Load temps is what really matters anyways. Sb overclocks pretty simple, just change multiplier and vcore. You will probably want to stay below 72C and 1.35v in normal operation. Tom's uses 1.35 for safe long term use, many others go to 1.4-1.45 with safe temps and it will probably still last after 5+ years.
 

matt_b

Distinguished
Jan 8, 2009
653
0
19,010

I don't think of IBT as overrated. It tells me a worse-case-scenario of what temperature a CPU could potentially reach and lets me know stability status of my system a lot faster than Orthos or Prime ever did/will. There is a reason that IBT produces "insane temps" - it stresses your system more than just about any other program out there. Some people like to still use the tried and true, and I cannot fault that - use what you feel most comfortable with.

For reference purposes (I didn't see them posted), the I5 2500 specs per Intel are:
Max Tcase = 72.6* C
Max VCore = 1.52v
 

aqe040466

Distinguished
Nov 29, 2011
886
1
19,065




I think there is something wrong with your motherboard. I am using i5 2500K OC'd to 4.7 GHZ idles at 37 deg celsius and at full load using prime95 it goes not more than 60 deg celsius. I used CM 212 hyper plus cooler. Motherboard is Biostar TZ68A+RCH.