overclocking 2500k heatsink

Thaifood

Thaifood

Thaifood

Associate
Joined
8 Aug 2012
Posts
781
hey people.. planning to over clock my 2500k even more.. planning to push my system to its limit.. at the moment i got it running at 4.4ghz and max temp i see in bf3 is around 68c..
i am running it on a Asus P8Z68-V LE and at the moment i got a Arctic Cooling Freezer 7 Pro Rev 2 cooling it... (i got low profile ram)
basically the question is, what is a good air cooler which i can change to?
now i dont plan to spend anymore than £40..
any help would be brillaint..
thanks
 
I'm using a Thermalright Macho in my Fractal Arc Midi case, the Macho is similar to the True Spirit (TS is slightly better performing - Macho is designed for low noise) to cool my 2700k @ 4.8GHz, admittedly I have 2 fans fitted in push-pull and decent case cooling but I get good enough results - currently Idling at 4.8Ghz 36c on the CPU 43c max on the cores. Stress testing will see the cores get to the low 80s whilst gaming they are usually mid to high 50s. I'd budget for a decent paste also, I used Gelid Extreme which is great price/performance wise.

macho.jpg
 
Last edited:
It's not as good as the True Spirit 140. If you Google you'll find XBit Labs did a review and at the same fan speed (1000rpm) the Freezer i30 was 7c hotter than the TS-140 under load (85c vs 78c). TBH you really should just go with the True Spirit 140.
 
PieEater said:
It's not as good as the True Spirit 140. If you Google you'll find XBit Labs did a review and at the same fan speed (1000rpm) the Freezer i30 was 7c hotter than the TS-140 under load (85c vs 78c). TBH you really should just go with the True Spirit 140.
Click to expand...

Haven't got one yet. Haven't had the time.
Bit will be ordering it next week
 
well got the thermalright True Spirit 140 CPU Cooler installed and got the cpu running at 4.4ghz (1.25volts). did intel buren test on standard for about 30 minutes and max temp on one of the cores was 79c... is that a bit high? wish i took some screen shots with the old cooler now as honestly i cant remember what it was..
just trying to see whats a decent temp.. if it isnt, will have to try reapplying thermal paste.. maybe i put too much on, but it does say use a pea size ammount
 
Thaifood said:
well got the thermalright True Spirit 140 CPU Cooler installed and got the cpu running at 4.4ghz (1.25volts). did intel buren test on standard for about 30 minutes and max temp on one of the cores was 79c... is that a bit high? wish i took some screen shots with the old cooler now as honestly i cant remember what it was..
just trying to see whats a decent temp.. if it isnt, will have to try reapplying thermal paste.. maybe i put too much on, but it does say use a pea size ammount
Click to expand...

I only use IBT in short runs to test stability, its best not used for long period IMO. Try running Prime Large FFTs for a an hour or two (obviously keeping tabs on temps), temps will probably be a lot cooler.

I get 80c from a short run on IBT, topping out on prime at about 69c in the recent hot weather we've had. My A50 only cost about £25 but looks very similar in size and style to your new cooler so you should get similar if not better temps as you cooler is more expensive so probably better quality.
 
I would say that 79c at 1.25v is high for the TS140 but it depends on other factors such as case cooling, what speed your CPU fan was running at, and what paste you used. Are you getting the 1.25v figure from BIOS or reading what is actually being drawn via CPUz?

When doing a default run of IBT on my 2700K at 4.8GHz my volts peak at 1.424 and my hottest core gets to 84c, I would expect the TS140 to be a couple of degrees cooler at the same volts which I don't think yours would be.

So initially I would use CPUz to confirm the peak voltage when running a default IBT test (as bainbridge suggests - use Prime for long runs) making sure all your fans are running at 100%. If your peak voltage is indeed 1.25v and your CPU fan is at 100% and your max core is still 79c then you may be right that you have put too much paste on. Did you spread the paste or just fit the cooler straight on the blob of paste? Spreading can cause air pockets which prevents the heat transferring between the CPU and cooler properly. A full on pea sized blob is possibly too much, some people say the amount should be the size of a grain of rice so I normally go for something in between.
 
PieEater said:
I would say that 79c at 1.25v is high for the TS140 but it depends on other factors such as case cooling, what speed your CPU fan was running at, and what paste you used. Are you getting the 1.25v figure from BIOS or reading what is actually being drawn via CPUz?

When doing a default run of IBT on my 2700K at 4.8GHz my volts peak at 1.424 and my hottest core gets to 84c, I would expect the TS140 to be a couple of degrees cooler at the same volts which I don't think yours would be.

So initially I would use CPUz to confirm the peak voltage when running a default IBT test (as bainbridge suggests - use Prime for long runs) making sure all your fans are running at 100%. If your peak voltage is indeed 1.25v and your CPU fan is at 100% and your max core is still 79c then you may be right that you have put too much paste on. Did you spread the paste or just fit the cooler straight on the blob of paste? Spreading can cause air pockets which prevents the heat transferring between the CPU and cooler properly. A full on pea sized blob is possibly too much, some people say the amount should be the size of a grain of rice so I normally go for something in between.
Click to expand...

It's the number I got from the asus overclocking tool in windows.. I will do anther run and look at CPUz.. And I did spread the paste myself..most I would have just put a small rice size amount and put the cooler onto the CPU then and let it spread itself?
 
bainbridge said:
I only use IBT in short runs to test stability, its best not used for long period IMO. Try running Prime Large FFTs for a an hour or two (obviously keeping tabs on temps), temps will probably be a lot cooler.

I get 80c from a short run on IBT, topping out on prime at about 69c in the recent hot weather we've had. My A50 only cost about £25 but looks very similar in size and style to your new cooler so you should get similar if not better temps as you cooler is more expensive so probably better quality.
Click to expand...

What version of prime? Sounds like maybe the one you're using doesn't have AVX instructions as I've found its within a few C of IBT if not the same, also if its an older version without AVX it'll register much lower temps and not test stability properly. Or are you only running large FFTs? it won't really test stability, its like testing the limits of your car but only driving at 20mph. You want to do a blend test to get a somewhat accurate picture of stability and max temps.

Long runs with IBT are no problem, I test my builds cpus with 100 runs at max (maybe 500+ cpus since IBT came out), few hours at 80C isn't anything and won't affect the chip life in any measurable way.
 
Last edited:
Thaifood said:
And I did spread the paste myself..most I would have just put a small rice size amount and put the cooler onto the CPU then and let it spread itself?
Click to expand...

I don't really understand what you've said above but it sounds like you applied a fairly large amount of paste and then spread it yourself over the CPU. Proper metal to metal contact between the cooler and CPU is better at transferring heat than metal-paste-metal. The paste is there to fill in any gaps because it will be much better at conducting heat than air. The idea of putting a blob of paste on the CPU and then fitting the cooler directly on top is that the paste will be forced to go where it is needed and in the process push out any air pockets (simplified explanation). So you may well get better results if you clean off the old paste and re-apply it (about half a pea) without spreading it.
 
PieEater said:
I don't really understand what you've said above but it sounds like you applied a fairly large amount of paste and then spread it yourself over the CPU. Proper metal to metal contact between the cooler and CPU is better at transferring heat than metal-paste-metal. The paste is there to fill in any gaps because it will be much better at conducting heat than air. The idea of putting a blob of paste on the CPU and then fitting the cooler directly on top is that the paste will be forced to go where it is needed and in the process push out any air pockets (simplified explanation). So you may well get better results if you clean off the old paste and re-apply it (about half a pea) without spreading it.
Click to expand...

Sorry was on the train and haven't had my morning coffee yet lol..
Yes I spread the paste myself.. Sounds like I did too much as from looking at the figures they are not much different to my old cooler. Wen I get time will dismantle the cooler and put a rice size amount of paste and let the cooler/CPU contact spread the paste out for me
 
Thaifood said:
Sorry was on the train and haven't had my morning coffee yet lol..
Yes I spread the paste myself.. Sounds like I did too much as from looking at the figures they are not much different to my old cooler. Wen I get time will dismantle the cooler and put a rice size amount of paste and let the cooler/CPU contact spread the paste out for me
Click to expand...

There's a nifty video where a dude simulates various methods using a clear piece of glass as the cooler contact and you can see all the air bubbles from when he tries the self spread method. Pea and press is the way forwards. If you can, a little twist while it's pressed down helps get some air bubbles out.
 
Justintime said:
What version of prime? Sounds like maybe the one you're using doesn't have AVX instructions as I've found its within a few C of IBT if not the same, also if its an older version without AVX it'll register much lower temps and not test stability properly. Or are you only running large FFTs? it won't really test stability, its like testing the limits of your car but only driving at 20mph. You want to do a blend test to get a somewhat accurate picture of stability and max temps.

Long runs with IBT are no problem, I test my builds cpus with 100 runs at max (maybe 500+ cpus since IBT came out), few hours at 80C isn't anything and won't affect the chip life in any measurable way.
Click to expand...

The version of prime I have must be very old by now as I've been using it for years. Admittedly this may not stress the system in the bang-up-to-date way you refer to, but it does load the system and I can then monitor temps and voltage.

My definition of stability is a system which does everything I want it to do without crashing, which mine has done admirably since I built it nearly 2 years ago and tested it using the method I described above, it has only gone over 70c for a few short ibt runs and I only grudgingly did them because it seemed "the thing do do".

You state you've run ibt for 100 runs on 500 cpus with temps in the 80s without any measurable damage, but that doesn't mean damage hasnt been caused. I would say doing this would reduce the lifespan of the cpu as it has been run hot for an extrended period. Personally I wouldn't buy a cpu I knew had been sitting for hours at 80c. Is this not the very reason why we have beefed up coolers? To keep the cpu cool?
 
80c is nothing to a cpu, i see OEM builds running in that range under constant daily load for years in server rooms without any damage whatsoever, the equipment is replaced long before any damage is done. The cpu might last a few hours less over its ten year or more lifespan, so i'm not too bothered.
I used to be all worried about a cpu hitting high temps, same for hdds etc.. but when you work with these things in vast quantities for years you realise that sometimes its more down to luck than environment, for example i've upgraded hdds in server units that had fan failures with alarm turned off by the technicians that were still 100% functional after being run at 70c+ for months according to SMART data etc.. yet have replaced numerous failed units in systems that were well ventilated and never went past 40C etc..

Chances of a cpu failure due to heat is immensely low and even degradation due to heat only is not really an issue, most degraded cpus were highly overvolted and that what kills most cpus over a short time.
 
Last edited:
You must log in or register to reply here.
Back
Top Bottom