Sign in with
Sign up | Sign in
Your question

High core temps on new i5 3570K?

Last response: in CPUs
Share
December 17, 2012 6:02:01 PM

Just put together a new rig.

Gigabyte G1 Sniper 3 Z77 board (Uefi Bios)
i5 3570k 3.4hz
Noctua NH-D14
8gb Samsung Green Ram
750W Corsair HX750
2x Samsung SSD 830 (1x 128Gb + 1x 256Gb)
GTX 480 from my old pc as the new 680 is currently RMA'd back to Gigabyte.

Temps are running at up to 47 degrees at idle... which I feel is not right. Turbo boost is turned on in the bios and the CPU is @ 3.8 ghz of its own free will, but still 47 degrees seems high. System temp is listed as 26-28 degrees. Ambient temp is 21-22.

Bios is showing cores 0, 2 @3 @ 42-47 and core 1 @ 30...

Real temps 3.70 is showing huge fluctuations in temps, it is contsatnly jumping from 23 up to 40 down to 20 up to 30 etc etc, never stays still for a second and is jumping up and down in a splt second by +10 dgree increments, which I do not know if it is faulty software or a CPU issue.

So, this is my first own complete build. Done loads of bits before, but this is the first time i have instlled a processor or heatsink.

The Noctua was a b**** to fit as it is huge, heavy and has very awkrad spring loaded screws to fit it to the mount.

Any ideas if this is caused by poor paste appication/spread or i something more sinister?

More about : high core temps 3570k

a b à CPUs
December 17, 2012 6:14:42 PM

Verify your temps using a few programs, but I would say the issue is in the application of your heatsink. I have a Noctua as well, and I did not think the installation was horrible, but mine is a downdraft as opposed to the massive tower cooler D14. How much thermal paste did you use? You can run into issues using too much as well as too little. Also, once you seat the heatsink on the thermal paste on the CPU you do not want to lift it at all, as it will create air bubbles in the paste. If you remove the heatsink at all you need to remove and reapply thermal paste, even if you simply set it on the cpu and then took it off (with paste already on of course).
a c 344 à CPUs
December 17, 2012 6:18:44 PM

Assiming your room is not abnormally hot, say 23c. then there is something amiss.

1. Are the fan/s on the cooler turning?
Are the fans on your case moving some air?

2. If they are, then I suspect a problem with the installation.
Reread the instructions to be certain that you have done the install correctly.
A fair amount of spring pressure may be needed.

How much thermal grease did you use?
A common mistake is to use too much which will act as an insulator.
The purpose of grease is to fill in microscopic pits in the mating surfaces driving out any trapped air bubbles.
A small drop in the center will spread under heat and pressure.
Try cleaning off the old grease with alcohol and a lint free papar such as a coffee filter.
Then apply only a small drop.
There are many methods, and all work reasonably well, so don't get hung up on one particular method.
Related resources
Can't find your answer ? Ask !
December 17, 2012 6:23:15 PM

chugot9218 said:
Verify your temps using a few programs, but I would say the issue is in the application of your heatsink. I have a Noctua as well, and I did not think the installation was horrible, but mine is a downdraft as opposed to the massive tower cooler D14. How much thermal paste did you use? You can run into issues using too much as well as too little. Also, once you seat the heatsink on the thermal paste on the CPU you do not want to lift it at all, as it will create air bubbles in the paste. If you remove the heatsink at all you need to remove and reapply thermal paste, even if you simply set it on the cpu and then took it off (with paste already on of course).


The issue with instll was that it takes a large amount of pressure to get the screws to take, and when one has taken it wants to spring back up on the other side. as a result it probaly squished to the left, and then to the right, rather than staying flat. if tht makes sense. It is also difficult to screw down as he provided driver is only high enough above the edge if you have the hands of a 3 yeas old girl.

I used a pea sized amount 4-5 mm across as the instuctions said.

I do not have another driver as the oter are magnetic, and i assumed that using a magnetic headed driver near a cpu was a bad plan...
a c 190 à CPUs
December 17, 2012 6:23:19 PM

+1 for Geofelt

I personally like HWMonitor for seeing the temperatures.
December 17, 2012 6:27:08 PM

geofelt said:
Assiming your room is not abnormally hot, say 23c. then there is something amiss.

1. Are the fan/s on the cooler turning?
Are the fans on your case moving some air?

2. If they are, then I suspect a problem with the installation.
Reread the instructions to be certain that you have done the install correctly.
A fair amount of spring pressure may be needed.

How much thermal grease did you use?
A common mistake is to use too much which will act as an insulator.
The purpose of grease is to fill in microscopic pits in the mating surfaces driving out any trapped air bubbles.
A small drop in the center will spread under heat and pressure.
Try cleaning off the old grease with alcohol and a lint free papar such as a coffee filter.
Then apply only a small drop.
There are many methods, and all work reasonably well, so don't get hung up on one particular method.


fans on cooler are on. PWM fans with two speeds running at thier slower setting i asume.

3x 180mm 100+ CFM fans in base of case (silverstone FT02) are running on high. Exaust fan (Kaze Juni 100CFM + is running at 1800 rpm on top of case above heatsink.

I used a pea sized glob. mayeit was too big.
a b à CPUs
December 17, 2012 6:31:14 PM

Actually, I used a magnetic driver for my build, and I believe I queried on here before, apparently the charge on the driver is not nearly enough to damage any components. I even had a response from a PC lab tech who stated he only used a magnetic driver. That sounds like a little too much for me, I used a 2-3mm "grain of rice" sized dot. And, if you think the heatsink was "teetering" back and forth while you tried to screw it in, that could have created air bubbles in the paste.
December 17, 2012 6:48:57 PM

IntelEnthusiast said:
+1 for Geofelt

I personally like HWMonitor for seeing the temperatures.


Well i am not sure exactly wht an intel enthusisast rep is... but sure sounds offical so i geuss if thats waht you reccomend then I will give it a go. I have downloaded it and it certainly gives more, and clearer info.

chugot9218 said:
Actually, I used a magnetic driver for my build, and I believe I queried on here before, apparently the charge on the driver is not nearly enough to damage any components. I even had a response from a PC lab tech who stated he only used a magnetic driver. That sounds like a little too much for me, I used a 2-3mm "grain of rice" sized dot. And, if you think the heatsink was "teetering" back and forth while you tried to screw it in, that could have created air bubbles in the paste.


I did read after i did it that some recomned a rice grain size... and I agree that it seems i may have f***** it up this time.

I am pretty nervous about remving the paste... i have soem decent paste remover i bought along wiht the components...just in case. Also some "surface purifier" to use after. I am wondering if I need to remover the cpu first, or leave it in situ?

a b à CPUs
December 17, 2012 6:56:32 PM

He's a rep for Intel, your getting advice straight from the horses mouth (or whatever that analogy is lol).
a b à CPUs
December 17, 2012 7:00:16 PM

If you are very careful you can leave it in the slot, just make sure you aren't using excessive amounts of remover on your cloth (like not dripping or anything lol). You can take a look at youtube, or if Newegg has a vid watch that, for a demonstration of removing thermal paste.
a c 89 à CPUs
December 17, 2012 8:15:42 PM

dandare said:
Well i am not sure exactly wht an intel enthusisast rep is... but sure sounds offical so i geuss if thats waht you reccomend then I will give it a go. I have downloaded it and it certainly gives more, and clearer info.



I did read after i did it that some recomned a rice grain size... and I agree that it seems i may have f***** it up this time.

I am pretty nervous about remving the paste... i have soem decent paste remover i bought along wiht the components...just in case. Also some "surface purifier" to use after. I am wondering if I need to remover the cpu first, or leave it in situ?

you dont need "paste remover", that is the first time i have ever heard of that. you just need a clean cloth or tissue to remove it, and if your anal about it the get some iso alcohol on a rag to clean any residue. pea size or rice size is crap. put some thermal paste on the cpu and smear a very thin layer over it with something hard and a flat strait edge, like a hard piece of plastic, scraping off any excess. I suggest noctua nt-h4 as it has no curing time and works better than as5.
a b à CPUs
December 17, 2012 8:18:34 PM

Rice size worked fine for me, letting the weight of the heatsink and the pressure of the screws spread it, don't know why you consider it crap, many would question your method as well. You don't need paste remover obviously, I am sure it is just branded alcohol, but of course suppliers want to push their product.
December 17, 2012 8:30:17 PM

chugot9218 said:
Rice size worked fine for me, letting the weight of the heatsink and the pressure of the screws spread it, don't know why you consider it crap, many would question your method as well. You don't need paste remover obviously, I am sure it is just branded alcohol, but of course suppliers want to push their product.


I did not say it was crap.. where did you get that from? sorry? confused.

I said I mucked up the installation, i.e. by using too much. - inferring that i should have used less than i did.

I would not really call mine a method either. first time I ever did it, just followed the instructions in the box.
iam2thecrowe said:
you dont need "paste remover", that is the first time i have ever heard of that. you just need a clean cloth or tissue to remove it, and if your anal about it the get some iso alcohol on a rag to clean any residue. pea size or rice size is crap. put some thermal paste on the cpu and smear a very thin layer over it with something hard and a flat strait edge, like a hard piece of plastic, scraping off any excess. I suggest noctua nt-h4 as it has no curing time and works better than as5.


being that i had never done it beofre i thought i may muck it up and i saw the remover on the website. Artic Clean Thermal Material Remover. I just opened and sniffed it and it smells of lemons... much like lem oil for guitar necks actually... it describes itself: Emulsifies and removes thermal compounds and pads.




a b à CPUs
December 17, 2012 8:32:46 PM

Not you, the post above me. You did no permanent damage which is good, I know I was pretty nervous installing my heatsink, feels like lots of stuff to break all at once haha.

When I was looking for the "best" method, you would not believe the flame wars I found. I did find several levelheaded posters who pointed out that success is possible with each method, the temp is what matters, as long as it has worked for someone I would consider it a valid method.
a c 344 à CPUs
December 17, 2012 8:58:40 PM

dandare said:
Well i am not sure exactly wht an intel enthusisast rep is... but sure sounds offical so i geuss if thats waht you reccomend then I will give it a go. I have downloaded it and it certainly gives more, and clearer info.



I did read after i did it that some recomned a rice grain size... and I agree that it seems i may have f***** it up this time.

I am pretty nervous about remving the paste... i have soem decent paste remover i bought along wiht the components...just in case. Also some "surface purifier" to use after. I am wondering if I need to remover the cpu first, or leave it in situ?


I think you have a arctic clean two bottle kit.
One is used to soften and remove the old material, and the second gets any residual cleaner off.
It is a good product so go ahead and use it.

Leave the cpu in the motherboard. I think it is safer there. Just don't slop the cleaner all over the place. Remove some old material, and discard the cleaning paper and do it again, just a little at a time.
a c 89 à CPUs
December 17, 2012 11:21:47 PM

chugot9218 said:
Rice size worked fine for me, letting the weight of the heatsink and the pressure of the screws spread it, don't know why you consider it crap, many would question your method as well. You don't need paste remover obviously, I am sure it is just branded alcohol, but of course suppliers want to push their product.

if i put a rice size grain on my cpu it would not spread all the way out and would leave spots uncovered where air pockets would form. the rice size method was for back in the athlonxp days where there was no metal heat spreader on the cpu and you applied the paste directly to the die.
a b à CPUs
December 18, 2012 12:25:33 PM

Well, worked fine for me, my temps are great.
December 18, 2012 6:43:28 PM

thanks to all above for answers. I can say that the pea sized amount was too much. it has actually come out of the sides of the heat sink in two places...

The over all spread was good, but just too much. I just spent over an hour painstakingly cleaning the cpu and heatsink, and as stated above by geofelt the product is good stuff. the cleaner is a citrus thing that wroks really well, and the surface prep brought them bothe back to new looking.

I used less paste, a lttle larger than a grain of rice, but smaller than a pea...

Now the issue I have is that when I fired it all back up I have the same crazy jumping temps that i saw before. Min temp on core 2 is 15 dgrees vs 28 on core 3

max temp recorded after 1 min of pc on time was 44 degrees, but is jumping up and down like a thing possessed, and the pc is just idle.

Really stsrting t worry that there is something wrong here, which my stress levles cannot take... my brand new gtx 680 is currently RMA'd nd if it turns out that the mobo or cpu are broke too I may have a fit.

Any suggstions?

the boost mode is on in the bios and the cpu is at 3798 according to that, but still 44 seems high, as I have seen temps reported at 23-25 idel for this cpu and most particulalry this Noctua cooler. i wonder idf the fans are not working right?

December 18, 2012 6:52:29 PM

dandare said:
Just put together a new rig.

Gigabyte G1 Sniper 3 Z77 board (Uefi Bios)
i5 3570k 3.4hz
Noctua NH-D14
8gb Samsung Green Ram
750W Corsair HX750
2x Samsung SSD 830 (1x 128Gb + 1x 256Gb)
GTX 480 from my old pc as the new 680 is currently RMA'd back to Gigabyte.

Temps are running at up to 47 degrees at idle... which I feel is not right. Turbo boost is turned on in the bios and the CPU is @ 3.8 ghz of its own free will, but still 47 degrees seems high. System temp is listed as 26-28 degrees. Ambient temp is 21-22.

Bios is showing cores 0, 2 @3 @ 42-47 and core 1 @ 30...

Real temps 3.70 is showing huge fluctuations in temps, it is contsatnly jumping from 23 up to 40 down to 20 up to 30 etc etc, never stays still for a second and is jumping up and down in a splt second by +10 dgree increments, which I do not know if it is faulty software or a CPU issue.

So, this is my first own complete build. Done loads of bits before, but this is the first time i have instlled a processor or heatsink.

The Noctua was a b**** to fit as it is huge, heavy and has very awkrad spring loaded screws to fit it to the mount.

Any ideas if this is caused by poor paste appication/spread or i something more sinister?


The high temps are caused by too many transistor (transistor density) on a small die (die shrink)(Intel's clever way of earning more cuz you can then manufacture more cpu per waffer which also saves cost and earn big) which causes less surface area for the CPU die to transfer heat to the IHS which in turns releases the heat to the environment via a heat sink and a fan so theres nothing we can do about it accept buying a better cooler. :sweat: 
December 18, 2012 7:13:49 PM

Yuna said:
The high temps are caused by too many transistor (transistor density) on a small die (die shrink)(Intel's clever way of earning more cuz you can then manufacture more cpu per waffer which also saves cost and earn big) which causes less surface area for the CPU die to transfer heat to the IHS which in turns releases the heat to the environment via a heat sink and a fan so theres nothing we can do about it accept buying a better cooler. :sweat: 


The NH-14D is one of the top 3 air coolers around. That is why I chose it. i "should" be getting idel temps in the 20's not 40's
December 18, 2012 7:48:24 PM

OK, so just OC to 4.6 and hit 90 degrees in about 30 seconds running prime. which is not good. Something is not right here. PWM fans on the Noctua hit +1300 rpm... but temps were sky high.

Now I am not an experienced Overclocker, all I did was adjust the multiplier from 34 to 46, no other changes, but this all seems too hot for me. please help.
a c 344 à CPUs
December 18, 2012 8:53:13 PM

dandare said:
OK, so just OC to 4.6 and hit 90 degrees in about 30 seconds running prime. which is not good. Something is not right here. PWM fans on the Noctua hit +1300 rpm... but temps were sky high.

Now I am not an experienced Overclocker, all I did was adjust the multiplier from 34 to 46, no other changes, but this all seems too hot for me. please help.

4.6 if stable is very good. That 90c would not surprise me, that is a very agrressive overclock to start with.
Also, not all chip samples will do a high overclock. I might expect 4.3 as a conservative target.

I suggest you do not use the motherboard overclocking utilities; they can be overly agressive.
For a conservative OC, start raising the multiplier gradually. Leave everything else on auto.
Monitor with cpu-z. Verify that your multiplier is ok.
See what the voltage is. You should not get high heat until the voltage is increased.
I like realtemp to monitor temperatures. TJMAX should be about 105c.
At that point, the cpu will downclock to prevent damage.
When idling, expect the multiplier to drop to 16. That is normal; speedstep doing it's job.
December 18, 2012 9:56:36 PM

geofelt said:
4.6 if stable is very good. That 90c would not surprise me, that is a very agrressive overclock to start with.
Also, not all chip samples will do a high overclock. I might expect 4.3 as a conservative target.

I suggest you do not use the motherboard overclocking utilities; they can be overly agressive.
For a conservative OC, start raising the multiplier gradually. Leave everything else on auto.
Monitor with cpu-z. Verify that your multiplier is ok.
See what the voltage is. You should not get high heat until the voltage is increased.
I like realtemp to monitor temperatures. TJMAX should be about 105c.
At that point, the cpu will downclock to prevent damage.
When idling, expect the multiplier to drop to 16. That is normal; speedstep doing it's job.


Thanks for the advice. That is bascilay what I did. left all else on auto, as I did not really know what it does. I have the Gigabyte G1. Sniper 3, which I bought because it was meant to be a top board, good for stable overclocks, and run cool.

I also bought a silverstone FT02 case, whcih came top, or close to, in most benchs I looked at for air cooling.

This is why I am concerned with high idle temps... lokking at xaples online the Noctua shuld give 25-28 idle ish... not 40. Any ideas on why this might be?

EDIT*** so I unclocked it + updated to the latest F8g Beta Bios, just to see if it would help.

I booted back up and same thing is still hapening. HW monitor shows min temp of 20 and max of 48 on core 2 within 1 min of booting system. temps are jumping in this rage every split second... is this normal? should I re-apply the paste again?

I also checked CPU-Z and the clock multiplier is shown at 38, which is the boost clock. actual frequncy shows as fluctuating quite alot, not sre if this is normal either.

OK, so readng this thread it seems to be that I should not worry about idle temps http://www.tomshardware.co.uk/forum/337920-10-idle-temp... - my CPU fans on the nocta do idle at about 384rpm, so pretty low.

Also this thread suggests that disabling power saving would help stabalise: http://www.overclock.net/t/756114/cpu-temps-jumping-aro... - are they talking about bios seting?








!