What temps should I be getting with an i5 3570K cooled by a H80i?

Toggle sidebar Toggle sidebar
S

Singularity_v2

Honorable
Jun 22, 2013
11
0
10,510
Hello everyone,

I'm currently running an i5 3570K cooled by a corsair h80i. I have recently removed the pump, removed the stock paste, and re-applied arctic silver thermal paste. I'm seeing idle temps between 32 - 40 C. Under full load (3D Mark Fire Strike Extreme), I'm seeing temps between 60 - 70C. These temps are with the stock clock and 3.8 ghz turbo boost. I just feel like these temps are way too high for a liquid cooled unit as I have seen many air cooled units out perform these temps.
I would like to bump the clock up to 4.2 ghz in order to bring my physics scores up in 3D Mark and to get my cpu performance closer to the level of my gpu (GTX 970 FTW) but am afraid that the overclock is going to send the already high temps skyrocketing. Any thoughts or advice would be appreciated.

i5 3570K cooled by h80i
ASRock Extreme 4 Z77
16GB Corsair Vengeance 1600 mhz
GTX 970 FTW @ 1404 mhz
Corsair Force 128 GB SSD
Seagate 2 TB 7200 HDD
 
Sort by date Sort by votes
S

Suribachi

Reputable
Dec 8, 2014
257
0
4,860
Are you able to tweak the fans on the rad by adjusting the fan curve? Have you tried using a push/pull fan config on the rad? How is the airflow in the case?

Liquid cooling is not as good as people hype it up to be. In most cases, you are going to get an extra 800Mhz-1.2Ghz out of the processor over air.
 
Upvote 0 Downvote
clutchc

clutchc

Titan
Ambassador
Apr 22, 2009
43,150
411
131,790
That's not too bad. Seems a bit high for stock, tho. What do the max temps rise to running Intel Burn Test? The i5 in my sig below hits 63-69C running IBT with the similar H60 cooler I have.
From the posts I've read lately, the H80i must be less efficient than the H60 (?). Seems others have had rather warm temps with their H80i too. I'm not sure how well that Corsair Link works keeping the pump at full RPM. The H60's pump is connected direct to a non-controlled fan header and always runs at full RPM.
 
Upvote 0 Downvote
S

Singularity_v2

Honorable
Jun 22, 2013
11
0
10,510


The fans on the radiator are set up in push/ pull blowing air out of the back of the case. I have 2x 120mm intake fans on front of case, a 140mm fan as intake on side of case, a 120mm fan as exhaust on top of case, and the push pull radiator fans acting as exhaust as well.I have the fans set to balanced in corsair link as putting it on performance effectively turns my computer into a jet engine.
 
Upvote 0 Downvote
S

Singularity_v2

Honorable
Jun 22, 2013
11
0
10,510


I just ran Prime95 for about 15-20 minutes and the max temps were:

Core #0 = 78 C
Core #1 = 81 C
Core #2 = 77 C
Core #3 = 76 C

This seems really high to me for this cooler. breaking into to 80's worries me.
 
Upvote 0 Downvote
clutchc

clutchc

Titan
Ambassador
Apr 22, 2009
43,150
411
131,790


Try IBT @ the 10 pass default run so we can compare. i don't like running P95 if I don't have to. Besides, with a Haswell I'd have to run an older version anyway.
http://www.techspot.com/downloads/4965-intelburntest.html
 
Upvote 0 Downvote
Karadjgne

Karadjgne

Titan
Ambassador
Dec 26, 2012
30,741
4,559
142,790
What version of p95?

Running 26.6 p95 small fft, i5-3570K OC 4.2, Corsair h55, 1/2 hr test I'm @63°C on core #1 (hottest).

I'd say with your temps, there's a problem somewhere. Could be fans not maxing, a bad pump mount (that pump needs to be tight, as in needs a screwdriver for last turn), could be too much paste/not enough spread, could be CL not reporting correct temps etc. You'll have to verify/test against others like realtemp or speccy, hwmonitor, even mobo software fan controller software etc.
 
Upvote 0 Downvote
S

Singularity_v2

Honorable
Jun 22, 2013
11
0
10,510




I have been recording my temps with CoreTemp as I do not trust Corsair Link. I think I'm going to remove the pump and re-apply the thermal paste. I have been very cautious on tightening the pump down and have just been using my fingers to tighten the hold downs as they are thumb screws. The pump is firmly attached with no play in it when using this method but I will use a screwdriver to add a little more tension to the pump during re-installation this time. Ill let you know how it goes.
 
Upvote 0 Downvote
clutchc

clutchc

Titan
Ambassador
Apr 22, 2009
43,150
411
131,790
You shouldn't need a screwdriver unless you have weak fingers. I have never used one on my H60s' thumb screws. Or any thumb screws for that matter, unless I just couldn't get my fingers in to do the job. If you can't get the H80i temps down better than what you have, I'd consider RMA'ing it if you still can. They are not unsafe, just not as good as they should be.
 
Upvote 0 Downvote
S

Singularity_v2

Honorable
Jun 22, 2013
11
0
10,510


I just took the pump off and re-applied thermal paste (arctic silver 5) and tightened the pump hold downs much tighter than they were on the previous installation (used a screwdriver after tightening them flush by hand) and am happy to say that idle temps have dropped by 3-5 C and prime95 temps have dropped by 5-8 C. The hottest core was at 81C on prime95 and now the same core only reached 76C. Hoping that this will continue. Idle temps have dropped below 30C (27-29C).
 
Upvote 0 Downvote
Karadjgne

Karadjgne

Titan
Ambassador
Dec 26, 2012
30,741
4,559
142,790
That's a good start. Congrats.

The reason I asked about what version of p95 you are running is because after version 26.6, the authors changed the way the program tests the cpu. They now use an extremely aggressive avx instruction set test that's honestly junk as there isn't any program that uses even close to the amounts of avx that's pushed, and is giving totally out of control temps. This has had a huge impact on Haswells, but I'm not sure as to exactly how much impact this'll have on Ivy Bridge. Just to be sure, use version 26.6, its tested, its reliable and tests a much broader scope of instructions. If nothing else, its one more thing to check off the possible causes list.
 
Upvote 0 Downvote
S

Singularity_v2

Honorable
Jun 22, 2013
11
0
10,510


I am using Prime95 version 28.5 build 2. So this version pushes the cpu more than anything else? I was wondering because I have personally never seen my CPU pinged to 100% in any other application other than Prime95. Even 3D Mark Firestrike Extreme doesn't even max all cores to 100%.
 
Upvote 0 Downvote
Karadjgne

Karadjgne

Titan
Ambassador
Dec 26, 2012
30,741
4,559
142,790
Yeah, I'd dump that version and get 26.6. It may not make any difference to an Ivy Bridge cpu, but it'll establish a working baseline temp that the ppl around here know works as it should.

Well, 3dmark doesn't punish the cpu as much as the gpu, whereas p95, IBT, occt, aida64 etc punish the cpu and ram, each will push a cpu to 100% load, and occt linpack will also push 90% ram at the same time, effectively pushing a cpu to 115%-120%. Honestly I can't think of any 1 test that'll 100% load cpu, gpu, ram in 1 run, probably because that'd also push the limits of more than a few psu's.
 
Upvote 0 Downvote
S

Singularity_v2

Honorable
Jun 22, 2013
11
0
10,510


Sounds good, Ill find that version on Prime95 and replace the one I have. I'm still wondering of the practicality of Prime95 and IBT. No game on teh market is going to max all 4 cores of my cpu anytime soon.

Since I got the temps a little lower I overclocked my cpu to 4.2 ghz with a +.185 VCore. Idle temps are 36 C to 44 C. I fired up Tomb Radier (Maxed Out), Battlefield 4 (Ultra Preset), and The Witcher 2 (Ubersampling On) and have seen a max temp of 79 C on the hottest core. Do you think this is a safe temp for this particular overclock?
 
Upvote 0 Downvote
Karadjgne

Karadjgne

Titan
Ambassador
Dec 26, 2012
30,741
4,559
142,790
That's kind of the general purpose of apps like p95 and IBT, to push the cpu beyond what's considered normal hard usage. If a pc can maintain stability under such duress, it won't crap out at anything less stressful. You are OC the cpu, asking for stability, then giving the pc a worst case scenario.

79° is still a little high, but tolerable. 70° is considered the safe limit for normal usage, 80° for p95 punishment.

At 4.2GHz, I'm running with vcore set at 1.19 which puts me stable at 1.14v according to cpu-z, speccy and hwmonitor. At 4.2 realistically you should be running at stock voltages or slightly less. Intel bumps up stock voltages kinda high because even they are subject to the cpu lottery so to guarantee stability on every chip, vcore is set high.

Just dropping from 4.3 to 4.2 moved me from 1.208 to 1.14 which resulted in temps under p95 small fft dropping from 70° to 63°.

Before you go jumping into games, I'd spend some time dropping your voltages, running stability tests. Take it in baby steps. Do this until you hit instability, then bump it up a notch. Just dropping these voltages 0.2v could drop your temps 10° or more.
 
Upvote 0 Downvote
You must log in or register to reply here.

TRENDING THREADS

Latest posts

Moderators online

Share this page

COMPANY
RESOURCES
FOLLOW
Top Bottom