Sign in with
Sign up | Sign in
Your question

2500k idle temps with Scythe Mugen 2

Last response: in CPUs
Share
July 3, 2011 8:37:54 PM

I see a lot of threads about how the 2500k is supposed to idle around or even below 30C, but I just don't see it happening myself.

I'm using the Fractal Design Define R3 with just the two stock fans. I have them plugged into the bundled fan controller, running them as low as possible. With that setup, the cores are idling at 35-40C and during Prime95 blend stress testing, the highest I have seen is 69C. If I turn the case fans all the way up, the CPU doesn't go higher than 65C.

Is this normal? I've seen people claim that the 2500k shouldn't go higher than the 50's during Prime95 stress testing.

I'm running all default settings in the UEFI and am not overclocking, although I plan to do that eventually.
a b à CPUs
July 3, 2011 8:58:16 PM

Temps are dependent on 4 main things (ignoring ambient temps):

1) Proper CPU/heatsink installation with good thermal compound.
2) Good air flow
3) Software (programs) running
4) Extent of overclock
July 3, 2011 9:31:04 PM

I know. I also forgot to mention that my ambient temp is 24-25C. When I say idle temp, I mean in Windows with < 1% CPU load and a 1.6GHz frequency in power saving mode.

So you don't think those temps are unusual?
Related resources
a b à CPUs
July 3, 2011 10:10:33 PM

Kralnor said:
I know. I also forgot to mention that my ambient temp is 24-25C. When I say idle temp, I mean in Windows with < 1% CPU load and a 1.6GHz frequency in power saving mode.

So you don't think those temps are unusual?



Could be better. My i7 2600K is overclocked somewhat and it's idling at about 29 C and it's above room temperature in here now (76-77 F). Your load temps are really what matter though
July 3, 2011 10:15:46 PM

bearclaw99 said:
Could be better. My i7 2600K is overclocked somewhat and it's idling at about 29 C and it's above room temperature in here now (76-77 F). Your load temps are really what matter though


Aye, of course, load temps are the important metric. I was just wondering why my idle temps were so high, it seemed strange.

For the record, what case do you use and how is your airflow setup?
a b à CPUs
July 3, 2011 10:43:53 PM

Kralnor said:
I know. I also forgot to mention that my ambient temp is 24-25C. When I say idle temp, I mean in Windows with < 1% CPU load and a 1.6GHz frequency in power saving mode.

So you don't think those temps are unusual?

I'll put it this way. They are on the high side, and I would be concerned.

Sometimes, a CPU/heatsink re-installation adhering to good practices would help.
a b à CPUs
July 3, 2011 11:12:40 PM

by any chance did you put the entire tube of thermal compound on the processor before you installed the heatsink?

or did you use the TINIEST drop like you should have ?
July 3, 2011 11:28:25 PM

Ubrales said:
I'll put it this way. They are on the high side, and I would be concerned.

Sometimes, a CPU/heatsink re-installation adhering to good practices would help.


I might try that later on if I don't find any other reasons. Installing the Mugen was a bit of a PITA, so I'll try exhausting other possibilities first.

Outlander_04 said:
by any chance did you put the entire tube of thermal compound on the processor before you installed the heatsink?

or did you use the TINIEST drop like you should have ?


I only used a tiny drop, only just enough to cover the whole surface with a very thin layer, definitely less than 1mm. I spread the compound across the surface using a plastic card.
a b à CPUs
July 3, 2011 11:46:54 PM

Kralnor said:
I might try that later on if I don't find any other reasons. Installing the Mugen was a bit of a PITA, so I'll try exhausting other possibilities first.

Good strategy! I agree, it is a PITA!

"I only used a tiny drop, only just enough to cover the whole surface with a very thin layer, definitely less than 1mm. I spread the compound across the surface using a plastic card."

A pea sized quantity spread around well is adequate. The thickness of the film should be around 0.1 mm, not "less than 1 mm".
July 4, 2011 9:10:39 PM

I think my temperature problem is related to the lack of general airflow more than anything else.

I just re-assembled my old C2D E6600 (OC'ed to 3GHz) in an Antec P180 (stole some parts for my current build) and for some reason, the top two default case fans weren't spinning and the CPU was idling just above 40C (granted, that was after a semi-heavy workload of installing a lot of Windows Updates).

After I fixed the two case fans, the core temps dropped to ambient, which was 24C. What an amazing difference from just two exhaust fans.

I think I'll look into adding another exhaust fan in the R3. Anyone got recommendations for a good low noise 120mm case fan?
a b à CPUs
July 4, 2011 9:49:30 PM

Kralnor said:
Aye, of course, load temps are the important metric. I was just wondering why my idle temps were so high, it seemed strange.

For the record, what case do you use and how is your airflow setup?



NZXT Tempest EVO case with about 6-7 case fans
July 8, 2011 3:48:56 PM

So I tried adding another exhaust fan at the top of the R3. (I used the bottom 120mm stock fan from the P180). Turns out it made zero difference to my idle temps.

I guess I'll have to try re-mounting the Mugen.
a b à CPUs
July 8, 2011 4:17:39 PM

Run all fans at full speed and check temps.
July 8, 2011 7:42:14 PM

I noticed that "Power Saving Mode" was disabled in the UEFI. After enabling it, temps dropped by a few degrees.

Running the Scythe fan at 1400RPM and both case fans at max, idle temps on the cores are 33-36C. Same ambient temp as above.

It seems to make very little difference if I run the Scythe at 700 or 1400RPM, only a few degrees in variance.

Also, according to ASRock Extreme Tuning Utility, the CPU Ratio was hovering around 35-37x, while CPU-Z claimed it was 16x. Not sure which to believe...
a b à CPUs
July 8, 2011 7:47:43 PM

Kralnor said:
.....It seems to make very little difference if I run the Scythe at 700 or 1400RPM, only a few degrees in variance.

This means that the heat transfer curve is peaking around 700 RPM for the given set of conditions, and then the curve is flattening out. Further increases in RPMs have a marginal effect on the heat transfer (and resulting temps).

An 8 to 10 degrees C over ambient is normal for idling temps.
July 9, 2011 11:46:01 AM

I still think its your case there buddy. Open up the front door and leave it open. While you can have more exhaust fans, it does no good if the inside of the case doesnt cool down with incoming air from an intake fan.

The front door barely gives the front fan enough room to pull the slightest bit of air in. So open up the front door and turn everything on high and check your temps again.
July 9, 2011 12:22:36 PM

That shaves a few more degrees off. Down to idling at 31-34C with the front door open and everything blowing at full speed.
July 10, 2011 9:44:09 PM

I finally decided to re-mount.
After removing the Mugen, I noticed that the TIM seemed to be a bit too thin on the surface of the CPU - in certain small areas near the center it actually didn't cover fully. This time, I used a drop in the center and didn't attempt to spread it myself.

However, re-mounting doesn't seem to have made much of a difference to idle temps. They might drop a bit further once the TIM has settled in, though.

Load temps seem to be very slightly better. With the front door open and all fans blowing at full speed, the cores peak at 49/51/55/51 at stock settings after an hour of Prime95 blend torture testing.
The test in my original post was performed with the front doors closed and the case fans running slower though (same speed on the CPU Scythe fan), so the difference is perhaps at most a few degrees under the same conditions.

Also, my cabling is a bit of a mess, although not downright horrible. Maybe I can shave off a few more degrees if I tidy up things.

I think I'll just call it a day and perhaps settle for a lesser overclock.
July 11, 2011 4:09:59 PM

Kralnor said:
I finally decided to re-mount.
After removing the Mugen, I noticed that the TIM seemed to be a bit too thin on the surface of the CPU - in certain small areas near the center it actually didn't cover fully. This time, I used a drop in the center and didn't attempt to spread it myself.

However, re-mounting doesn't seem to have made much of a difference to idle temps. They might drop a bit further once the TIM has settled in, though.

Load temps seem to be very slightly better. With the front door open and all fans blowing at full speed, the cores peak at 49/51/55/51 at stock settings after an hour of Prime95 blend torture testing.
The test in my original post was performed with the front doors closed and the case fans running slower though (same speed on the CPU Scythe fan), so the difference is perhaps at most a few degrees under the same conditions.

Also, my cabling is a bit of a mess, although not downright horrible. Maybe I can shave off a few more degrees if I tidy up things.

I think I'll just call it a day and perhaps settle for a lesser overclock.


You must understand that the case you are using is not meant for an overclocker/gamer type of rig. I was in the same boat as you, the case is priced very well, has alot of good features and its really freaking awesome looking cause its so minimalistic. However, that case has like 3 fans or so, one of which is severly blocked off due to the closed door. So it basically has two and a half fans. That is nowhere near enough ventilation for a gaming/O.C. friendly case. Hell, look at the antec gaming series or HAF series. They have like 5 fans a piece stock with room for more lol.

This case was built to be quiet, which means reduced airflow and increased temperatures. It is not a problem with your cpu or cpu cooler or thermal paste or whatever. Hell, its not even a problem with the case, since it was designed to be this way. It ends up being a problem with the fact that you might have chosen the wrong case for what you wanted to get out of your rig.

Even so, idling around 31-34 is not bad at all, especially for this case.
July 11, 2011 4:57:29 PM

Actually, the case only comes with two fans - one at the front for intake (which is completely covered by the front door) and one at the upper back for exhaust. I of course understand that this is very far from the high-end cooling offered by for instance the HAF series and thus temperatures won't be as good.

With that being said, I haven't chosen the wrong case. A quiet rig means more to me than raw performance and enhanced overclocking at the expense of increased noise.

Still, I have seen numerous reports of people hitting 4.5GHz with very similar setups. Even if I can't do that (not all CPU's can, no matter the cooling), I'd be happy with just anything above 4.0.
a b à CPUs
July 11, 2011 4:58:21 PM

Just for information, my i7-920 OCed to 3.82 GHz, in a HAF 932 idles at around 31 degrees C at 23 degrees C ambient. I consider this good.
July 11, 2011 5:22:41 PM

Kralnor said:
Actually, the case only comes with two fans - one at the front for intake (which is completely covered by the front door) and one at the upper back for exhaust. I of course understand that this is very far from the high-end cooling offered by for instance the HAF series and thus temperatures won't be as good.

With that being said, I haven't chosen the wrong case. A quiet rig means more to me than raw performance and enhanced overclocking at the expense of increased noise.

Still, I have seen numerous reports of people hitting 4.5GHz with very similar setups. Even if I can't do that (not all CPU's can, no matter the cooling), I'd be happy with just anything above 4.0.


Sorry, I didn't mean for it to sound that way. It's not that you chose the wrong case, I am sure that you knew what you were getting in a silent case like this. However, I also don't think its realistic for you to think that you would be able to have a processor that is know for being a ridiculous overclocker, be much cooler than what you have it now in a case that is not known for keeping stuff cool at all. While the silent factor may be more important to you than the cooling, you can't expect to get a 4.0ghz + (which is a very high overclock btw) with that case. Add to the fact that MAYBE you got one of the unlucky cpu's that wont oc more than that, and I think you should stop banging your head against the wall as to what may be wrong with your cpu/cooler and know that those people that got up to 4.5ghz probably had a case/motherboard/cooling solution that was able to make that happen.
July 11, 2011 5:50:17 PM

phishy714 said:
Add to the fact that MAYBE you got one of the unlucky cpu's that wont oc more than that, and I think you should stop banging your head against the wall as to what may be wrong with your cpu/cooler and know that those people that got up to 4.5ghz probably had a case/motherboard/cooling solution that was able to make that happen.


I am not comparing my setup to vastly different ones that have hit 4.5GHz. I have noticed people with the same case, fans and HSF hitting that very clock speed with acceptable temperatures under Prime95 load (less than 75C).

Of course, I understand that hardly any two CPU's are created equal and it might just be that this one won't even hit 4.0GHz (which I don't think is considered a very high overclock, but let's agree to disagree about that).

I think I'll let the thread rest here until I get around to attempting the overclock.
July 11, 2011 6:09:25 PM

Good luck either way! Hope you get the results you are looking for. I desperately want a i5-2500k so if you don't like yours, feel free to sell it to me :) 

lol. Let us know how it goes for ya
July 13, 2011 5:05:51 PM

At first I tried 4.0GHz with all voltages on auto, but that failed very quickly in Prime95 blend - didn't last longer than a few minutes before giving a 0x0124 BSOD, which seems to be an indication of not feeding enough Vcore - it was getting 1.15v at its peak.

Tried 4.0GHz with a fixed Vcore of 1.2v and level 1 LLC (which is supposedly the most stable setting for ASRock boards). Passes 5 runs of IBT and the hottest core reaches 72C at the motherboard's least aggressive fan profiles, so it doesn't look that bad.

Will need to run overnight of course as that is nowhere near enough testing ;) 

phishy714 said:
I desperately want a i5-2500k so if you don't like yours, feel free to sell it to me :) 


Haha, nah, sorry. Even if I have to settle for 4.0, I think I'll be pretty happy with the upgrade, even if my trusty C2D E6600 was still going strong.

EDIT: So using more aggressive fan control for the Mugen didn't help all that much. It's stable after 9h15mins of Prime95 blend, but the hottest core reaches 70C. I contacted another 2500k / R3 user with the -exact- same cooling setup, and he keeps it under 60C at 4.5GHz, so that really makes me wonder...

EDIT2: Dropping LLC down a level (from 1 to 2) shaved 2C off the hottest core. Ran another 8h of Prime95, still looking good.

EDIT3: Tried a Vcore of 1.25, then 1.275, and then 1.3, but even that wasn't enough to boot into windows at 4.5GHz. I think the heat generation will simply be way too high from having to up the Vcore so much.
July 15, 2011 12:27:38 PM

Right, so instead of continously editing my previous post, I figured I would make a new one ;) 

Still couldn't get into Windows at a Vcore of 1.325v and 45x multiplier so I upped to 1.35v. That made me able to run Prime95 very briefly, but I quickly stopped as the hottest core reached 75C in less than a minute - that is my personal limit as I want to stay on the safe side.

4.0GHz seems feasible enough, the temps are acceptable and I might be able to drop Vcore a few more notches to lower temps further. I'll see how much further than that I can go.

EDIT: 4.3GHz seems to be out of the picture too. With a Vcore of 1.25v the hottest core quickly reaches 75C and I get a 0x0124 BSOD during IntelBurnTest in less than a minute.

EDIT2: Seems like I can't go higher than 4.0GHz without hitting too high temperatures. At 40x multiplier, Vcore of 1.175v and LLC level 1, it Prime95 blend passed for 1h and it did 5 runs of IntelBurnTest. The hottest core peaked at 73C. It also seemed okay with Vcore of 1.170v, but 1.160v gave BSOD during stress testing. Can't lower LLC to anything less than level 1 (highest setting on this board) without BSOD'ing either.
!