Sandy bridge -- temperatures high and changing quickly?

Cooling Processors quietly

Moderators: NeilBlanchard, Ralf Hutter, sthayashi, Lawrence Lee

Post Reply
NSParadox
Posts: 24
Joined: Thu Jun 26, 2003 2:25 pm
Location: San Mateo, CA
Contact:

Sandy bridge -- temperatures high and changing quickly?

Post by NSParadox » Mon Jan 17, 2011 11:05 am

I recently built a new system. Relevant parts:
Sandy Bridge i5-2500k
Scythe Mugen 2 rev B with stock Scythe 120mm fan
Asus P8P67M Pro motherboard (as far as I can tell very similar to the rest of the P67 family)
Gigabyte 460 GTX 1GB
Sonata Proto case with Antec NeoPower 650W power supply (80+) and stock Antec 120mm two-speed case fan set to low (not sure if it's 7V or 5V)

Subjectively, the system is pretty quiet (not quite silent).

According to Asus' utilities that are bundled with the motherboard, here are the temperatures I'm seeing:

Idle
CPU: 32 C (Real Temp 3.6 says 40-42 C for Cores 1-4)
Motherboard: 30 C
CPU fan ~200-300 RPM - I don't believe it's audible
CPU info from CPU-Z: 1.648 GHz, 103 MHz BCLK, 1.0V core voltage

Load with Prime95, 4 threads for 15 minutes
CPU: 69 C (Real Temp 3.6 says 72-81 C for Cores 1-4)
Motherboard: 34 C
CPU fan ~1400 RPM - System less quiet but not at all loud. Loudest sound may be coming from power supply as opposed to CPU fan but I haven't taken the case off to see if I could isolate the sound.
CPU info from CPU-Z: 4.327 GHz, 103 MHz BCLK, 1.312 core voltage

The odd thing to me is how quickly the temperatures ramp up/down. When I turn off prime95, my CPU temperatures fall back down to 34 C in a matter of seconds. When I turn it back on, it's back up to 60+ C in less than 5 seconds. I'm wondering if this indicates there's a problem with my CPU sensing, or maybe with the thermal paste job? This is the first time I've mounted a heatsink anywhere near the size of the Mugen and the heatsink definitely twisted/shifted a bit while I was trying to bolt it on upside-down. I applied a rice-size grain and fairly evenly distributed the thermal paste with an old credit card before installation.

So, questions:
- Should I be concerned about the temperature? How does this compare to other people's Sandy Bridge systems using fairly quiet cooling?
- What would explain the sudden ramp-up/ramp-down for temperatures. Could I have a heatsink or thermal pasting problem?

PS - I've overclocked the CPU from stock 3.3 ghz to 4.3 ghz, although I saw very similar results at clock (very fast temperature ramp-up and +/- 5C idle/load temperature). So far, the system is stable. I can run Prime95 for 10 hours straight with the overclock and there are no errors.

Thanks for your help!

quest_for_silence
Posts: 5275
Joined: Wed Jun 13, 2007 10:12 am
Location: ITALY

Re: Sandy bridge -- temperatures high and changing quickly?

Post by quest_for_silence » Mon Jan 17, 2011 11:58 am

Briefly, it should be quite correct (even on Clarkdales you'd have similar - even if not identical - behaviour).
You should just better tune your fan control in order to have a quieter/smoother profile than actual one.

karkee
Posts: 207
Joined: Wed Apr 09, 2008 1:15 pm
Location: belgium

Re: Sandy bridge -- temperatures high and changing quickly?

Post by karkee » Tue Jan 18, 2011 1:04 pm

I have a this at the moment:

I5 2500K
Asus P8P67 Pro
8gb kingston 1333 (used from old setup)
Prolima Megahalems
Antec solo

I did some tests and at stock clocks im idle around 31/32 for all cores, load im getting temps 47/48° on all cores, sometimes core 2 tends to run a bit hotter.

matchu
Posts: 70
Joined: Wed Mar 30, 2011 4:16 pm

Re: Sandy bridge -- temperatures high and changing quickly?

Post by matchu » Sat May 14, 2011 8:39 pm

Hello

I have a similar build with a similar problem

System
i5 2500k @ 4.4GHz
MSI P67A-G43 mobo
Scythe Mugen 2 Rev. B w/ stock Scythe cooler
Antec P183 V3 with a single Slip Stream 120mm / 1200rpm exhaust in the back

Tweaking
Both scythe fans are set to 700rpm at idle and to ramp up to 1100rpm to keep temps below 70.
The core voltage is set to 1.25V in the BIOS, but is usu slightly lower during use
again, clocked to 4.4GHz

Temperatures
Ambient = 22C
Idle = 40C (700rpm)
Prime95 = 60C (1100rpm)
Prime95 (previously at 1.35V) = 71C (fans maxed 1500rpm)

My behaviour is similar to yours, in terms of temperatures changing very quickly and going very high. I also wonder if my numbers are within "normal limits". But if I had to guess, the high voltage and high clock speed are probably responsible for these temperatures.

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Re: Sandy bridge -- temperatures high and changing quickly?

Post by ces » Sat May 14, 2011 9:35 pm

These temps don't seem to be out of line with SPCR test results, especially given that they are overclocked. See SPRC test results for non-overclocked 2500K:
http://www.silentpcreview.com/article1156-page3.html
(don't forget to add your ambient temps to these numbers... if you forget to do that it makes your numbers look worse than they are)

See also these temps on a non-overclocked i7-965 Extreme Nehalem core, LGA1366, 3.2GHz, with some of the best air cooled cpu heat sinks in existence:
http://www.silentpcreview.com/article1198-page4.html

If you think the temps are ramping up too high too quickly, try re-seating the cpu cooler.
Last edited by ces on Sat May 14, 2011 9:41 pm, edited 2 times in total.

m0002a
Posts: 2831
Joined: Wed Feb 04, 2004 2:12 am
Location: USA

Re: Sandy bridge -- temperatures high and changing quickly?

Post by m0002a » Sat May 14, 2011 9:39 pm

matchu wrote:I have a similar build with a similar problem.
I don't believe that others have a problem, since it is perfectly normal for temps to ramp up and then down quickly depending on load. However, your idle temps (40 C) are about 8 C higher than most others (including mine) for 2500K, so you may have a heatsink mounting problem, or it may be an issue (or difference) in the MSI bios sensors.

m0002a
Posts: 2831
Joined: Wed Feb 04, 2004 2:12 am
Location: USA

Re: Sandy bridge -- temperatures high and changing quickly?

Post by m0002a » Sat May 14, 2011 10:04 pm

Here are the temps of my INtel i5 2500K with Asus Sabertooth P67 mb at idle. No overclock, CPU Fan 900 RPM (PWM):

Ambient 23.5C

Real Temp 3.67
Core 1 - 31C
Core 2 - 28C
Core 3 - 25C
Core 4 - 34C

Asus AI Suite II
CPU 33C

CA_Steve
Moderator
Posts: 7650
Joined: Thu Oct 06, 2005 4:36 am
Location: St. Louis, MO

Re: Sandy bridge -- temperatures high and changing quickly?

Post by CA_Steve » Sun May 15, 2011 7:35 am

Having a fast thermal response from load to idle is a GOOD thing. It means your cooler is making great contact with the CPU and the cooler/fan are capable of quickly dissapating heat. Modern CPU's shed and add load power in nanoseconds (3GHz CPU, eh?), while you view the monitor in tenths of seconds - things are happening much faster than you can view them.

The Real Temp numbers look high at both idle and load. Did you calibrate the s/w? If not, do so and try again.

* edits for clarification in italics.
Last edited by CA_Steve on Sun May 15, 2011 9:22 am, edited 1 time in total.

ces
Posts: 3395
Joined: Thu Feb 04, 2010 6:06 pm
Location: US

Re: Sandy bridge -- temperatures high and changing quickly?

Post by ces » Sun May 15, 2011 8:26 am

CA_Steve wrote:Having a fast thermal response is a GOOD thing. It means your cooler is making great contact with the CPU. Modern CPU's shed and add load power in nanoseconds
While the CPU doesn't have a lot of thermal momentum, the CPU heat sink will have more. So wouldn't you expect that the better the thermal connection between the CPU and the heat sink the less abrupt will be the thermal swings of the CPU?

CA_Steve
Moderator
Posts: 7650
Joined: Thu Oct 06, 2005 4:36 am
Location: St. Louis, MO

Re: Sandy bridge -- temperatures high and changing quickly?

Post by CA_Steve » Sun May 15, 2011 10:16 am

I edited my OP for clarification.
ces wrote: While the CPU doesn't have a lot of thermal momentum, the CPU heat sink will have more. So wouldn't you expect that the better the thermal connection between the CPU and the heat sink the less abrupt will be the thermal swings of the CPU?
Here's a bit of a ramble.

I expect a decent cooler to provide an abrupt swing from load to idle. I think in most normal applications with a decent cooler, the temperature transition from idle to load will be seen to happen in just a second or two. And load to idle will get to within 5C of baseline idle temp in about the same time.

Getting the heat out of the CPU depends on:
- making good thermal contact between CPU and cooler
- The cooler's ability to do it's job
- the case's airflow/heat retention.
- ambient environment.

Let's say, your system provides decent performance for the above.

A CPU with:
- stock cooler might run at 75C load.
- a decent air cooler might run at 60-65C load.
- water cooled ~45C


Will there be a big difference between the stock and decent cooler in the temp curve going from idle to load? I think they will be similar, just that the stock cooler temp ends up higher. In a water cooled system, the overwhelming capability of the water block might slow the transition a bit, but I've not played with one.

So, here's a couple of common issues and common reasons why they occur.

1) higher than expected CPU idle temperature.
a) bad expectations by the user.
b) possibly bad contact between cooler and cpu.
c) did you bother to calibrate your temp monitoring s/w?
d) high temps elsewhere, so it's not a high delta from ambient
-ambient room temp is high
- internal case temp is high
2) it takes a really long time for my CPU temps to drop from load to idle values. (many many seconds to get within 5-10C of initial value)
- poor connection between cpu and cooler
- poor cooling capability of heatsink/fan
3) higher than expected load temp.
a) bad expectations by the user.
b) possibly bad contact between cooler and cpu.
c) did you bother to calibrate your temp monitoring s/w?
d) high temps elsewhere
- ambient room temp is high
- internal case temp is high

Summing up, just do 4 things.
1) calibrate your temp s/w.
2) See if your CPU idles within 5-8C or so of ambient case temp.
3) See if your load to idle temp curve drops to within, say 5-10C of idle value in a few seconds.
4) See if your load temp delta is within:
- Intel spec
- similar to what others see with similar systems.

Deucal
Posts: 100
Joined: Sun May 02, 2010 3:10 am
Location: Iceland

Re: Sandy bridge -- temperatures high and changing quickly?

Post by Deucal » Tue May 17, 2011 6:55 am

the new Intel CPUs dynamically change their voltage from idle - load - idle. The old way of OC was that the voltage always staid at the specified one for the OC.
Look at the voltage readout.
How did you OC it? did you use the MBs inbuilt OC settings? Or did you change the voltage and multiplier your self? You might me using too much voltage for that OC, some MBs go the safe way to make the OC work and volt too much compared to what you might actually need for your specific owned CPU.

Modo
Posts: 486
Joined: Wed Apr 16, 2008 3:32 am
Location: Poland

Re: Sandy bridge -- temperatures high and changing quickly?

Post by Modo » Tue May 17, 2011 9:31 pm

This is WAD, perfectly safe. I see the same behaviour on my i5 2500. Similar things happened with my E6550 in the previous rig. Now Intel made the idle power even lower, so the swing is bigger.

Post Reply