Bad temperatures or expected results? H100 + 2500K

I wanted my 2500K to have lower temps, I'm kinda shocked with the 34-35ºC average idle temp. I have installed and reinstalled the H100 twice, using MX-2 TIM, just to make sure everything was right.

Everything is set with stock settings, with Turbo Boost enabled.

DDR3 @ 1333 MHz. Setting RAM to 1600 MHz makes the CPU temps go up to 43-45ºC range and H100 fans become noisy 24/7.

Case: HAF-X with stock fans (1x 230mm front, 1x 200mm side, 1x 200mm top, 1x 140mm back).

H100 set at Performance profile.

Ambient temperature: 25~26ºC

Bottom line: at load (LinX) the max readings are 54, 54, 58, 58ºC

I have attached screenshots that show each step during the tests.

1. Turned on PC and idling for 10 minutes

2. Quick LinX test of 2 minutes

3. Running LinX for 13 minutes

4. 10 minutes later (idling)

5. Summary of results

Don't be fooled by those lowest 26ºC readings reported by RealTemp. It never stays on the 20's range, it's always in the 30's when idling.

Any advice or is that's what to be expected?

PS.: I have the feeling that with this Gigabyte motherboard the system runs hotter than with my previous motherboard (ASUS P7P55D Deluxe).

your idle temp of 34-35c is still one degree below body temperature.

putting aside software which can be inaccurate, what are the temps in the BIOS after 10 min? thats the closest you will get to actual temp without a probe.

your idle temp of 34-35c is still one degree below body temperature.

putting aside software which can be inaccurate, what are the temps in the BIOS after 10 min? thats the closest you will get to actual temp without a probe.

There are 2 different readings in the BIOS. After 10 minutes:

One screen shows the temps of the 4 cores: 43ºC, 37ºC, 42ºC, 43ºC

The other status screen shows CPU voltage (1.176V~1.212V) and DDR3 voltage (1.344V) and another CPU Temperature value: 37~38ºC

PS.: H100 becomes a bit louder after 3-4 minutes

I'm confused about a few of your conclusions, first of all the "average idle temp", what reading or result is the average idle temp?

Your ambient temps, ~25C, are the basis for all other temps, average CPU temps will always be above that. A CPU idle temp of 5C - 10C above ambient is great. That temp is also dependent upon the CPU power saving features, SpeedStep, C1E, C-states, and memory voltage, among others.

Looking at the Summary results, I see minimum CPU and core temps of 25C - 27C, at or just above ambient. Your max CPU and core temps of a bit below 60C are great. The average of those two temps, shown in the far right column, is not the average idle temp, and a rather worthless figure, IMO. The 100% CPU usage condition that only happens during benchmark testing is a worst case CPU use situation. Actually, if a persons CPU usage caused it to vary between idle and 100%, each happening half of the time, an average temp of the mid 30's C is very good.

I use a H60 with a i7-2600k, and for normal use is set the turbo multiplier to 42. I also have SpeedStep and C1E enabled, so the core speeds vary from 1600MHz to 4200MHz. Even at idle, one or more cores will jump from 1600 to 4200 when something happens in the OS, etc. The core temps will follow that, of course. So a very stable and low idle temp is difficult to achieve in that configuration. Yours is similar, if at 3600MHz max.

CPU temperature reporting in Sandy Bridge CPUs is different than the previous generation Core i CPUs, like the one you used with your P55 board. It is much faster in Sandy Bridge CPUs, and the core temps are much closer to the overall CPU package temp. My X58/i7-900 series CPU board is much more sluggish in core temp changes, and the cores are reported as at least 10C higher than the CPU package. The boards BIOS/UEFI also determines how the CPU and core temps are reported, so comparing different manufactures boards and different CPU/chipset platforms is not valid.

I'm confused about a few of your conclusions, first of all the "average idle temp", what reading or result is the average idle temp?

When the PC was idle for the first 10 min I had a look at the stats file created by AIDA64 and the average among the cores (#1 to #4) was 34-35-36's...

Your ambient temps, ~25C, are the basis for all other temps, average CPU temps will always be above that. A CPU idle temp of 5C - 10C above ambient is great. That temp is also dependent upon the CPU power saving features, SpeedStep, C1E, C-states, and memory voltage, among others.

 

Looking at the Summary results, I see minimum CPU and core temps of 25C - 27C, at or just above ambient. Your max CPU and core temps of a bit below 60C are great. The average of those two temps, shown in the far right column, is not the average idle temp, and a rather worthless figure, IMO. The 100% CPU usage condition that only happens during benchmark testing is a worst case CPU use situation. Actually, if a persons CPU usage caused it to vary between idle and 100%, each happening half of the time, an average temp of the mid 30's C is very good.

I know, it it's not an average idle temp, it's just an average considering the first 10 min idle period, the 2+13 minutes running LinX and then other 10 min idle period.

I use a H60 with a i7-2600k, and for normal use is set the turbo multiplier to 42. I also have SpeedStep and C1E enabled, so the core speeds vary from 1600MHz to 4200MHz. Even at idle, one or more cores will jump from 1600 to 4200 when something happens in the OS, etc. The core temps will follow that, of course. So a very stable and low idle temp is difficult to achieve in that configuration. Yours is similar, if at 3600MHz max.

 

CPU temperature reporting in Sandy Bridge CPUs is different than the previous generation Core i CPUs, like the one you used with your P55 board. It is much faster in Sandy Bridge CPUs, and the core temps are much closer to the overall CPU package temp. My X58/i7-900 series CPU board is much more sluggish in core temp changes, and the cores are reported as at least 10C higher than the CPU package. The boards BIOS/UEFI also determines how the CPU and core temps are reported, so comparing different manufactures boards and different CPU/chipset platforms is not valid.

I'll try some overclocking, in fact I already tried 4.2GHz, with default Vcore (1.220V detected in BIOS) and max core temps were around 64-65ºC.

I didn't know about those differences on SB CPUs and the previous generation.

The Gigabyte software picks the reading named "CPU" to report the CPU temperature, and they are always the lowest reading. I always considered the core temps as a better parameter. Does that still stand with SB CPUs?

http://img155.imageshack.us/img155/8504/3254dt565rrd.png

After all I think I was worried for nothing, everything seems to be working just fine. :):