GA-EP45-UD3LR & E7400 temps

[mm67]

I am trying to find settings to run E7400 at 4 GHz, so far looking good except one thing.


Core temps look usable but what is the Cpu temp on Everest ? I mean what is it actually measuring. I'm running Prime95 blend test and I noticed that when Prime95 is running small FFT's core temps go up and that Cpu temp goes down, all temps are close to each at that time. But when Prime95 is running large FFT's core temps go down and that Cpu temp goes up, a lot. Worst I saw was core temps at 57 and Cpu temp at 79. Should I be worried about that sensor's readings ?

[Chike]

Re: GA-EP45-UD3LR & E7400 temps

I temember such anomality with other board and processor, do you have the latest BIOS installed?
I have the same board waiting to be paird with E7500 probably in a week or two I'll see then, but I think the ASUS P35 showed the same.
Wouldn't worry about it too much, obviousely CPU temp cannot be higher than it's cores.

[mm67]

Re: GA-EP45-UD3LR & E7400 temps

I've been trying to duplicate this behavior on my GA-EP45-UD3 board but it just doesn't do the same. One thing that is different with these boards is that on UD3 board I replaced standard NB heatsink with this one: Thermalright HR-05 SLI/IFX and a 80 mm fan. And if I understand large FFT test correctly then that is the one that also stresses MCH.

[profJim]

Re: GA-EP45-UD3LR & E7400 temps

Double check your temperatures by running ET6, HWmonitor or SpeedFan while Everest is running..
Cpu temps are "always" lower than core temps by at least 5 (C) at idle.
On my systems this temperature difference is much greater when stress testings, especially with IBT, LinX, and OCCT PSU test. It sounds like Everest is not properly reporting the correct cpu temperature.

In the bios PC Health Status panel, you can set the CPU Warning Temperature to 60, 70 degrees or higher. Your system needs to have a motherboard speaker hooked up to hear this warning. My system will beep during IBT or LinX tests if this setting is set to 60 (C). This always happens when RealTemp or Coretemp reports cpu core temperatures at this setting or higher.

You can download the datasheet and thermal design specifications for your processor from the Intel website for more information.

[mm67]

Re: GA-EP45-UD3LR & E7400 temps

I already did that, HwMonitor and ET6 show same temp as Everest and Cpu warning temperature is set to 80, I already heard that warning beep couple of times. That Cpu temp actually shows higher reading than core temps on both my Gigabyte systems when temps get high, like over 60. On low temps that reading is always lower than core temps.

This is my UD3 running large FFT's

[profJim]

Re: GA-EP45-UD3LR & E7400 temps

The Everest core temps are probably identical, is the Everest cpu temp the same as the other program's cpu temp readings?
With a cpu temperature warning setting = 60, you should find that the warning beep happens when cpu core temperatures exceed the warning threshold,

[mm67]

Re: GA-EP45-UD3LR & E7400 temps

Yes, Everest, HwMonitor and ET6 all show same reading for Cpu temp. Bios also sees the same temperature as I have already heard that warning beep couple of times. Warning is set to 80 and Cpu temperature reading is close to that all the time when running large FFT's. At same time core temps however are about 60 degrees.

Decided to lower OC a little bit and see what happens, same thing. In this screenshot there is temp readings of ET6, Everest, HWMonitor and HWInfo32, all show Cpu temp as being lot heigher than core temps while running large FFT's.

[Lsdmeasap]

Re: GA-EP45-UD3LR & E7400 temps

What does the BIOS PC health section say for CPU temp? If it is the same as these programs then they are likely correct (As in being detected by the programs correctly), and your board may just have a faulty sensor or the BIOS version you are using is messed up

No matter what the issue is though the temp has to be incorrect otherwise your Core temps would be much higher too if the CPU was really that hot.

[mm67]

Re: GA-EP45-UD3LR & E7400 temps

I really have no way to see bios Health section while cpu is under load but on idle it is showing readings around 30 degrees, same as all programs on Windows. And since I have set bios high temp warning beep to 80 and I hear the warning beep when programs show Cpu temp sensor being near 80 I'd say that all programs detect temp correctly. I agree that cpu temp can't be really that much higher than core temps, but the strange thing is that this only happens when running Prime95 large FFT's. If the sensor was faulty wouldn't it always behave the same way.

[Lsdmeasap]

Re: GA-EP45-UD3LR & E7400 temps

Ahh, sorry I thought it showed that at idle too. Ya I guess there isn't anyway to check that.

You could try older/newer BIOSes and see if they all exhibit the same reading, if so then it is probably just a faulty sensor

Well large FFt's are going to get your CPU hotter than other tests, so that would be why maybe it is the only one that brings the faulty reading that high.

I bet you see the same if you use Intel Burn test or LinX too, but core temps would remain way under that amount telling you for sure it is not a correct reading

[mm67]

Re: GA-EP45-UD3LR & E7400 temps

Both small FFT's and Linx look similar, Cpu temp is close to core temps, maybe 2 - 3 degrees higher than core temps.


But when running large FFT's Cpu temp reading is 10 - 15 degrees higher than core temps. I really can't understand how this is possible.

[Lsdmeasap]

Re: GA-EP45-UD3LR & E7400 temps

Ya, that doesn't sound right, but just appears to be a faulty sensor or reading to me and your large FFT's test seems to show that it has to be

[mm67]

Re: GA-EP45-UD3LR & E7400 temps

Running with lower FSB and higher multiplier seems to be helping with temps, this is starting to look like usable 4 GHz settings for E7400.

[Lsdmeasap]

Re: GA-EP45-UD3LR & E7400 temps

Ya those temps are looking good now!

[mm67]

Re: GA-EP45-UD3LR & E7400 temps

Settings were starting to look real good, passed 20 runs of Linx, passed 12+ h of Prime 95 small FFT's and 12+ h of large FFT's. Then last night Windows had crashed with error code 0x00000124 after running 11,5 h of Prime95 blend test. Guess there is some more tweaking to do.