Discrepency with Speccy on Windows 10 and Windows XP

[DaveH]

Hi, I've been using Core Temp for ages to monitor my CPU temperatures in my system tray, but since I installed Piriform's Speccy system monitor, I've been aware of a discrepancy between them on the temperature readings.
Now I'm well aware that Speccy is notorious for giving hugely inaccurate temperature readings with some processors, in fact there is a huge thread on their forum dedicated to this!
However, I've now become aware that the discrepancies I'm seeing on my dual processor dual boot system are not consistent.

On Windows 10 in fact Speccy and Core temp pretty much agree, and they both agree that Processor 0 runs about 10°C hotter than Processor 1.
This is mirrored on Windows XP on Speccy, but not on Core Temp, where the processors appear to be running at pretty much the same temperature.

The processors are X5460 Quad Core Xeons, and the motherboard is a Supermicro X7DAL-E+ server board.
Anyone any idea why the Core Temp readings would be different on the two operating systems?
The versions of Speccy and Core temp are the same, although on Windows 10 they are 64 bit and on Windows XP they are 32 bit.

This is what I'm seeing on Windows XP -

WinXPImage.jpg

And this is what I'm seeing on Windows 10 -

Win10Image.jpg

Both grabs were taken with the machine just idling, but of course it was impossible to guarantee exactly the same conditions on both operating systems.
Of course I have no idea which readings to trust!
I suspect that it's the Core Temp readings on Windows XP which are suspect, which is why this post is here and not in the Speccy forum!
Any help and advice gratefully received.
Cheers, Dave.

[DaveH]

Anyone?

[DaveH]

I guess not.........

Could using an older version of Core Temp be better on XP?
Although the latest version still works on XP, compatibility may not be as good as with later versions of Windows, and it does seem to be that installation of Core Temp that is suspect.

[The Coolest]

It looks like WinXP sorts the CPUs differently. You can see that CPU#0 on XP is CPU#1 on Win10 and CPU#1 on XP is CPU#0 on Win10.
And Core Temp follows the affinity index of CPUs.

[DaveH]

Thanks for the reply @The Coolest!

I have never been aware that the processors are being detected a different way around on the two operating systems!
I'll have to look into that, I always assumed the processor IDs came from the BIOS, so shouldn't be different.

That's not really the problem though, it's the fact that according to Core Temp on Windows 10 one processor seems to be running about 10° hotter than the other, whereas on Windows XP they appear pretty much the same. Speccy says there is about a 10° difference on both operating systems, certainly on the average core temperatures, and the same way around, with processor #0 hotter than processor #1.

I did try an older version of Core Temp on XP BTW, which produced the same result.

[The Coolest]

Can you post a register dump from both Win10 and WinXP?
After taking another look, it seems that on XP the order of cores could be wrong.
I don't have a multi-CPU machine to test this with, so it's possible that it's a long standing bug.

[DaveH]

Well that would certainly go a long way towards explaining the discrepancy!
Here are the two files, I hope they give some clues as to what's happening.
Thanks, Dave.

CoreTemp-Dump-Win10.txt

CoreTemp-Dump-WinXP.txt

[DaveH]

@The Coolest
Did you find anything?

[The Coolest]

Yes, I see that the core/package order is messed up on XP. Not sure why, though.

[DaveH]

Thanks, that could explain a lot of things!
Is this only because it's a dual processor machine?
Surely it's not wrong on all XP installations!

[DaveH]

@The Coolest
Just to say that after a lot of experimentation and head scratching, I'm now using HWiNFO to monitor my processors.
I still have Core Temp installed, I'm just not using it now in the system tray.
While not as light on resources as Core Temp, HWiNFO does show the same on Windows XP and Windows 10, and it agrees on both systems that one processor is running about 10° hotter than the other, which agrees with Speccy.
Why this is I do not know, I physically swapped the processors over and the same one was still reading hotter by the same amount, so I guess that's just how it is.
Thanks for the help.
Cheers, Dave.