Atom D510 Detection problem 0.99.7

[Hominidae]

Hi folks,

I am running a Supermicro X7SPA Board with an ATOM D510 CPU.

CoreTemp 0.99.7 (Win7_x64) reports Temps with
an offset of approx -145 C !!
compared to the IPMI-sensor reports of the MoBo.

The CPU model is also not reported...and VID is not matching..

[The Coolest]

Could you please post a Register dump (Tools menu)?

[Hominidae]

Could you please post a Register dump (Tools menu)?

..see attached file...

[The Coolest]

Recognition will be fixed in the next release.
VID is not the same as vcore, so that part is actually working properly.
But from Intel documentation, it seems that this CPU does not have a functional DTS.
Try to apply a -128 offset (in the Options --> Adjust offset), this should give you 'normal' readings, and see if the temperature changes with load on the processor.

[Hominidae]

Recognition will be fixed in the next release.
VID is not the same as vcore, so that part is actually working properly.
But from Intel documentation, it seems that this CPU does not have a functional DTS.
Try to apply a -128 offset (in the Options --> Adjust offset), this should give you 'normal' readings, and see if the temperature changes with load on the processor.

Thanks for the fast response...

...applied the suggested offset...Coretemp showed temps with approx 15 C offset to MoBo IMPI sensors.
...started transcoding a video with all 4 CPU-Threads on 100%....Temps rising approx 10 C
...Then CoreTemp dissapeared from the Tasklist without notice
...Restart CoreTemp...it reported negative Temps...removed the offset an Temps were back to 15C offset to IPMI interface
...after approx 1 min, temps are back with HIGH offset reporting 153 C now.

...guess we´ll have to wait for the next release

[The Coolest]

That is very strange. My assumption is that some other program may be in the way and causes Core Temp to "think" that the
temperature being read from the CPU is invalid. When that happens Core Temp adds a 128c offset due to the idea that invalid
temperature readings most often occur when the processor goes over TjMax and the digital sensor wraps and starts counting
back down from 127.

I recommend you try to turn off other monitoring utilities and see how Core Temp behaves when it's the only thing that's running.

[grave_digga]

I have nearly the same problem but only with Core#0.




Looks like this is a general problem with the older Atom processors. Here is also the dump file, maybe you fix that in the next release. Thanx for this great tool!

[The Coolest]

I've posted a Core Temp FAQ, you can see an explanation about this issue and strange temps in section #8.
http://www.alcpu.com/forums/viewtopic.php?f=63&t=892