technical details Tj(max)

[spindot]

Hello.

First of all, thanks for the nice tool and your efforts.

I'm wondering how the maximum junction temperature Tj(max) will be determined by Core Temp V0.95.4 beta. And especially what is the difference to V0.95 related to the E6x50 Core 2 Duo CPU's (G0 stepping).

The beta determines 100 °C for a E6750, the stable version determines 85 °C. The 100 °C seems to be more reliable to me.

My first assumption to determine Tj(max) was reading bit 30 of the undocumented MSR 0xee (as descripted in c't computer magazin 11/07, page 218). If the bit is set, Tj(max) should be 85 °C, otherwise 100 °C. But it turned out that this bit is set for a E6750.

Further on I stumbled over an interesting thread in an Intel forum:
http://softwarecommunity.intel.com/isn/ ... 29027.aspx

Especially the last but one post in this thread unveils that the bit 30 in MSR 0xee is only valid for some mobile processors. But unfortunately I think the other mentioned details aren't enough to determine Tj(max) by reading one or more MSR independent from processor type. But it is also possible that I have not understood all details.

So the third possibility for the determination of Tj(max) may be a lookup table with the processor ID (if this is sufficient?) as the key. But where is this relationship documented? Do you need to have a kind of data sheet for each CPU type? In my opinion this would have the great disadvantage that this table has to be extended for each new CPU type.

Due to the difference in Tj(max) determination for the E6750 from Version 0.95 to 0.95.4 beta I would assume that Core Temp uses a method similar to the lookup table method. But maybe I'm wrong here and you can point me to a description how Tj(max) can be determined independent from the (Intel) CPU type by just reading some MSR's or something similar.

Thanks in advance.