Core 0/Core 1, big temp difference.

simonnance · 21 Mar 2007 at 10:38

OK, ive undervolted my E6400 nicely..... and Core 0 is at a nice 25C, but Core 1 is hitting 31C!

Temps with TAT and Coretemp, no proccesses taking up the second core (i normaly see that type of temp diff when running Folding on one core). Using a well seated Ultra-120 on a lapped IHS.

6C diff is 20% more.... that's a LOT. Not dangerous, but very strange, and only really noticable on low volts at idle (TAT load is 31/34, so still a dif, but much less of one).

Any explanation?

pastymuncher · 21 Mar 2007 at 11:30

Try remounting the heatsink. It may not quite be on straight.

simonnance · 21 Mar 2007 at 11:34

it's seated pretty well...... the Ultra-120 has a very good mounting mechanism with regards to it sitting propperly, and the TIM is applied propperly.

anyway, arent the 2 cores on the same die? So rather than 2 points of contact with th eIHS there should be just one for both cores, so if there was poor contact with the HSF it should effect both.

Kirth Gersen · 12 Oct 2007 at 21:01

I'm finding the same - 6C difference in TAT - any further ideas?

weescott · 12 Oct 2007 at 21:09

Its normal. Don't worry about it.

gash2024 · 12 Oct 2007 at 21:22

I have it with my E6300, speedfan shows core 2 as being higher.
God knows why i think its just a gay thing that the software cant get its head round.

Kirth Gersen · 12 Oct 2007 at 23:32

Ta - mind now at rest...

loltim · 12 Oct 2007 at 23:41

Same on my E6750

marscay · 12 Oct 2007 at 23:43

you will most likely find the heatspreader is to blame, if you lap it you will get more uniform difference between the cores plus slightly better temps overall.

mk17 · 12 Oct 2007 at 23:43

I think you'd do well to see all cores in any multi-core system actually match one another - but its actually a hardware issue, i.e. not s/w based. fact is that material difference matters, and o/s allocation doesn't balance ideally.

I'm not saying that the actual sensors are perfectly perfect, but the differences are not entirely software related.

fwiw, between 5-10 degrees is margin of error, imo, creeping above that and its worth considering lapping one, the other, or both. especially under idle, margin shouldn't vary too far under load, I think

.

mk17 · 12 Oct 2007 at 23:49

Sorry, should mention, that your temps seem pretty good tbh - 'specially under load even for an undervolted system. Plus, I've never had the nads to lap a cpu. Lapping a heatsink I've done, but not both.