E8400 TEMP ISSUE,VERY DETAILED,looking for answers！！！！

[harlam357]

However...the exhuast feels cooler then what came out of my older Opty machine, which also load temped at ~61C. Dunno. There's obvious observer bias here, though, lol. (aka I don't want to believe those temps )

Ditto... I replaced my Opt165 with this setup using the same case & heatsink. I basically just did a board, cpu, memory swap. Now my Ultra-120 is also more inline with the rear exhaust fan in my P180 than my DFI SLI-DR was (socket was lower on the board). Comparing to another Opt170 I've got setup, it's exhaust is much warmer than this E8400 setup.

I hadn't really thought about it before... I just assumed 45nm would be cooler. They obviously clock better but maybe there really is a heat/density issue and who knows what role high-K is playing in that effect.

 

[gluety22]

wierd

Funny I seem to have the same problems...

oh no hotter!

my cheap gear

 

[Coreyhm1]

Heres waht the CPUZ dump says

load


idle

 

[harlam357]

Man... you got an excellent chip. If I could hit 4.0GHz @ 1.28v maybe my chip wouldn't read so hot.

it could be a bad one,i tried 3 e8400,all on 1.3v 4.05ghz
idle:1st one at 49/49,2nd at 29/42,3rd one at 42/42
max load:1st 70/69,2nd 57/56,3rd 60/59

And if this isn't any indication that the core reading cannot be trusted I don't know what is. I have a hard time believing that a bad hsf mount or a bad ihs contact can cause a 10C swing between cpus under the same conditions.

 

[thehacker010]

Coreyhm1, is that 1.28v at load after Vdroop? What is your voltage set as in bios?

I just started a prime95 test a few hours and it is going steady at 4.0GHz (450x9) @ 1.280v load.

My temps are reading as 27*c idle and currently at 53*c load. It might go up a few more degrees by the end of the day, but it has pretty much leveled off. If it appears stable after 8 hours, I am going to call it a day.

 

[Coreyhm1]

not sure if this board has much or any vdroop, its set at 1.28v in bios, i think the real voltage is 1.26v.

I honestly havn't paid much attention, i started out at overclocking with 1.28v, it might be stable with lower voltage, i doubt it though.

 

[Blkout]

Some of you seem to be forgetting the extra cache compared to the E6 series. The E8 series is going run cooler IF the cache were the same but with an extra 2MB of cache, that creates more heat also. I'm still guessing when all is said and done and these temp issues get sorted out, the E8 series is only going to run slightly cooler than the E6 series at the same given clocks simply because of the extra cache generating more heat. However. they still overclock better than the E6 series because the more efficient design.

 

[Blkout]

not sure if this board has much or any vdroop, its set at 1.28v in bios, i think the real voltage is 1.26v.

I honestly havn't paid much attention, i started out at overclocking with 1.28v, it might be stable with lower voltage, i doubt it though.

1.28-1.30v under load and 1.32-1.34v at idle would be about right for 4.0GHz.

 

[harlam357]

I just talked to a guy on the Folding Team who is running a QX9650 @ 4.33GHz on 1.38v and the stock heatsink. He is only seeing ~60C load across the cores as reported by SpeedFan. I believe SpeedFan reads the same registers as CoreTemp with regards to individual core temperature so the readings should be analogous vs. CoreTemp.

Since a quad accounts for twice the silicon and thus twice the heat that just means that these temp diodes on the 45nm duals are WAY off. Either that or his quad should be at the meltdown point.

 

[Coreyhm1]

1.28-1.30v under load and 1.32-1.34v at idle would be about right for 4.0GHz.

its 1.28v idle, and about 1.26v load (cpuz says 1.26v idle and 1.24 load). It may be wrong, its set to 1.28v in bios.

 

[Blkout]

its 1.28v idle, and about 1.26v load (cpuz says 1.26v idle and 1.24 load). It may be wrong, its set to 1.28v in bios.

Going to be a little while longer before we'll know for sure. New BIOS, new CPU-Z and new Coretemp should fix these issues.

 

[thehacker010]

One thing I find a bit odd is that Speedfan is constantly 10*c lower than Coretemp. Perhaps someone can enlighten me on why this may be. The only thing I could guess is speedfan is using a different TjMax value.

Here is a pic taken a few minutes ago while prime95 was still running in the background.

 

[harlam357]

I think in that case SpeedFan is reading the bios cpu temp value since it's not showing a per core reading.

 

[Omsion]

I think in that case SpeedFan is reading the bios cpu temp value since it's not showing a per core reading.

And that temp is off-die, so it won't be as accuarate...normally. Who know's now with all the possiable misreadings.

 

[Coreyhm1]

This is interesting

 

[Omsion]

This is interesting

<IMG>

That sensor is off the CPU, though.

 

[imposter]

New e8400's will not work with coretemp. ITs not software issue at this moment.

 

[thehacker010]

New e8400's will not work with coretemp. ITs not software issue at this moment.

I haven't heard that before. Then again, I haven't looked at the thread showing everyone's pack date and temps.

Just so you guys know my pack date was 12/20/07, which would make it an "old" one.

 

[OnDborder]

Mine is rma'd.. I have too much of a difference in temps and the temps do not flucuate, they seem to be locked at the same temps.

 

[harlam357]

New e8400's will not work with coretemp. ITs not software issue at this moment.

Pack data of 1/9/08 here... does that make mine "new"?