Q9650 stuck @ 44C

[Phunk0ne]

Finally decided to replace my C1 9550 for my E0 9650, after it has been sleeping in it's box for 2 weeks.

I was instantly blown away at how cold this quad stays in Idle and stress (see sig for set up)
But funny enough, core3 shows it is stuck @ 44C while all other cores reach a stunning 25C @ idle. But when stressed with IBT, it hardly reaches 48C.

when all the other 3 cores climb their way up in producing heat, core3 stays @44, untill the cores reach an approximate of 45 and then I see core3 move upwards in heat.
after 5 runs of IBT @ max all cores max out around 48ish.

Should I RMA this chip, or is it safe to say that, while core3 is stuck @ 44C, and all other cores show a nice line of temps, that I can safely assume that core3's temp is around the 25ish Celcius as well?

 

[JMapleton]

Wow, that is terribly cold. What are you using to guage temps?

 

[BonzaiDuck]

Originally posted by: Phunk0ne
Finally decided to replace my C1 9550 for my E0 9650, after it has been sleeping in it's box for 2 weeks.

I was instantly blown away at how cold this quad stays in Idle and stress (see sig for set up)
But funny enough, core3 shows it is stuck @ 44C while all other cores reach a stunning 25C @ idle. But when stressed with IBT, it hardly reaches 48C.

when all the other 3 cores climb their way up in producing heat, core3 stays @44, untill the cores reach an approximate of 45 and then I see core3 move upwards in heat.
after 5 runs of IBT @ max all cores max out around 48ish.

Should I RMA this chip, or is it safe to say that, while core3 is stuck @ 44C, and all other cores show a nice line of temps, that I can safely assume that core3's temp is around the 25ish Celcius as well?

Mapleton has his point, there.

Intel had published a statement about these Penryns somewhere on their web-site, quoted in the "Help" screens for RealTemp. [Paraphrase]: "The sensors are not intended for reporting idle temperatures."

How that would play out if you choose to RMA the processor, I don't know. I think some folks chose to RMA. I heeded the disclaimer and figured that "they don't give-a-s***" about the functionality of the sensors at the low end of idle. But the behavior of your single whacky core parallels the behavior of both cores on my E8600. And I'm a lot less worried about the processor's integrity: it passes Linpack and PRime95 testing at my chosen over-clock . . . just fine. If the core tJunction temperatures can be relied upon, I'm maxing out in the 60's centigrade depending on the room ambient.

 

[VirtualLarry]

If it overclocks well, then don't RMA it. Just be aware that your idle temps are off on that core. On my OCed Q6600, one core is always reading about 5C lower than the other cores. But the chip overall OCs good. I wouldn't want to get rid of it.

 

[Phunk0ne]

decided to keep the CPU, and yes, thats really "cold".
I use RealTemp and Coretemp to check on the cores. I'm can positively say the stuck sensor is just... well really stuck.
But this chip overclocks like a champ.

I have it running @ 3.6Ghz @ 1.25V, I think I can even lower the volts at these speeds, but that is for upcoming Sat/Sun.
At this speed, it hardly breaks 48C @ IBT max stress, that's just nuts. Idle it sits around 28C @ 20ish room temp.

 

[BonzaiDuck]

Originally posted by: Phunk0ne
decided to keep the CPU, and yes, thats really "cold".
I use RealTemp and Coretemp to check on the cores. I'm can positively say the stuck sensor is just... well really stuck.
But this chip overclocks like a champ.

I have it running @ 3.6Ghz @ 1.25V, I think I can even lower the volts at these speeds, but that is for upcoming Sat/Sun.
At this speed, it hardly breaks 48C @ IBT max stress, that's just nuts. Idle it sits around 28C @ 20ish room temp.

I had always been a stickler for "perfection." Almost to the extreme of anal-retentive. But before there were "core" sensors, there was tCase -- the sensor that measures temperature at the center of the IHS processor-cap.

For the 65nm C2D and C2Q CPUs, tCase was supposed to lag behind the core sensors by about 10C degrees. It only stands to reason that it would.

Now, with Intel disclaiming the usefulness of the core sensors for idle temperature measurement, I find the following phenomenon with my choice of options. At load temperatures, my tCase reading is about equal to the core sensor readings. At idle and temperatures up to 51C, my core sensors remain stuck. Since tCase seems to behave properly at the idle and low-end temperature range, I'll trust it before I trust the core sensors for reading temperatures above the stuck threshold.

For that, you're not really worse off for anything, because Intel thermal specs reference tCase -- not the tJunction core sensors.

 

[dust]

Same issue here, got a sensor stuck, the second core reports 35c without moving at all while the comp is idle, climbs to 40-42c under heavy load. It doesn't bother me though, since the other cores report well and the load temp is @ ~60c. It's common sense also that the second core is fine but the sensor isn't. As long as the rig is stable I don't care. It's been running for six months already without issues and it'll probably be sooner useless than dead.

Try to push it more, @3,6 doesn't justify the replacement of the 9550 as I'm sure that one could do this easily enough. The 9650 are meant to hit 4-4,2.

 

[Thor86]

This is one main reason I didn't bother getting into Penryns. Stuck sensors would drive me batty.

 

[BonzaiDuck]

Originally posted by: Thor86
This is one main reason I didn't bother getting into Penryns. Stuck sensors would drive me batty.

Yeah -- I know what you mean. But the lag in BIOS revisions, per coding revisions for reading the Penryn sensors, was just as bad. Even so, there was always TCase, and that's the basis for the Intel spec.

I could see waiting six months for bleeding-edge hot-dawgs to post their results and problems with motherboards. I wouldn't have imagined this problem with sensors from Intel itself.