Hmmmm - Strange temps Q9550 ??

Pooh · 21 Mar 2010 at 18:45

I've recently moved from a Q6600 to a Q9550 and I've got he 9550 clocked @3.73 at a vcore of 1.3V on a water cooled rig.

Both the CPU & water block have been lapped and have a very good thermal paste applied as per the manufacturers recommendations. Now they say that the temps will drop after the paste has gone through a few heat cycles and reach maximum efficiency after some 200 hours of running however

I'm a bit confused re the idle temps as they seems bit high and the difference to the load temps is what's getting me.

When it's running Prime95 and all cores are loaded at 100% there's not much movement. I'm happy with the load temps but I thought the idle temps would be a lot lower. Even if I set the CPU at stock clock and volts there's hardly any difference in the idle temps from stock to 3.73

Any thoughts

Rroff · 21 Mar 2010 at 18:51

Thats water cooling for you imo - atleast with the cheaper stuff it doesn't bring idle temps down much but can handle a massive amount of heat without a big rise in temps when the loads on.

My Q9550 temps are pretty similiar to yours - I'm using a Coolermaster V8 with fan mod - idle is 39, 37, 38, 38 atm with 1.35 vcore - OCCT never gets above lower 50s and IBT doesn't get much above 60.

Mr Paul · 21 Mar 2010 at 19:07

45nm temp sensors are generally pants on the Core2 line. Sticking sensors being the major problem. Do a sensor test in realtemp and post a pic of the cool down temps.

pastymuncher · 21 Mar 2010 at 19:23

It look's like stuck sensors. The 45nm cpu's seem to be terrible for this. I have had a E5200, E6300 and E8500 all with stuck idle sensors. Once the temp get's over whatever it's stuck at it will display the correct temperature which is why you have little difference between idle and load temps.

El Jimben · 21 Mar 2010 at 20:21

I just bought a C1 Q9550 from the MM and I think all of the temp sensors are crocked to be honest. Cores 0 and 1 show movement but 16 degrees difference at idle when the chip is lapped and the base of my TRUE is flat makes me think core 3 must stuck too. They all move under load but only core 0 looks like it's giving a realistic temperature. I bought this from Tute and he said that the difference between cores when under load was only a few degrees so I wonder if the chip might've taken a knock in the post and it's screwed up now?

Pooh · 21 Mar 2010 at 20:43

Here's the cool down results. The 1st small FFT run gave me an error in core 2 so I pumped a bit more juice through vcore, now running at 1.33 and stable

scooby555t · 21 Mar 2010 at 23:40

my 1 idles at 32 32 37 39 also under water at 3.60 and there is very little movment at 100% stress
so i think you are ok

setter · 21 Mar 2010 at 23:47

I had a q9550 a while back, 1.200 vid, ran it at 3.8ghz on 1.216 vcore, temps were 56-54-46-47, i reckon core's 2 and 3 had stuck sensors, never worried about it because cores 0 and 1 ran so cool, was a great cpu, only sold it to move to i7.

Marine Iguana · 21 Mar 2010 at 23:52

My Q9550 was the same had two sensors stuck so never bothered to rely on em

Mr Paul · 21 Mar 2010 at 23:53

Pooh said:
Here's the cool down results. The 1st small FFT run gave me an error in core 2 so I pumped a bit more juice through vcore, now running at 1.33 and stable

Yeah, looks like cores 1, 2 and 4 have stuck sensors near idle. I would use core 3 for idle temps. Load temps all 4 are ok.

Pooh · 22 Mar 2010 at 01:06

Thanks Guys

I need to run a bit more juice through mine to get it stable at 3.8 as I've got a P35 mobo and 8Gb Ram.

The chip vid is 1.1625, so quite a good'un from what I can gather

Is there any way of 'un-sticking' them

Would a run at high temp, maybe free them off, or are they just goosed

Mr Paul · 22 Mar 2010 at 08:13

No, they are goosed