Made a move from Q6600 B3 to Q9450...

[ShadowFlareX]

It sucks to find your Q9450 runs hotter than a Q6600 B3, about 10oC at most. First time I slapped the Q9450 in there, I used the Arctic Silver's line method to apply the AS5. Convinced myself that I probably did a piss poor job, I reseated my TRUE-120 but now with the thin-layer-over-the-IHS method... and is still giving me the same results.

I did a search regarding Q9450 temps, found out it could be stuck temp sensors on the core, but it doesn't look like it.

It's running at undervolted stock speed, 1.15V BIOS. Rampage Formula that I have only allow 1.1V BIOS as the lowest VCore, tried that but only took 5 secs when the cores started failing one by one on Prime95, it's sad. Core Temp reports VIDMax of 1.225V and VIDMin of 1.150V, pretty much an average chip, but worse than average that it couldn't run stable with 1.1V at stock speed.

Back to the heat issue, I've been hearing about Real Temp having a lower TJMax and a number of people trust this more than Core Temp. So I gave Real Temp a go, and it's reporting 10oC lower than Core Temp, and it's as if I'm seeing my ol' Q6600 B3's temperature again.

Room Temp: 32oC / 90oF

Core Temp Q6600 B3 @ stock, 1.15V
Idle: 46, 47, 41, 43
Load: 55, 56, 51, 53

(Unfortunately I never tried out Real Temp on Q6600)

Core Temp Q9450 @ stock, 1.15V
Idle: 55, 56, 49, 52
Load: 67, 64, 61, 62

Real Temp Q9450 @ stock, 1.15V
Idle: 44, 46, 39, 41
Load: 57, 55, 51, 52

Since I like Core Temp's interface better (I can see all 4 core temps on the system tray), I adjusted Core Temp's TjMaxOffset to -10 in Settings.ini so it shows pretty much what Real Temp is showing. I'm not sure if reducing Core Temp's TjMax by 10 (so making TjMax 95oC) is a wise choice, what do you think? Of course I like seeing lower temps, but is it accurate?

 

[Idontcare]

Don't discount the fact that CPU temps are going to scale by power-consumption divided by die-size as a rough rule of thumb.

When you make the chip smaller in die-size but crank up the transistor count and clockspeed such that it consumes just as much power then the temperature has now choice but to increase unless you increase the thermal conductivity by a commensurate amount.

I'm not saying your temps are to be expected, I do not own a Q9450 but I do own plenty of B3's and G0's. All I am saying is that your expectations of having lower temps just because you went from 65nm to 45nm chips is probably part of the reason you feel a bit of buyers remorse (seemingly).

 

[Markfw]

I see the same thing. My X3350 (Q9450 in disguise) runs @ the same temp or higher than my Q6600 G0 stepings.

 

[Pabster]

Don't worry about it, your temps are fine.

 

[frodbonzi]

I'm using water-cooling... but my q9450 has yet to pass 40 degrees, even at full load... and I'm OC'd to 4ghz...

Still doing some tweaks, but so far at 1.2V, looking pretty good... never had a 6600, so can't compare...

 

[JK949]

I went from a 6600 to a 9450 too and experienced the same temperature increase. It turns out that a VERY LARGE batch of 9450's have defective sensors and
report anywhere from 10 to 15c above the actual temperature of the cpu. Intel did not issue anything in response to this because they don't see it as a major
problem unless people are over clocking them, which would void the warranty anyway. Mine is showing 58 overall and 55 to 60 for the cores.
If I take 12c off of that it is identical to my 6600 over clocked to 2.68. The real test was touching the heatsink. Barely warm to the touch.
All this is with the throttling disabled so it runs at 2.66 all the time.

 

[bryanW1995]

make sure that you're using 95c as tjunction when you compare your 45nm quad temps. in fact, just stay completely away from temps since all temp monitoring programs are guessing at the actual diode temp. instead use distance to tjunction. my Q6600 loads about .15v higher than my x3350, runs 100 mhz slower, has a 3-4c lower ambient temp, and typically has the same distance to tjunction. oh, and it's tuniq is lapped while the x3350's tuniq isn't. My 9450 runs at about the same distance to tjunction as the others but it's stuck with a zalman instead of a tuniq like the others.

 

[ShadowFlareX]

Originally posted by: bryanW1995
make sure that you're using 95c as tjunction when you compare your 45nm quad temps.

Yep, already did that, Core Temp is using TjMax of 95oC

Thanks for the reply guys, I'm at ease now to know about these facts. I have headroom for overclocking now, yay.

 

[ShadowFlareX]

Right, finally managed to squeeze my system into an AC'd room, bloody hell idle and load temps dropped 10oC! I'm seeing idle temps of 37oC and 55oC Prime load.

Managed to overclock it to 3.6GHz with little effort, load temps were reaching 65oC with 1.4V (BIOS). Was happy and didn't push it further. Kinda wishing I got the X3350 instead, it would've done 3.6GHz @ 1.3V. Didn't notice its lower voltage advantage in the forums until after I bought it, oh well.

Settings I had for 3.6GHz:
VCore: 1.4V BIOS / 1.37 Real
PLL: 1.5V
VTT: 1.30V BIOS / 1.28 Real
NB: 1.31V BIOS / 1.21 Real
SB:1.5V