Q9450 will these temps work?

Ok so I've tried reseating my Thermalright Ultra 120 Extreme heatsink about 7 times now with different amounts of thermal paste and using different methods. I tried to spread out a small out over the whole processor, I tried putting a thin line across the middle, tried a small blob in the middle, and finally I spread a thick coating on at least twice as much as I'd put on before. Both the CPU case temp and individual core temps were for the most part the same no matter how I applied the thermal paste. The Ambient temps in my apartment are ~70 Farenheit. When I'm at stock settings and voltage my core temps are about 45-50C idle and CPU case temp is about 30C idle. AT Prime95 load the core temps are about 60C and the CPU case temp is about 40C. After researching what other people have been getting for temps my numbers seem a bit on the high side. But I'm air cooling of course with a Scythe S Flex 1600 RPM fan attached. The strange thing is my temps don't increase all that much when I overclock my processor to 3.6 ghz (450x8).


Here are a couple pics to show you where I'm at. The first one is at prime95 load and the second one is at idle. Are these temps ok for 24/7? I've read alot of people can get 30's and 20's Idle and It confuses me whether they;re talking about core temps or case temp, but according to the overclocking guide over at toms hardware the case temp shouldn't be more then about 5C difference compared to the core temps but I'm seeing about a 15-20C difference there.

The temps do seem a bit on the high side, but with that said are well within safe temps. Have you tried the penny mod?

Looking at your cpu temps in Everest, and comparing to the core temps, it looks like you've got a chip that behaves like an e6600 I had for awhile....that chip always idled high (the cores only), BUT, under load, the increases in temp (cores!) were actually LESS than normal....so, in all, the temps stayed "nicely" away from the Tjmax, and that's what is important, the distance away from Tjmax, under load; >20C is where you would like to be.

M

Originally Posted by dekruyter

Looking at your cpu temps in Everest, and comparing to the core temps, it looks like you've got a chip that behaves like an e6600 I had for awhile....that chip always idled high (the cores only), BUT, under load, the increases in temp (cores!) were actually LESS than normal....so, in all, the temps stayed "nicely" away from the Tjmax, and that's what is important, the distance away from Tjmax, under load; >20C is where you would like to be.

M

That's because idle temps mean nothing for the error rate can be far to large.

I get ~38-40 idle and ~60-64c under load.
My q9450 is at 3.33Ghz @ ~1.2v using a AC7 cooler

Your idle temps do seem quite high for your OC, although the load temps do not seem too bad...

There is quite a bit of fluctuation in temps across your cores, perhaps either the cpu and/or heatsink are not even/level?

Your temps are fine for 24/7, however you should be able to get them lower than that. It really seems that lapping your cpu and/or heatsink could really help the situation. 5c variation across cores is a lot imo


edit: Would you mind posting your voltages/bios settings? It appears we have the same CPU/mobo/ram and I have not bothered taking my OC past 3.33 (416x8), and would not mind a little help getting to 3.6

Originally Posted by BlueFireIce

That's because idle temps mean nothing for the error rate can be far to large.

I don't understand... ok, idle temps mean nothing, ok; but what about "error rate can be far too large" ? What error rate?, and what does it have to do w/ temps?

Opps, never mind...I think you're saying what I have, the idle temps are inaccurate; actually they are not "designed" to be a measure of real idle temps, right?

I think he is refering to the information I've seen on the realtemp calibration page. Its listed under the testing section.

Originally Posted by dekruyter

I don't understand... ok, idle temps mean nothing, ok; but what about "error rate can be far too large" ? What error rate?, and what does it have to do w/ temps?

Opps, never mind...I think you're saying what I have, the idle temps are inaccurate; actually they are not "designed" to be a measure of real idle temps, right?

Yes, the sensors are made to report TjMax, not temps, and the lower the temp the larger the error is in the reading, the only way to really know the temps is the mill a slot in the IHS and put a temp probe in it. However for the most part RealTemp and the like are good enough, but you should only care about load temps unless you are hitting TjMax at idle, other wise it does not really matter.

Originally Posted by BlueFireIce

Yes, the sensors are made to report TjMax, not temps, and the lower the temp the larger the error is in the reading, the only way to really know the temps is the mill a slot in the IHS and put a temp probe in it. However for the most part RealTemp and the like are good enough, but you should only care about load temps unless you are hitting TjMax at idle, other wise it does not really matter.

Yeah I know Unclewebb was telling me that too a while back and he also said that TJMax itself can vary among each processor even in the same stepping and batch. Theres so many variables to consider why my temps are so high. I can prime95 for 24/7 and intel burn test for many hours too but IBT gets me into the mid to high 80's so I dont like to run that for too long. What kind of temps are you getting BlueFireIce I'm just curious. Do you know about how much that lapping helped tempwise? I hear it can knock off a solid 10C or so. My TRUE has lots of small indents and scratches in addition to the machine finish that's on all of them so I'm mulling over wether I should do that next but I'd rather not void my CPU warranty if I don't think temps will improve much.


I've never heard of the penny mod what is that?

My temps aren't usually as high at idle as that pic shows its usually about 50C. I don't mind reseating my Heatsink again with a new TIM application but I'm looking to get 10-15C knocked off but that seems like a pipe dream at this point.

My voltages ATM are 1.35 Vcore 1.44 VTT 1.45 NB 1.8 Vram in BIOs and 1.312 Vcore 1.4 VTT 1.47 NB 1.87 Vram in Everest

Lapping did a good deal for me, I got a good 8-10C off my temps and I could have done a better job also. I get about mid 50's with prime or mid to upper 60's with IBT all depending on room temp. The penny mod can be seen here, some TRUE's don't have good pressure on the base and cause a lose fit, mine was tight, even after lapping, so it did nothing for me, but some people get nice drops in temps from it.