Core 2 Quad Q9450 running to hot?

Nvidia101

[H]ard|Gawd
Joined
Jul 17, 2005
Messages
1,194
My liquid cooled CPU temps seam a bit high since this is not air cooling.

core 0 - 50 C
core 1 -36 C
core 2 -43 C
core 3 -42 C

On idle.

I ran prime 95 for 5-10 minutes and noticed

core 0 - 62 C
core 1 -54 C
core 2 - 55 C
core 3 - 53 C


I have the Velocity Micro E2250 with Core 2 Quad Q9450 . When I opened it from Box. I noticed the liquid CPU cooling heat sink fell half way off. Still attached so it did not damage anything.....
 
You might have gotten a better reply if you had placed this in the Overclocking & Cooling forum, or even the sub forum for watercooling. I don't know all that much about watercooling but it does not seem to have much of a rad in there for cooling, nor do I know how well it performs so I can't say if these temps are out of spec. Is the CPU OCed? Did you reapply the TIM before reseating the water block?
 
You might have gotten a better reply if you had placed this in the Overclocking & Cooling forum, or even the sub forum for watercooling. I don't know all that much about watercooling but it does not seem to have much of a rad in there for cooling, nor do I know how well it performs so I can't say if these temps are out of spec. Is the CPU OCed? Did you reapply the TIM before reseating the water block?

not overclocked. I did not re-apply nothing as I did not have any extra paste so just what was already on it..
 
Then you need to take it off and clean both the CPU and the water block and reapply the TIM. Might help your temps.
 
Then you need to take it off and clean both the CPU and the water block and reapply the TIM. Might help your temps.

whats wrong with the TP already on it? it cant really be the couse of high temps can it?

also how do I know the liquid cooling works? should the tubes be cold? I always had fans so this is all new to me.
 
As I said, I don't know anything about this cooler, those could be 100% normal temps for it. The TIM spreads out when a water block/HS is put on, if it comes off that gets all messed up and you get uneven contact, not only that but seeing it could have been off for so long it could have also been contaminated with anything. The rule of thumb is if a water block or HS is removed for ANY reason you reapply the TIM, it keeps you from having trouble down the line. Even if it's not the problem, you will never know until you reapply, you start with the simple stuff first. TIM is cheap, another cooler isn't.

Pick up some OCZ Freeze or something, I don't know what people are calling the "best" right now, as I use Coollaboratory Liquid Pro, and have not seen anything come close to it in my set up. Might want to pick up some ArctiClean, it works really well.
 
It depends on your fan(s). If you are running your fans silently and you are ok with the temp, then everything is fine. 50c is normal if your goal is to minimize noise by turning the fans down. If you have only a single 120mm rad then you'll need to push alot of air through it to get it cooler than that. 55c at load is a stable temp though if you never plan on oc'ing, so if your rig is silent, then just enjoy it.
 
Update:


I just ran prime 95 100% CPU only for 14 minutes and got this:

core 0 - 70 C fluctuates between 70-71
core 1 - 63 C
core 2 - 65 C
core 3 - 65 C

It did 9 tests. 0 errors or problems.

Also the CPU is not overclocked running at standard speed of 2.66GHZ per core.
 
I have a X3350 (Q9450) --> Zalman 8700 NT HeatSink --> Stock Settings

100% load - 55C - 60C on all cores --> using Realtemp

Not sure which program you are using to check your temps, but CoreTemp and RealTemp have base numbesr so that will affect your temp numbers a little.
 
I also have a X3350, My CPU is also watercooled and overclock to 3.6. Do what BlueFireIce said redo your thermal paste. Are you still using the factory push pins? If you are get a bolt kit that uses nuts and bolts to hold your waterblock on. You are probably not getting good connect between the cpu and the waterblock with the push pins. The push pin loosen up after a while and the water block does not have good connect with the cpu,. Even with my overclock and prime95 running ,I am still in the low 60c's with a room temperature of 27c [80F ].
 
45nm Core 2 Quad processors have temperature sensors that get stuck at idle and at full load they don't necessarily use the same TJMax for each core. Both of those facts come from Intel's presentations last year. That's why Intel has never recommend using their sensors to report accurate core temperatures.

Run the RealTemp Cpu Cool Down Test and post your results. It will show you how screwy your sensors are. Most of your issues have very little to do with thermal paste and a lot to do with the quality of sensors that are on your chip.
 
Idle means doodoo, Yes its where most of the time pc is at, but loads matter the most in a gaming rig.

One thing im wondering is your cooling, It might be a liquid cooling system, but it can still cool like crap. Your temps are indeed very high for stock. Have you tried uninstalling the liquid cooling and trying an air cooler? It might just be the liquid cooling is just subpar to cool the Q9450, That or the base is not making good contact with the cpu, Have you tried removing the old tim and replacing it with as5/mx-2? I recommend doing so and seeing if that fixes it.
 
I also have a X3350, My CPU is also watercooled and overclock to 3.6. Do what BlueFireIce said redo your thermal paste. Are you still using the factory push pins? If you are get a bolt kit that uses nuts and bolts to hold your waterblock on. You are probably not getting good connect between the cpu and the waterblock with the push pins. The push pin loosen up after a while and the water block does not have good connect with the cpu,. Even with my overclock and prime95 running ,I am still in the low 60c's with a room temperature of 27c [80F ].

yes it has factory push pins. I have never done this before. took me like 30 min to put the water block correctly or at least I hope so without one of those annoying pins popping out. not to comfortable unless you have small hands...


45nm Core 2 Quad processors have temperature sensors that get stuck at idle and at full load they don't necessarily use the same TJMax for each core. Both of those facts come from Intel's presentations last year. That's why Intel has never recommend using their sensors to report accurate core temperatures.

Run the RealTemp Cpu Cool Down Test and post your results. It will show you how screwy your sensors are. Most of your issues have very little to do with thermal paste and a lot to do with the quality of sensors that are on your chip.

I just ran the test and got this:


test1fo9.jpg
 
My overclocked 9550 runs at 35c idle....on air....so imo your temps are to high :(

What does yours load at?

I also have a X3350, My CPU is also watercooled and overclock to 3.6. Do what BlueFireIce said redo your thermal paste. Are you still using the factory push pins? If you are get a bolt kit that uses nuts and bolts to hold your waterblock on. You are probably not getting good connect between the cpu and the waterblock with the push pins. The push pin loosen up after a while and the water block does not have good connect with the cpu,. Even with my overclock and prime95 running ,I am still in the low 60c's with a room temperature of 27c [80F ].

What are your voltages?


OP: With my Q9650 at 1.25v and 4.2ghz...when i load Prime95 its high 50's...never above 60....idle temps however are still mid/high 30's. Even for a Q9450 your temps seem high though. What kind of water cooling setup do you have?
 
nissanztt90

My cpu voltage is 1.325, Memory is at the stock setting [ 2.1 ] FSB is .05, [G]MCH is normal [ stock setting ]. My biggest problem is my room temperature, I have two computer running in a small bedroom. When I bring the room temperature down to 20c to 22c my cpu load temperature will come down to 56c to 58c. Even with the high room temperature my cpu temperature are 60c to 62c. Also like unclewebb said I am not sure if my sensors are reading correctly.
 
Well I think i need to reapply my thermal paste.

What are you using to load the CPU? IBT or Prime95?
 
Prime95. Like a said I am not sure how accurate my temperatures are. I also lapped my cpu and my D-Tek Fuzion v1 water block
 
Prime95. Like a said I am not sure how accurate my temperatures are. I also lapped my cpu and my D-Tek Fuzion v1 water block

I think your temp readings are fine. Assuming ambient temp differences are minimal (fairly warm in my room) my IBT temps at 1.25 are only mid 60's...and mostly low/mid 50's (peak is 58, but doesnt normally stay that high) with prime at 1.25v so 1.32v would probably load low/mid 60's for me as well with p95.
 
I just ran the test and got this:

Did you run Prime95 Small FFTs on all 4 cores? Did you have anything else running in the background like a virus scan or did you keep using your computer while you were testing? The reason I ask is because your results don't look quite right. This test can tell me a lot about what's going on with your sensors but you need to run it carefully for best results.

When running this test, at full load you should see your highest temperature and then it should gradually step down in temperature as the load is reduced. The pattern yours shows is that you might need to re-do this test if you want to learn about your sensors. In about 20 minutes I'll do a test on my Q6600 including a temperature graph to try and show a typical example.

Your core 0 and core 1 sensors are reporting basically the same temperature at 41% load as they are reporting at 98.6% load. That's somewhere between very unusual and impossible if you were running Small FFTs on all 4 cores.

Your data is not conclusive but it looks like core 2 and core 3 are going to be your best sensors. Core 1 probably has some slope error and core 0 seems to be using a significantly different TJMax value. A TJMax difference of 10 degrees between two cores on a 45nm Quad is not unusual at all. The majority of your problems are all sensor error related.

Here's how my Q6600 looks.

q6600testtb1.png


I'm air cooled with the fan on low so there is a much bigger spread from idle to full load. It shows that as the CPU Load decreases the temperature decreases. At each step of this test there should be a distinct value and Distance to TJMax should constantly increase at each step. It's easy to see that in my case, core 0 and core 1 are very well balanced and they also happen to be my best sensors.

With data like that, it's much easier to see the differences in slope from one sensor to the next. When the slopes are similar between two sensors but there is a constant difference of 5C or 10C then that is a good indication of differences in TJMax. Post another test and I'll try to come up some calibration numbers you can use so that your reported core temperatures are closer to reality and a lot more consistent.
 
core 0 - 50 C
core 1 -36 C
core 2 -43 C
core 3 -42 C

On idle.

I ran prime 95 for 5-10 minutes and noticed

core 0 - 70 C
core 1 -64 C
core 2 - 65 C
core 3 - 63 C


Those are my current temp with stock intel cooler at 3.2ghz q9450 so i backed it sown to stock till my TRUE gets here.
 
Just run CoreDamage for 24 minutes. Room temperature is 27c [ 80F ]. The pump is set to 5 and the fan controller has the fans are at their lowest setting [ slowest ] . At their highest the temps were

65c
63c
62c
64c

according to Real temp.
 
In realtemp, the Tjmax is 95C, while in CoreTemp, Tjmax is 100C. That is why RealTemp always shows 5C lower than CoreTemp does. Not sure which is the right Tjmax, but to me, 100C is more reasonable.
 
Back
Top