I just tried the stock Intel Cooler on my Q9300 again............and real temp reports core temps of 49/47/43/48 at idle. I used AS5 too. Its night and nice and cooled down in here and all case fans are on 100%. My HDT-S1283 doesn't do any better except with running Prime95 small FFTs. The core temps while running Small FFTs went up to 90 C in just a few seconds. That is ridiculous. I am going to order the bolt-thru kit in the next few days and try it out with my HDT Cooler. Nothing that I do seems to work. I need any suggestions that you might have for me.
Lowering the core voltage is only a partial remedy, but a great idea for anyone to do. I was able to get my Q9450 down to 1.8 volts at stock speed and was 100% stable under long term full load (video editing).
Never thought it would run stable like that, but it was pretty neat, and it ran about 5c cooler, too.
Anyways, either there is too much thermal paste being used, not enough paste, or you do indeed meed to upgrade the mounting hardware to the backplate method to see if that helps.
What voltage should I most likely try? Or just keep lowering it until I get the best temps? And also I just tried reducing the Clock to 266x6 Multiplier and it was the EXACT SAME TEMPS......I seriously don't get it. It seems more likely that the DTS are faulty (as common in most 45nm procs) than the Cooler. Because I get the same temps almost with the stock or the S1283. But I gotta try what shoota said.
I was able to get my Q9450 down to 1.8 volts at stock speed and was 100% stable under long term full load (video editing).
I really hope you mean 1.18 or 1.08V, because at 1.8V, your Q9450 will be toast any time now.
@OP: You may want to also consider using Real Temp instead, and also read the interesting documentation and explanation of inaccurate temperatures on 45nm Intels:
http://www.techpowerup.com/realtemp/docs.php
Those idle temps look like they are really just miscalibrated. Try touching the heatsink - it's probably cool. What are load temps (Prime95 torture test) like?
Wait a sec........zipz0p? You should re-read my post. I am using Real Temp and it goes to about 90 C in full load under prime95s Small FFTs test. I just lowered the voltage to 1.125v in BIOS and it might of made a 1-2 degree difference but I'm not sure. Is it possible that it could be too low of a voltage?
If it idles in the mid 40s, and goes to 90 while working, it sounds like the heatsink isn't attached properly. (or the thermal paste isn't right.)
Take the motherboard out of the case, and put the heatsink on then. Don't forget to apply the correct amount of TIM. It might seems like a PITA to take the motherboard out, but it really is the best way.
---------------
The voice of REASON
Do NOT feed the TROLLS!
Always a DEMON!
But aznguy0028...do you think that it is a HSF problem or a proc problem? The 90 C is only with the stock cooler. it gets in the high 50s with the Xigmatek. But even 50 C at stock on full load is still too hot.
And I only got the Q9300 2 weeks ago. And how does RMA work? I also might try and use less paste and see.
EDIT: I really would not have to take the Mobo out if I don't need to, but if I have to I would. I don't understand how the Stock Intel Cooler could not be attached right cuz its pretty simple.
Sorry abouto that jman. I definitely didn't read that carefully enough. Not enough sleep
90C at load is wayyyy too hot. But 50C at full load with the Xigmatec isn't necessarily too hot, that sounds fine. I get up to almost 60C full load with a similar cooler at 3.2GHz on my Q9450. Depending on the program I use, my idle temp is listed around 20C, 30C or 40C.
high 50's is fine under 100% load. that is completely safe.
most people would agree with me that you should stay below 70 on full load...i try to stay 65 or under. but 70 seems to be a point that most people agree on here.
my Q9450 goes to around 58-61 on full load in Prime95 too, its no big deal. the place where i live is 100+ everyday during the summer.
i don't think you should have too much to worry about, post more info if anything else troubles you, im sure people on this forum will be able to help, myself included.
since you have the Xiggy, it is highly recommended that you get the bolt through kit, much better contact w/the CPU.
I just reseated the Stock heatsink and put AS5 on again after cleaning the old paste off. I was idle for 10 mins and got core temps of 40/39/36/39 with Real Temp. Thats better for sure. But the method I used for applying the AS5 was i put a line on the top of the proc and took a credit card and smoothed it out into a nicer thin line. I am still at 1.125 voltage in the BIOS. I think it was at 1.25 or something before. Also I noticed for some reason my multiplier was set to 7 instead of 7.5, which it should of been. So it could only get up to 2333MHz. But now it does 2500Mhz.
I also have a really important question. What OC could be safely attainable with this? I was wanting to overclock the whole time but couldn't do it with those previous temps. I'm gonna also run Small FFTs again and see if it does any better.
EDIT: This is much better now with stock cooler. I ran prim95 Small FFT for 15 mins and got 64/61/57/59. Not too bad. But it is a big improvement from 90 C.
Another EDIT: Now running at 374x7.5 for 2800MHz. Under Small FFTs Real Temp reported 74/72/67/68. That seems stable I guess..But this is also at stock Vcore of 1.225v. But for some reason CPU-Z says the core voltage is 1.152v, but it says 1.225v in BIOS. Before my Xigmatek did a better job at full load and was about high 50s on full load in Small FFTs at stock settings. I think that if I put the HDT-S1283 back on and use the same method of applying the paste that I might finally have something. But maybe I should sleep cuz it is 2:30 AM, but I need to fix this..lol.
Message edited by jman07_93 on 07-13-2008 at 11:29:51 AM
90C is hot but it won't burn up your chip in 5 mins like you would think. I know someone who's been folding at >90C 24/7 for months. The only problem with temps that high at stock is it means you won't be able to OC much or you'll start throttling, which isn't dangerous, it just renders your OC totally useless and perhaps slower than at stock. But it's good to see you got it down where it should be.
Sounds like you're on the right track. And if you don't have the retention bracket for the Xigmatek, you should get it. Pushpins suck suck suck.
I don't think you need to smooth the thermal paste out. Often, people put too much on, though. Just a very little dab will do the trick. Put it in the center of the CPU's heat spreader, then place the Xigmatek directly on top and bolt it down without twisting it too much. The thermal paste should spread itself out and help the cooler make optimal contact with the CPU heat spreader. It can take some "burn in" time for the paste to spread out and seep into the micropores in the metals and get the best possible contact and coolest temperatures.
As for safe OCs, just try not to go over 1.35V and stay within the temperature guidelines mentioned above. You should pretty easily be able to break 3GHz as long as you can keep that FSB stable.
Yep. I'm slowly feeling much better about it. Tomorrow........wait...I guess it would be later today since it already is tomorrow..lol. But I'm going to try the Xigmatek again and see. My guess is it will work better this time and I can OC a bit more. I actually don't really think that anything is defective. I think that I might just be doing a few things wrong here and there.
EDIT: I am definitely going to buy a bolt-thru kit for sure. I just have to wait 1-2 days cuz I don't have enough money in my paypal account balance and I don't have a credit card and they don't take eChecks on Sidewinder Computers. I have noticed that taking the HSF off and putting it back on is becoming a breeze to do.
Message edited by jman07_93 on 07-13-2008 at 12:07:06 PM
It's entirely possible that there is a problem with one or both surfaces being irregular, as well. Say for instance your CPU had a bad scratch, pit, etc.
