The Official PC building thread -3rd Edition

Ahh, I figured it out. It goes in a 3.5" bay. Hmm and I can hide the wires as well. This is good. Believe it or not, I actually did a decent job of cable management.

Unfortunately my camera is broken. I will try to use my phone.

 

]

Still 10C sounds like quite a variance for cores that are mostly in the center of the heat spreader and very close to each other. I don't know if the sensor problem is also happening with the 65 NM but a quick test can help alleviate any concerns. As a rule most sensor should never vary more than 2C and because of their proximity to each other usually one or two sensors is adequate to monitor temperatures.


Here is a 65 NM quad core.

ignore that photo might not be right!

 

Alright, well I check it with real temp?

I will make sure after I get my new screws. When I have everything tightened up then it can only be the stuck sensors I guess.

 

Sorry. LOL It was the right photo but here is a link to Arctic silvers guide for quad cores.

My point was that the pressure loss by a single screw would be more on the outside of the heat spreader. One check of errors by using real temp which is something nice to know so one doesn't reinstall what was right the first time. I know because I did exactly that.

http://www.arcticsilver.com/pdf/appinstruct/as5/ins_as5_intel_quad_wcap.pdf

 

Yeah, but the thing is I'm not at all comfortable like this. I will re-install with all four screws no matter what.

 

I agree, I would too. I was just suggesting the test for peace of mind until you can do it.

 

Agreed. I wouldn't feel happy with a cooler that big installed without all four screws.

 

so a sensor test should look like this then right? not any variance in the cores?



yes thats at idle..lol and i set teh TJ Max as the same as everest and such.lol i hate my temps.

 

Those are the tests. I have stuck sensor on my E8400 which can vary by as much as 10C. I just take the reading from the hotter of the two cores.

 

Makes sense. If I have a difference of 14ºC from Min to Max, what does this mean?

 

Here is an example of a stuck thermal sensor.

It's just a recent measure of your recent temperature range. Capture programs use a lot of processor resources so that alone can cause temps to raise.

 

Fair enough, I was torrenting at the time so I suppose that would have something to do with it. In the example you've shown, is that 7C difference between the sensors indicative of a stuck sensor?

 

Yes, the difference under load can be as much as 10C and on idle around 6C. If you notice the sensor test shows a 4C difference but that is also measured from idle. If I heat things up a bit the difference would usually be about 5 to 6C.

 

while we're comparing notes, here's what I get:



as soon as asus comes out with a new bios, I'm taking her up to 3.6ghz. I have plenty of headroom temp wise. there's a bug with my current bios (0403) and if LLC is enabled it will cause the pc to randomly reboot at idle for no reason a couple times a day (this is very common with everybody using 0403 bios). with LLC disabled or on auto, there is no problem whatsoever. there is a beta bios available (0406) I might try. I was playing in the bios earlier today and was able to get the performance level down to 6. it was at 10 on auto (lower is beter).

 

Soph,

why is real temp showing your clock speed so low, even under stock?

 

As nice as that board is, one does wonder why Asus didn't implement that feature properly from the word go. It's a classic case of fixing the bugs after the product's gone on sale.

Mort: I have speedstep on, maybe he does too?

 

it's just with this particular bios (0403). it was fine with the original bios (0219) and all other prior bios's and is supposedly fixed in bios (0406).

that would explain it

 

Out of curiosity, what caused you to upgrade to 0403?

 

C1E and EIST are on, no point in running it all out if I'm not using it. This way I can turn all my fans down and run things quieter but I can raise them for encoding. My actual overclock is above 3.8.

Here is a shot of CPU-Z with Orthos running