The Official PC building thread -3rd Edition

The most basic water cooling systems comes close to, matches, or beats most if not all good air coolers. I was doing water cooling the hard way and still am, but then along comes Donald (64026402)with a closed sealed water cooling system which I've never heard of or used. When I hear there was a couple on sale I quickly purchased two more which are still in their boxes a year and half later. LOL

I think that kits are still quite a bit better than most sealed units but sealed units are getting better. You can still purchase sealed water cooling units that already have the liquids installed and sealed, and all that you have to do is install the CPU block much the same way that you do air cooling systems. Setting one up isn't as hard as one might think, but if you want my advice go for a kit.

 

Those are very decent temperatures.

 

Put it this way Bigwill, the PS3 version of COD4 uses medium quality aliased textures at 1024x600. My PC uses high quality Anti-Aliased, filtered textures at 2560x1600. There's no comparison in the quality of the image. You already have half the gear for encoding etc, you get the CPU, RAM, video card etc. All that's left to turn it into a gaming PC is a graphics card - you can always hook it up to your HDTV if you don't want to buy a large widescreen monitor.
I don't wish to start a console vs PC debate, but the only reaosn I would own a PS3 would be to play the games it has that the PC doesn't, and the same goes for the 360. They are inferior substutions for games that the PC does have, very much so.

As far as watercooling goes, you're best off doing it the 'hard way' so to speak, you'll get far better results.

 

I'm not sure why you have that much temp variance between the cores. could be a sensor problem I suppose. what is your ambient temp?

here is what I get with coretemp:

 

Really?? What I want to know is why one core is running so much hotter than the rest. I'm not OCed, and at idle, I thought those were high. It's weird because before I OCed the first time, my temps were actually a little lower. Than I OCed, they went high, and after I came back down to stock, my temps rose like 2C. I don't know...I wanted to OC to about 3.2 GHz, but I'm not sure how high the temps will go...I don't want to go too high....

 

I covered that in an earlier post. The temp sensor diodes on some of the new 45 NM cores get stuck and don't report correct temperatures, but all that you need is for just one sensor to be working to insure safe and accurate temperature measures. I know this because one of mine got stuck and so I went surfing to find out why. It has no bearings on the quality and function of your CPU just go with the highest reading since it will reflect the temperatures of all cores. Look at the uniformity of mort's image.

Mort

Open Coretemp then options and settings, and then under TJunction max offset type in -10 and it will reset your Tj max to 95.

 

thanks Soph, I wish I could do that with everest too.

 

I've looked to see if Everest did it too but I don't use Everest that much so I'm not all that familiar with anything beyond what is clearly labeled on the menu. I know that's it's a good app but so are all the others that are in my arsenal.

 

wait So the highest reading reflects the temp of the others? Isn't my 56C a little hot for stock speed 2.66 GHz? Mort's at only 45C for his highest reading and he has his OCed to 3.4...

 

Mort's working on better cooling, with better cooling equipment so that he can overclock, so he should have better cooler temps. He's also probably used thermal paste more effectively. Your temps aren't extremely negative, and for normal use not even negative, but they could be improved considerably if you plan on overclocking. If not then leave them alone. If so then let's start a discussion about cooling your system.

 

mrk44,
What does it do under full load?? I would be curious to see that if I haven't already.. lol. If Soph is correct then your cores shouldn't go over about 70C, of course that's a guess.

What I don't understand is the reason for the "sticking" of the values. I need to look that up for clarification.

good luck with your "values"...

...gm

 

that is a very good question and I was going to ask that too but forgot.

 

I extrapolated his load temps from his posted idle temps. Look at the screen shot that he posted.




Note that he has his Tjmax set at 105C so you can subtract his idle temps with CoreTemp by 10C to match those that many of us have been posting. If so then he has considerable load temp headroom before running into a concern, as long as he's not overclocking.

What I meant by some cores not posting the correct temperature is that some thermal diodes fail and don't report the correct temperature. This is referred to as a stuck sensor or diode. The important thing is that not more than one diode generally fails on a single CPU. This means that the rest are usually good, and a sure sign of a good diode is by its movement when stressed. Only a single diode is required to measure core temperatures on a multicore CPU since the should have nearly they exact same temperatures.

 

Sophocles: I do want to OC at least to 3 GHz.
I reinstalled my hsf, and no difference, temps are exactly the same at 0% load.
Here are my temps at 100% load:



Subtract 10C from each core cuz my tj max is at 105C.

 

Well there's the 70C I was talking about. lol. I did some reading and found this little thread at TH.. linky I think it helps explain some of what Soph is telling us and "eases" my mind at least on the subject... SO mrk44 you're not the only one with some "strange" readings.

Thanks Soph for your "help" here.

....gm

 

Hmm! All of your cores sensors are now off enough to send alarms. Something is wrong with how your heat sink is seated against your CPU heat spreader. Socket 775 retention clips can sometimes be confusing and even when done right they don't always provide adequate pressure between the heat sink and spreader. Once the clips have been seated to your motherboard there is a counter ( I think, could be clockwise) twist to tighten them up. If your clips aren't providing enough pressure the heat dissipation won't be effective. If your clips aren't in good shape from too many attempts then that can also reduce their effectiveness.

 

Yeah, I think I might be done with these coolers. I'm too lazy for them. I am really not happy with these temps, especially because I was planning on OCing to 3.4 GHz when I started my first build....when I first saw the temps I was getting I lowered my goal to 3.0, and those temps don't satisfy me either. I either want to go with water cooling or this hsf when it comes out.
Does anybody know if that water cooler I posted on the previous page is any good? To all of those who have it, which water coolers are you guys using?

 

One of the things that I did before installing this last water system was to replace the retention clips with some good hardware bolts and nuts. Here is a photo I took not long before I booted for the first time. Note the symmetry. LOL

 

Ok NOW I'm confused because they are still 21C off..... just like they were when he was at IDLE!!! His HIGH is only 70C and that seems within tolerances.. I little higher than I would want to see it but within tolerances none the less. Since he's using coretemp, his distance to TJMax is 35C according to what you said about using the "higher value" for his readings.

I'm learning stuff every day and I'm learning now so some explanation or links would help a lot.. I know, GOOGLE it... lol.

Anyway... I think the "stuck diodes" suxor and the cooler he is using is a bit difficult to get right... jm2c..

....gm

 

On a quad core if one or two cores are off by more than 2C each it is generally not a concern unless the other two cores agree. When all cores are off by 2C or more it suggests that the pressure of the heat sink against the CPU spreader is not even.