Watercooling, Tjmax and Q6660

[fruees]

Hey guys just a quick Question...

My Rig: Q6660 G0 at 3.3ghz, corsair H5O-1 cooler, Ip35 Pro, Hd4870x2, DOminator.

When I start my pc up, my lowest core is as low as 16 C, max core is 23. After a while the cores will get to around 33 max at idle.

I run small ffts in prime95 and get max 59 C.

Pc idling for 20 mins, max core is 43 and min core is 37 C!! wtf? what makes them jump up 10 C?

My questions are:

1. Is this normal for a watercooler to never get back down to original idle temps, after a stress test?
2. I have the latest rev. of Coretemp. Tjmax is 100 - now I read in one place it should be 90, in another place 95. Please, what should Tjmax be for my CPU and how do I change it?


Thanks so much for the help guys - I have finished a year long build and it's annoying to have a few doubts remaining.

 

[irkjab]

Well, what is your room temp, and what temp does the cpu get to if you just leave it idle for a while after startup?

 

[fruees]

room temp normal - U.k wintertime so the heating is on.

The problem is that even if I leave it for an hr, idle gets to 34C max but as soon as I stress it and then leave it, it stays at 43C Idle (these temps are the max core temps).

Can anyone pls help me with the TJmax question pls? what should it be?

 

[randomizer]

Well I'm no WC expert (heh... I don't even know how to WC) but I assume you're heating up the water over time so that's going to reduce its ability to cool the chip as much as it did when you first turned it on. I find the same thing occurs with air cooling as well.

As for Tj Max, 95-100C is fine. It really doesn't matter what you put it at because the CPU will begin throttling when Distance to Tj Max is approximately 0, not at any specific temperature.

 

[fruees]

Thanks Randomizer, so in terms of safety, TJ max will be fine either way.

But in terms of accuratley measuring the temp? Am I right in thinking that if I put TJmax down to 95 from 100 then the temps that I'm seeing in coretemp will drop 5 degrees?

I thought that my temps would be caused by the water gradually heating up, but I didn't think that it would be that extreme!

 

[randomizer]

Any change in the Tj Max assumed by the program will cause a change in reported temperature by that exact amount. Temp is calculated using a simple formula:

Core temp = Tj Max - DTS reading

Where the DTS reading ranges from 127 to 0, decreasing as the temperature approaches the actual (unknown) Tj Max (the DTS reports the distance to Tj Max).

 

[fruees]

Ah yes thanks so much for that Randomizer. Hence I believe I am right in saying that the real tmp of your CPU is impossible to get without a thermal probe in the middle, and everything else is just educated guesswork?

So when people report their temps, it's pointless paying them any notice because you don't know what their Tjmax is? (Could be + or - 5 C)
Mine is at 100, is this the same for any other Q6600 owners?

 

[randomizer]

Not only that, the DTS itself is imperfect and only really accurate near Tj Max. There is an unknown amount of error that can be somewhat corrected with some utilities which increases as the chip cools. Additionally, the actual Tj Max varies from chip to chip and from core to core in the same chip. With this many variables attempting to get accurate temperatures isn't practical and isn't what the DTS was designed for.

Core temps can give an indication of poor heatsink mounting, as can simply reading the distance to Tj Max (which should be kept high, not low). The latter will be the same regardless of what software you use. If you are idling with a distance to Tj Max of less than 40, chances are your heatsink is mounted poorly.

 

[fruees]

Wow, and here I was worrying about 5C difference! So essentially, there is no point worrying about slight temp increases/decreases from heatsink to heatsink because no-one knows what the real temp is!

Just to pick up on your point, you think I should keep my tjmax at 100, instead of putting it down to 95 to increase bragging rights and make me feel better about my hardware ( including mudddying a friends claims of 33C idle on a Q6660 at 3ghz with a Dark Knight, as opposeed to my higher temps with a supposedly superior cooler)?

 

[randomizer]

Wow, and here I was worrying about 5C difference! So essentially, there is no point worrying about slight temp increases/decreases from heatsink to heatsink because no-one knows what the real temp is!

Assuming everything else stays the same, including the CPU used and the ambient temp, differences in temperature when changing the heatsink does matter. While you don't know exactly how much it has changed by exactly (if the DTS changes by 1 it doesn't necessarily mean the temp changed by 1C due to inaccuracies), an increase or decrease in temp outside of margin of variation from ambient/load fluctuations is still worth noting.

The precise temp is not so important. Temperature deltas are more valuable, but still subject to many variables.

Just to pick up on your point, you think I should keep my tjmax at 100, instead of putting it down to 95 to increase bragging rights and make me feel better about my hardware ( including mudddying a friends claims of 33C idle on a Q6660 at 3ghz with a Dark Knight, as opposeed to my higher temps with a supposedly superior cooler)?

Why not try 97C for a compromise?

 

[Shadow703793]

I just looked at your temps (which is high for watercooling)....and then I saw that you aren't running REAL water cooling. The H50 is NOT real water cooling! The H50 is more or less bested by a high end cooler like a TRUE 120 or a Core Contact Freezer 120mm. I'm not surprised your getting those temps (59C).

As for temps, go with RealTemp. According to CompuTronix it's the best of the breed for the C2D/C2Q CPUs.

including mudddying a friends claims of 33C idle on a Q6660 at 3ghz with a Dark Knight,

FYI: I ran a Q6600 @3.3Ghz with a S1283 + 110CFM Delta fans + SpeedStep enabled, I ran at 30-35C at idle.

FYI: The H50 is....
1. Not real watercooling
2. Not better than a high end air cooler (ie TRUE 120 or a S1283 + better fans)
3. H50 dosen't have enough rad for high OC esp. on i7s.
4. Never meant to compete with real watercooling

Btw, for REAL watercooling info look here: http://www.tomshardware.com/forum/253958-29-watercooling-guide
Real watercooling will cost you $200-250 for just the CPU using top of the line stuff.

 

[fruees]

Shadow, I know I don't have real watercooling, but having researched for a while It seemed that The H50 was better than any air cooler, except for the new Noctua which is ridiculously big!!

I thought that my load temps were good actually, just concerned by my idle temps. Shadow, do you find that after testing o load, your temps increase at idle due to the hotter water? Also what was your tjmax?

Thanks so much for the help so far lads!