German vs. US (sm vs. lg)
 

http://www.gruntville.com/reviews/wc...cape/index.php

sure let pH off the hook, whew

those testa had different room temps, but it was pretty much like I expected it would.

Given that the P4 diode is the temperature measurement that I trust least, but...

Unsure about why they're surprised about the results when run in a "US" system. That the nexxos is tuned for ultra-low-flow with its micro-structure is plainly evident, but it's hardly going to perform like a bag'o'pus at higher flow rates. The results, assuming that they are reproducible on a more exacting testbed, are not too different to what I would have expected, with the micro-structure block's inherent resistance and structure preventing them from "breathing" and stretching their legs when presented with higher flow rates.

Would love to see Phaestus test one. Am even more determined to hunt one of the NeXXoS blocks down and get one to Pheastus (via me).

i think alphacool.de do ship them over to the US
i got the vga block from them, pretty good cooling my 5800@U

Really? More so than Athlons?

I love the sarcasm whenever there is an "important" review done by a bad tester.

This page is where I stop believing it (coincidentally, this is the first page I read.) Once again, the PolarFLO SF beats out the WW. Umm... Only 3-5C? SLK900A with a Tornado would be about the same as those other blocks...

OK, just to review (and maybe I missed something) watercooling in general involves several heat transfers.
1) heat source to block
2) through block to wetted side
3) block to water
4) via water motion to radiator
5) water to wetted side of radiator
6) through radiator walls to fins
7) radiator to air (small amount probably lost via direct radiation)

I don't think we'll see any differences between small and large diameter hosing in 5, 6, or 7. Yes there are "sweet spots" for some rads but I'm ignoring that :)
I also don't think we'll see a difference here in #1. TIM and clamping pressures seem to be the same.

I'm at work and away from my notes, but as I remember, it doesn't take much water flow to transport, say 100W of heat at, say a 4C heat rise, so I'm not sure that 8mm ID tubing is at a great disadvantage here.

So, what's left is:
2) through block to wetted side:
Again, no reason why a smaller diameter hose system has to have a thicker baseplate. The Innovatek Rev, rev 4 and the "pyramidal" one do, but is looks as though this alphacool one does not. IMHO the only reason to have a thick baseplate is for thermal damping, but (again very much IMHO) why not just store sudden heat loads in the water?

...and 3) block to water
So far, Cathar's whitewater and cascade designs appear to do this the best. the greater the water velocity (within reason) the better they transfer heat to water. The larger ID systems seem to all move greater volumes of water than the smaller ID systems (duh!) - and the greater volume gets translated into greater velocity in the water block. BUT I would guess that there's an optimum velocity, so a block "tuned" for lower flow might indeed perform better at low rates than, say, a whitewater at the same low rates.

I'm curious to see a real test as well...

Bob

PS: this is not to say that 8mm ID systems might perform better simply with the substitution of larger diameter hosing - unless the blocks are designed so that water velocity really doesn't matter much (like the Rev 3).

PPS: please tell me where I've gotten any of this wrong. I'm trying to learn about this stuff and getting "slapped down" is actually useful to me :)

The sweet spot is an accident. Higher flow is better for radiators. Flow is restricted by small tubing. Thus, 5 is a concern with smaller tubing. You guess wrong. Well, I suppose that the optimum velocity would be infinite metres per minute, but that's irrelevant. A WW is tuned for a 1048, is it not? Generally, though, something is best for what it's tuned for, so you're right in that aspect.

Athlons are just about as bad as the P4's.

Both are inconsistently calibrated from CPU to CPU. Both, to my experience, under-report the die temperature to a moderate or even extreme degree. Phaestus uses the Athlon diode to success by bypassing any mobo "interference". His results are still specific to his particular CPU though and if he changed CPU, he would need to recalibrate his readings.

With the Athlon diode on one of my CPUs is that it seems largely immovable for moderate heat loads - with any voltage/speed up to 1.75v/2.2GHz resulting in the near exact same readings for pretty much anything, and it's only by really bumping the voltage, MHz, and running BurnK7, can I actually get it to budge by a significant amount, so yeah, not linear at all for that CPU - and it varies from CPU to CPU.

On my P4, I can get it to trip the CPU's automatic critical thermal shutdown (135C as per the calibrated internal non-user-visible diode), while the user-visible diode doesn't read much above 60C.

Sarcasm wasn't intended in the comment - it's just a reality.

Re: radiator "sweet spots". My theory on this is that it is caused by degenerate conditions occurring in the inlet/exchange tanks resulting in certain tubing sections of the radiator being "favored" over others.

there is a company which sell the alphacool waterblock
you think they might lend it for testing?

http://snt-systems.com/catalog/produ...products_id=81