LiquidNinjas Waterblock Round-Up 2003

[Player0]

Please post your comments here.

[mdzcpa]

Good job Tom:)

[nikhsub1]

Tom, I must applaud you for your efforts! You are a braver man than me. Anyway, a few of your charts look a bit off. The 2 high flow charts with the varying heatloads, 1 at 2Ghz at 1.65V and the other one at 2.4Ghz at 1.85V. C/W's scale linearly in regards to changing heat loads. If 'block x' has a C/W of say .20 at the current flowrate, it will have a C/W of .20 no matter if the heat load is 30W or 1000W at the same flowrate. Your graphs certainly do not represent that and I would assume it comes down to the equipment you had available for the test. Here is yet another BillA graph to demonstrate my point:

http://overclockers.com/articles655/tempwattsWB75.gif

Very good effort though! My last parting wish was that a WW was in the review. You could add an addendum to the test, I will send you mine if you say the word.

[jimmyang]

Nice :D

I thoroughly enjoyed reading it and I even recommended it to some of my friends who are new to watercooling.

They find it easier to understand than most techie sites. This is a great article and an applaud for you Tom for going through all the trouble to get the review this far.

[WesM63]

Hey tom,
Congrats on finally getting that beast out :D It was well tested and I enjoyed reading it.

[ralf_c]

well done mate:thumbsup:

[Bellarine]

If 'block x' has a C/W of say .20 at the current flowrate, it will have a C/W of .20 no matter if the heat load is 30W or 1000W at the same flowrate.

Um. The chart you posted there doesn't support that theory. Picking any flow rate line, I can see that the C/W most DEFINITELY changes depending on heat input.

That makes sense, because as heat increases, it spreads to a larger surface area in side the block. As long as the block is designed to cool that larger surface area, it's C/W should decrease as heat input increases (ie, performs better under load).

This seems to be the case especially with waterblocks that do well with peltiers, such as the AquaJoe and the MCW5002.

Unless you were trying to say something different? But if you look at the graph you posted, and measure it, you can see that the C/W most definitely DOES change under heat load. From almost .9 at the beginning to .5 at the end.

Of course, that change greatly depends on the design of the block.

[Bellarine]

Oops, I seem to have posted with my wife's account again ;)

[Player0]

Just FYI, that was me up there:

[nikhsub1]

Originally posted by Bellarine
If 'block x' has a C/W of say .20 at the current flowrate, it will have a C/W of .20 no matter if the heat load is 30W or 1000W at the same flowrate.

Um. The chart you posted there doesn't support that theory. Picking any flow rate line, I can see that the C/W most DEFINITELY changes depending on heat input.

That makes sense, because as heat increases, it spreads to a larger surface area in side the block. As long as the block is designed to cool that larger surface area, it's C/W should decrease as heat input increases (ie, performs better under load).

This seems to be the case especially with waterblocks that do well with peltiers, such as the AquaJoe and the MCW5002.

Unless you were trying to say something different? But if you look at the graph you posted, and measure it, you can see that the C/W most definitely DOES change under heat load. From almost .9 at the beginning to .5 at the end.

Of course, that change greatly depends on the design of the block.
The C/W of a block is fixed, it never changes. As heatloads increase, so does deltaT, however, the C/W ALWAYS remains the same at any heatload and scales linearly. You should know this.

[jimmyang]

Originally posted by nikhsub1
The C/W of a block is fixed, it never changes. As heatloads increase, so does deltaT, however, the C/W ALWAYS remains the same at any heatload and scales linearly. You should know this.

:D

[Player0]

Originally posted by nikhsub1
The C/W of a block is fixed, it never changes. As heatloads increase, so does deltaT, however, the C/W ALWAYS remains the same at any heatload and scales linearly. You should know this.

The graph you just posted shows C/W changing. Please explain that.

[jimmyang]

edit* nvm ... care to explain more nikh?

[Player0]

Pick the blue line, which is a constant flow rate.

The first data is about 37/40 = 0.925 c/w

The last data on that curve is about 65/130 = 0.5 c/w

Looks like a big difference to me.

[Player0]

"You should know this"

I really hate that elitist attitude people get. No, I actually dont know either way on this. Ive discussed it before, but I dont actually remember which way is right now. Excuse me for forgetting.

It doesnt change anything, me knowing or not knowing. I've asked some people in the know and Ill see what they have to say. In the mean time, your graphic clearly supports MY side of the arguement.

[nikhsub1]

Tom, don't get me wrong. Not an attack, but I KNOW you know what you are doing. I know you have experience in this. My point of the graph was not to pick C/W's but to illustrate how linearly a block will scale as heatloads increase. Your numbers cleary do not show this. Take the White Water (or any block that BillA has tested) which has a C/W of .19 at 4lpm. That C/W of .19 will remain constant with the same flowrate no matter the heatload. A quote from BillA:

"A good wb, or a poor one, is such at any and every applied heat load (for the same sized heat source)"

What your tests should have shown was the difference in relative performace between the blocks INCREASING as the heatload is increased (at the same flowrate). Yours show something different and that is all I wanted to point out.

[jimmyang]

I am very interested in this subject but perhaps this will shed some light...

Linky (http://thermal-management-testing.com/waterblock_test_results.htm);)

Beers for the whole thread, on me!
:drinky: :drinky: :drinky: :drinky: :drinky: :drinky: :drinky: :drinky: :drinky: :drinky:

[Player0]

Nikh, if you've read my review, you'll notice that I said the exact same thing there. It's no surprise that people are going to come here and try and 'correct' me about things that I've already noticed and mentioned in my review. If anything, my data supports that there IS error in this kind of testing.

What I am trying to understand is this graphic. Yes, I've read the thermal-management-testing.com, and there it sais that c/w remains constant under different thermal conditions. Yet the chart show's different c/w ratings occuring at different loads.

Im not saying its wrong, Im just looking to further my understanding about WHY its right.

[nikhsub1]

Originally posted by jimmyang
I am very interested in this subject but perhaps this will shed some light...

Linky (http://thermal-management-testing.com/waterblock_test_results.htm);)

Beers for the whole thread, on me!
:drinky: :drinky: :drinky: :drinky: :drinky: :drinky: :drinky: :drinky: :drinky: :drinky:
And exactly shows my point (thanks for that link, forgot to look there... a quote:

"conclusively demonstrates that the "C/W" of a wb is unaffected by the applied power (so long as the die size is the same). Many wbs have been tested, and they all respond quite the same.

There is no such thing as a 'good' low power wb being relatively worse at higher heat loads. And yet another WCing fable bites the dust."

And the graph that plots the exact same C/W for varying heatloads:

http://thermal-management-testing.com/CWpwrWB75.gif

[Player0]

And here's hoping HardOCP doesn't take down my server again.

[BillA]

from that article,
Chart 7 shown above is of the die temp vs. power, NOT the C/W
but for the die temp to be linear the C/W must be constant
-> see Chart 10
http://thermal-management-testing.com/coolant%20temps.gif
the various 'curves' explained in the article

be cool

[Player0]

Ah, okay, thanks Bill, that makes more sense to me :)

[nikhsub1]

Originally posted by Player0
Yet the chart show's different c/w ratings occuring at different loads.

Im not saying its wrong, Im just looking to further my understanding about WHY its right.
I did read the review, and noted what you said. I just wanted to clearly point out exactly what the numbers should have looked like is all. The first graph is NOT to pick C/W's out of, but merely showing how linearly the blocks will scale. Why would [H] take the server down? Are you hosted through them?

[Player0]

The server locked up because its under tremendous load right now from review traffic, mostly from Hard im sure. We got the windows open, hopefully the 30F weather outside will keep us up. Probably break the traffic record today.

Hopefully we get that AMD dually server here soon. No, we host our own server.

[enderox]

Congrats on such a mega review.
A lot of blocks compared. A lot of time involved in testing and comparing.

The only point I wanted to make is that one has to take everything is this review with a grain of salt.
The comparisons are valid for his setup and his setup only.

That is one mega pump putting out 50w of its own heat.
I therefore think that one cannot assume that performance you'll get with a low flow rate pump (with the advantage of low heat output) will be anything like the restricted flow tests.

Cathar has seen little difference in using the Eheim 1250 vs the Iwaki pumps. The higher flow rates are negated by the extra heat output by the pump itself.

food for thought

Mike...

[Player0]

Yes, thats a good point about the pump, although I dont beleive the pump is putting 50w of heat in to the water. It's not submerged, and only the impeller has contact with the water in the pump. Yes, the impeller shaft does require water, but the water does not flow in and out of this shaft very much. The entrance for water here is only the size of a pin hole.

Im sure that the pump will apply more heat to the system than a smaller pump, but to what degree is very up in the air. I personally wouldnt think it would have a great deal of effect on water temperatures, especially with the size of my radiator. But I wouldnt say for sure, so, its a point well taken.

[nikhsub1]

Good point enderox. I agree with Player0 that 50W is not all going to the loop since it is inline. However, the amount of pump heat dumped into the loop will vary by how much restriction there is in the loop and the only changed component of the test was the block so... I doubt it makes much of a difference.

[Player0]

Just because a lot of people have been bringing this up, here is an important except from the review:

After a three month ordeal of obtaining review samples, putting together my new test bed, trying to get a good flow meter working, testing all the blocks and retesting them, and finally coming here to make sense of it all, the first thing I notice is the absolute limit that this kind of waterblock testing has reached. I would like to conclude that my results seem to be rough at best, even though great care was taken in to preparing the samples. The SlitEdge block performed poorly on my AMD system no matter what the circumstance. And I spent a good deal of time and effort trying to improve it's performance on my testbed without any luck. This block has been tested by others and has gotten much better marks. Noticing how the different blocks seem to react when I increased the wattage by a small amount, and the performance rankings seem to change dramatically shows me that there is more going on behind the scenes than I am able to account for. How the blocks would react on a P4 system, or with differing flow conditions, I wouldn't be able to even venture more than a rough guess from the performance in this review.

Yes, I realize how jumbled the numbers are, and yes, I know that its the testing system I used that is to blame. The limit of good results that can be acheived on such a simple testing system has definitely been reached at this point, and theres no further reason to continue testing on a system like this, where you can't account for all the variables.

I'll be working on a new test bed shortly, something a heck of a lot more refined.

[killernoodle]

So.... are ya gonna sell the blocks? :D

[Player0]

Uhm. Most of them are sold. I think I have to send the SlitEdge back. The RBXes and MCW5002 are spoken for. I unloaded the cascade and maze4 last week after testing. That leaves the AquaJoe and the Thermochill, which ill seel if anyone is interested.

[Mike Donatello]

First post from a long-time lurker. ;)

Your review was great, and comes during Week 3 of my "Great Waterblock Info Quest." I'm planning on moving to a water-cooled system in the next month or so and am totally on the fence RE: my setup. The review was a great help in narrowing down my list of choices. Thanks!

[Drake]

:D:DWELCOME:D:D Mike!

Glad ya came outta lurking ;)

[mdzcpa]

Welcome:)

For some reason I think I recognize you from somewhere. Mike Donatello sounds very familiar. AMDMB? Abit forums?

[Mike Donatello]

Thanks guys.

ABIT, ABXZone, XtremeSystems -- anyplace I can get good advice to keep from frying my stuff! Never played with water before and -- after frying an IC7-G last summer attempting a volt-mod -- I'm adopting a "slow and steady" approach to revamping my setup. Haven't even decided on which new case yet...



Originally posted by mdzcpa
Welcome:)

For some reason I think I recognize you from somewhere. Mike Donatello sounds very familiar. AMDMB? Abit forums?