"Overclockers.com" vs "Bill Adams" WB Test Results

[ILikeMy240sx]

Well, I think JoeC takes that into consideration. If you look at his test bed its pretty nice. It tries to have the same mounting each time.

 

[JFettig]

The change could be as simple as using different thermal goo.

Jon

 

[Smirabi]

to reply to something nikhsub1 said, "The goal is to isolate the blocks performance."

because the block's performance is a function of what seems like a million other things (motherboard model, die area, "ambient" water temp, ambient room temp, type of tubing, pump-related factors.... et cetera) how can you truly ever isolate a the proportion of the variance in C/W (and also temps) created by the block and only the block?

i have an answer. it's called regression analyses, a statistic used a lot by me in my psychology grad-schooling. and i'm working on how i can correctly apply it to this problem, and the overall problem of cpu-cooling. but until someone (hopefully me) creates a regression equation that can account for almost all of the variance in cpu temp (or in this case, waterblock C/W) then all we can do is hope that the tests reported by BillA and JoeC will generalize to our setup.

or does someone see something i dont that makes me way wrong?

 

[Diggrr]

No, what happens is that the throttling valve drops the extra pressure from the bigger pump. All other components behave exactly the same, dropping the pressure according to the flow rate.

What exactly is the type of valve used?
I'm using experience with hydraulic and pneumatic machinery, but a normal flow control valve will drop flow rate and pressure only until a second obstruction is met.
Say you had a Swiftec (open design) and a WW (more restrictive).
The flowrate may be set the same on the valve, but the pressure exerted on the water between the valve and the Swifty would be lower than that between the valve and the WW.
That added pressure would be giving the WW an advantage in testing because the higher pressure would increase the flow to a point [possibly] higher than it might see with the actual pump that the valve is set to simulate.

An Iwaki may well be able to exert far more pressure (far being relative to pumps) in this zone of tubing than an actual Eheim, even though the valve was set to what an Eheim might be rated for.

See what I mean?

 

[Caffinehog]

Re: Re: Re: "Overclockers.com" vs "Bill Adams" WB Test Results

Not if the testing is done properly Caffinehog. Bill uses a 'standard' system typical of most H20 systems? Have you ever seen a picture of his situp? Looks like a mad scientist lab! AFAIK, BillA has the most comprehensive test bench out there for testing water blocks. If the C/W in ANY test is based on anything EXCEPT the block itself, then that test is no good in my book. The goal is to isolate the blocks performance.

The point is, Bill tests things with a consistent setup. Other reviewer's setups will vary from this. Therefore, comparing two blocks reviewed by Bill is like comparing apples to apples. Comparing one of Bill's reviews to somebody else's review is comparing apples to oranges.

 

[nikhsub1]

Re: Re: Re: Re: "Overclockers.com" vs "Bill Adams" WB Test Results

The point is, Bill tests things with a consistent setup. Other reviewer's setups will vary from this. Therefore, comparing two blocks reviewed by Bill is like comparing apples to apples. Comparing one of Bill's reviews to somebody else's review is comparing apples to oranges.

I completely agree. That is not at all what you said in your first post though...

 

[SOShootME]

I believe Joe said he has has been in contact with BillA while he was putting his test bed together. Joe is trying to develop a testbed that will allow him to provide consistent results across all blocks that he personally tests. That will give us meaningful comparison numbers from block to block within Joe's data set. BillA's test results are another reference set.

Unless BillA and Joe use identical test beds in identical envoronments , you can't expect their numbers to match. As Diggr and others have mentioned, there are just too many variables.

 

[eobard]

Hi everyone, I'm still thread monitoring, I do it alot, and in this case I don't see anything that needs "dealing with". Thanks everyone. [/detour end]

 

[bigben2k]

What exactly is the type of valve used?
I'm using experience with hydraulic and pneumatic machinery, but a normal flow control valve will drop flow rate and pressure only until a second obstruction is met.
Say you had a Swiftec (open design) and a WW (more restrictive).
The flowrate may be set the same on the valve, but the pressure exerted on the water between the valve and the Swifty would be lower than that between the valve and the WW.
That added pressure would be giving the WW an advantage in testing because the higher pressure would increase the flow to a point [possibly] higher than it might see with the actual pump that the valve is set to simulate.

An Iwaki may well be able to exert far more pressure (far being relative to pumps) in this zone of tubing than an actual Eheim, even though the valve was set to what an Eheim might be rated for.

See what I mean?

...but you're missing the point! The pressure drop of a block is a direct result of the flow rate going through it. Nothing else can affect it.

The only difference that may exist in terms of pressure, is the overall level.

Let me try this example (numbers fictitious):
Let's say we have a big pump that provides 5 psi of pressure at a given flow rate. The block will drop 2psi, the chiller drops 1 and the throttling valve drops 2.

Now swap the block for something that's nowhere near as restrictive (aka a Swiftech open plate).

The block drops 0.5, the chiller drops 1, and the valve is set to drop 3.5. The flow rate is still the same.

Can you dig it?


The valve that Bill used (an array of valves actually) is a needle valve. I really don't know what JoeC uses. It's really hard to set the flow rate to something specific, with a single large design valve.



As for the difference in the results, everyone has to remember that a "C/W" depends on a couple of things: a measurement of "C" and a measurement of "W".

Measuring "W" is relatively easy, with the right tools (volt and amp meter). Measuring "C" is where all the difference lies:

If I build a testbench (and I am, really!), I have to decide where within the heat die, I'm going to position my temperature probe. The slightest difference between my position and any other tester's position is going to make a world of difference.

So C/W are unique to a testbench, see?" (Then you have to trust that the test bench and the procedure stays the same!)


In order for a C/W to become "universal" and "transferable", the results would have to be adjusted to match the temperature that an actual CPU would indicate. The problem is that it's really not easy to do! We all know that temps reported by motherboards can vary by 10 deg C, just on a whim.

So with that in mind, I created the WBTA (Water Block Testing Alliance), www.wbta.us , of which JoeC is now a member, as a place where all water block testers can come together, and try to get something a little more universal "out there".

(The Forums aren't up yet, but it's coming!)

pHaestus has done some great work in trying to read temperatures directly from the CPU, by tapping directly into the thermal diode of an AMD processor.

 

[BillA]

sorry Ben, wrong again

"Measuring "W" is relatively easy, with the right tools (volt and amp meter)."
- this is a measurement of the power applied to the heating element, not the heat input into the wb or hsf at all

be cool

 

[E_Man]

If its not already on the site, i would greatly appreciate if with the C/W results could contain an info box neer them that says the die size used, and any other key factors.

 

[bigben2k]

sorry Ben, wrong again

"Measuring "W" is relatively easy, with the right tools (volt and amp meter)."
- this is a measurement of the power applied to the heating element, not the heat input into the wb or hsf at all

be cool

True, very true.

There's a secondary loss that's very hard to quantify, somewhere between the heat die and the block.

 

[Diggrr]

Well, being the sceptical person that I am, I dibbed on some flow meters of squire Iron Hawk's to see for myself.
Nothing personal, but I'm a "show me" kind of guy, and I've wanted one for ages anyway.

By the way Bill, none of this is to say that your testing was flawed. Being the first, and only quantitative tester around made your results very much needed, and I thank you. I am sure all your results are comarable to one another, without doubt, but just wonder at the differences now that others are coming out with differing results.
I feel it's healthy to wonder, and I've got curiosity by the bucketfull.

 

[BillA]

Diggrr
you cannot imagine the 'lessons' you will get with flow meters, have fun

the 'testing scene' is progressing nicely, if in slightly different directions
JoeC's bench seems pretty good, though his "C/W" measurements will always be difficult to compare with those of others due to the single values always being at a different flow rate

pHaestus' progress can be seen here
while he is still missing a couple of pieces of equip; his results, I believe, will be generated in the same manner as mine

the "C/W"s of each tester will never be directly comparable, rather they will be offset by 'x' or 'y' amount, up or down
and some will be more compressed than others

but the wbs' relationships will be the same across the board

E_Man
my heat die is 100mm², JoeC's is 140mm², pHaestus' I don,t know

be cool