Here Here!!! 100% agree on the cost issue!

If you ask me, DangerDen should be performing these tests, so should resellers. It certainly would be in their interest.

Of course we'd still need impartial testers... like you pHaestus (sorry about the PH thing)!

About the temp probes, and accuracy:
My point was that since the rigs out there are all going to be slightly different, and if the variation in those results range in the +/- 3 degrees (as an example), given the exact same equipment, then the temp probes you're using right now would do just fine. We just don't have those numbers. What we have, is most people reporting their results within a +/- 1C, with few reporting at +/- 0.1C, so definitely, you would need at least +/- 1C accuracy.

The next step would be for you to go through the exercise of determining the sources of error in the testing conditions, and equating them in terms of a percentage. If for example, you found that a change in humidity from 20% to 80% (95% plus is common here in Houston, Texas) affects results by .3 degrees, then that's the accuracy of the test: 0.3 degrees. So you get a temp probe that measures +/- .1 degrees, and state your results with that accuracy. If ambiant temp also affect results, then the margin of error is ADDED to the previous one.

CPU compression is another one of those "beyond my control" variables. I think that the best thing to do is to just describe what the WB comes with, and specifically describe how you set it up (i.e. spring screws turned way down, then turned back 1/2 turn).

You've mentionned the pump induced heat many times, but I've been ignoring you! It's a good case for controlled environment testing, like you said.

Yes, the user wants to know what's best, but what's best for one, may not be best for another. If there's one thing that I found to be consistant, is that everyone is different! We have people in this forum from all over Europe, and they certainly don't have access to the same supplies that we do.

I do appreciate a lot of what you guys are saying, I just don't think the majority of people out there will really care if the temp is 0.1C off or not in your tests. Let alone care about the effects of Humidity (I dont even know what the humidity is in my area, and I would imagine that most other people don't either). The fact that you break this down to the nitty gritty detail, is great (whatever floats your boat I suppose). However, if after you completed your testing, and someone else did a test and was say only 1-2C off in there temp compared to yours (but their methods were more in line with what I proposed) do you think the majority of comsumers would really care?

To answer some of your questions bigben2k:

1. Will these tests be expensive? Perhaps, but I would imagine that someone from one of the major OC/Performance/Cooling websites (Birrman54, pHaestus, etc) will be able to get the equipment (WB's, Rad's, Pump's, etc) for free (just like people get Motherboards, Ram, CPU's, etc) for testing purposes (probably have to return them afterward though:cry: )

2. People will always have different gear (that's the name of the game with PC's), but if you had a baseline test system, you could just swap in a different CPU (and new Motherboard if needed) and show the difference with the newer CPU. So if you started with an AthlonXP, you could swap that out for a T-Bred, test it with your base system. This would give people a good comparison of what to expect with this new CPU. About the only time this wouldn't work out as well, is if you went from say an Athlon to a P4 Northwood (then you would have a different WB to test as well as the different CPU).

3. Most testing I have seen done has included a CPU under Idle and Load condititions.

4. I don't ignore the numbers in the tests I've read about online. My point was, that you don't need to read the numbers on 15 different sites, to see that one particular Heatsink is the best (in a varity of setups and systems). Your second comment is what I have been trying to get across. "...the best of each will not yield the best results" pHaetus is only testing WB's, so his test might show a WB performing better than all the rest, but when you put it in a system (realword not some lab), the same WB might suck.

5. I agree about the Rads, and I would even suggest that larger Rads perform better on highflow blocks, while smaller Rads perform better on lowflow blocks (course haven't seen any test on this yet).

6. Different rigs might change the numbers a bit, but you are correct, you could get the same results with some completely different Rig.

7. My points on tubing size have to do with the fact that you see numerous posts about people having 5/8 OD tubing (vs. say 3/8 OD tubing) and this gave them 1-3C better performance. Again, this would help people decide if they really needed the big 5/8 tube or could they use the smaller 3/8 tube and still have acceptable performance. I myself perfer the 3/8 OD tubing as it takes up less space, and is easier to work with (bendable, easier to poke it in out of the way places, etc). I agree with the 0.1C temp testing, it really isn't necessary for the Majority of people.

I think we're starting to see along the same lines... or at least understand our differences! Thanks for your patience PowerHouse, it's really nice to have this civilized conversation.

Lab vs realworld:
I think that the biggest objection to using a PC for testing, is that the cooling rig changes from tester to tester (i.e. different rad, different pump, etc). Most reviews of WB I've seen are individual reviews, not waterblock roundups.

In waterblock roundups (and I've only seen maybe 3), the test rig is of course the same. The problem is, the rig may not be optimal for each waterblock (which I think is your complaint, perfectly valid).

The other problem, and by far the biggest one, is that new products keep coming out. Now, I don't know about you, but I'd rather test a new individual component, under controlled conditions, find out its optimal running conditions, and see how it could be matched up with the other items I would have already tested, instead of running the whole set of combinations all over again.

to answer a few of your points...

1-If the equipment has to be returned, then it's not possible to test again under the same conditions.

2-Again, a very extensive set of combinations to test.

3-Agreed, but they don't give me much usefull information, not even a Best OF", in some cases.

4-Of course you ignore the numbers. You ignore them because you know that there is no way that your results will be exactly the same. You use the numbers to see who came out on top, but that's the extent of it.

5-Yes, high flow blocks do better with large rads, but no, low-flow blocks do NOT perform better with small rads, actually, they are more likely to perform better with a large rad too!

The lab, has to be setup in a way to simulate a PC. The point is not that a lab is different than a PC (yet it is). If it's any easier, think of the lab as another PC, that will give exactly the same results, no matter how often you run the same test.

More importantly, the lab gives one the ability to test individual components, again, with consistant results.

The purpose of testing individual components, is to give you the ability to match different components, and get optimal results.

Here's an example: The hottest radiator right now is the Black Ice Extreme (BIX) yet if people knew that a $20 heater core, picked up at any auto parts store will perform just as well, if not better than the BIX, then people would stop wasting $50+ on it, unless they just wanted something that looks pretty, and is made to fit very nicely in their case (my personal opinion).

6- (agreed)

7-Tubing size has a number of effects, under a number of different circumstances.
a)a larger tube will increase the water mass of the system, thereby increasing the thermal capacity (very often ignored)
b)a larger tube will REDUCE the flow resistance, but bear in mind that unless the internal pressure has been reduced to under about 5 psi, there will be no noticeable increase in flow. This can be measured as a function of flow rate (i.e. water travels at a smaller speed in a larger tube, given the same flow rate)
c) tubing size should be at least as big as the smallest opening of any of the components. (i.e. if you use 3/8 where all your components are 1/2, then the tubing MAY become a significant factor in flow rate.
d)The biggest flow restrictor, is by far the waterblock. It includes at least 2 90 degree bends, as water comes in and out of it, (an additional 90 degree if you use 90deg fittings right on the waterblock) and usually has at least 6 90 deg bends inside. Also, sometimes the water channels are not straight, are not smooth, have intrusions that cause turbulence, etc... all of which are good for cooling, but bad for flow.

A smaller tube (i.e. 3/8) is physically smaller, and does occupy less space, but I don't believe that in most setups, using a slightly larger hose is actually an inconvenience.

A m8 of mine pointed me in the direction of this article http://www.amdmb.com/article-display.php?ArticleID=179
Which looks into this exact issue. His conclusions get back to the old issue about ballancing the componants and properties are of greater value than simply increasing waterflow with bigger tubing or pumps.

not all blocks have that many 90 degree bends. the spir@l has none (within the block) for example. the Maxe2 doesnt have any (all the bends looks greater than 90 degrees, except for 2 that look close...).

anyhow........damn it took me forever to read this thread!

also, ive been wondering why no-1 has made a high-flow block with lots of surface area. theroretically, a block with higher surface area should transfer heat to the water a lot better. for example, make a block with the inlet on one end, and an outlet on the opposite end. .have like a diamond-shaped island in the middle. have a bunch of vertical fins on the sides of that diamond. id draw up something, but i dont have any imaging software (MSpaint wont work becuase it doesnt save in .jpg format).

CD, you're right on many points, but there are blocks of that design being made, by individuals!

A spiral may not have 90deg bends, but it does have a resistance to flow. I referred to 90 degree bends because they're an easy way to calculate flow restriction. Spirals are a whole different ballgame...

Some of the most ingenious designs call for posts coming up from the center of the water block (inside). They offer more surface area that way.

Their designs are limited by the tool though, so you can forget about the finned, diamond island. (nice idea though).

Thanks for the link TiTch, I'm glad to see some more people come around (to the dark side?!?)

For bigben2k:

Agreed, glad to see this doesn't turn into some kind of flamewar. I like good discussions (even if I end up being wrong, at least I can learn something from it).

If I had my choice, I would setup the following Systems and conduct these tests...

- One AMD (AthlonXP) and one P4 (Northwood) so everyone could be happy.

- Roundup the leading blocks (probably about a dozen or so I could think of, but more would show up I'm sure). Same with the Radiators, and Pumps.

- Miscellaneous items might include a couple Reservoir's (some pumps would no doubt be inline as well but there are reservoirs for the Eheim pumps too). As well as different size tubing (for the different blocks and Rad's. as well as some elbows, Y's, T's and Valves.

- Some Mid-Full tower case, but remove all the case fans except the one that is in the powersupply (of course most people have at least one case fan in their systems, so maybe leave one in:) ).

- You would of course need the hottest CPU's of the moment (AthlonXP 2100+, P4 2.53 standard speed in one test, overclocked in the other). I suppose you could use some kind of CPU simulator but I perfer the real thing (plus to simulate the XP you would have to have a device with the same dimensions as the CPU core to have a valid test).

- Once you put together the system (just like you would if you were putting the system in your case, but again everyone's case is different so this could have an impact on the tests as well, but more on this in a moment). Install your ideal of the perfect system, this would be your baseline system (Maze3, BIX, Eheim 1250 pump perhaps), which would probably change after you finished your testing depending upon your results (by change I would mean that you would probably use the best components after you tested everything as your base system - Just like most people would use the SK6 as their base system when testing new aircoolers).

- Testing could include recording the temps at Idle, with a moderate load (Websurfing, gaming for an hour, etc), and with a full load (Folding@home, Prime95, etc).

- First test would be the CPU blocks, swap out each block after the test (of course after a cool down period, etc so the tests are even). You would also record all the pertinent details before and during each test (room temp, water temp, cpu temp, etc).

- the second part of the tests would involve swaping out the Radiators. Instead of a BIX use a heater core, and perform the tests on each block again.

- The next test might involve swaping the pumps (perhaps have a set of low, med, and high volume pumps) and redo the CPU, and then the radiator tests).

- What about the variables in the test you ask, like tubing size/length, and number of 90degree bends. Well you could further separate the tests into two sections, one group with a small tube (3/8 OD) the other with a large size tube (5/8 OD). As for the Elbows, perhaps use a minimum on the initial tests, and after you finish and have your new Baseline system, you could show the effect of adding more elbows to a system. You would also be able to test the effects of systems with reservoirs as opposed to those without (and wether it makes a difference performance wise or not).


Ok, if you read this far (and I probably missed something) what you may notice is that there will be a ton of tests (hundreds depending on the amount of equipment you have to test). So you might want to just test 6 or 8 blocks at a time with maybe 3 or 4 different Rad's and 3 or 4 pumps. As you would finish the tests, you would have a clearer picture of which grouping of equipment performed the best (so you would have a baseline system for your next set of tests, a so called high water mark to beat).

Eventually you would have a set of equipment that through this testing performs the best, you would also have tons of details that would set this review apart from the rest on the net. You could further break it down like the best 3/8 OD setup, the best 5/8 OD setup, the best of the smallest Rad's and biggest Rad's, as well as which blocks performed best with each setup.

What would be funny (unless you were the one to do all this testing:cry: ), would be that if the difference between the best of each category, was only 1-2degrees.

Anyway, I have probably driven lots of people away from this particular topic with all the reading, but I do appreciate all the feedback, and opinions.

I bet you will be sorely disappointed in my waterblock testing then because there will be:

No computer running in a case
No multiple radiators
No different tubing/fitting sizes
No hottest running CPUs (I need a ceramic base AMD with Palomino core to validate my diode numbers and the best I can muster is ~ 75W)

I remain unconvinced that there is any real purpose other than busy work to the majority of these studies that you want to see, Powerhouse. If you vary the flow rate (and thus the pressure capabilities of your pump) then you simulate the effects of different pumps and different rad and fitting combinations as they relate to waterblock performance (water velocity vs. pressure drop is all that matters). There is no need for anything else. The radiator is for now a necessary evil (I have no recirculating water bath) so I will use the biggest best heatercore I can find.

You have spent several thousand on equipment: 2 motherboards and high speed processors, a case, all these radiators, fittings, tubing, pumps, waterblocks, etc, and then you will learn what? Using typical overclocker temperature monitoring gear you would need a temperature difference of more than 4.5C to be valid when rating performance (2 thermistors with 3C accuracy so the error propagation goes like (3^2+3^2)^1/2 or square root of 18 for relative error in measurement). It would take several months to do the testing that you want to see, and in the end you would get a skewed vision of the true performance of the blocks because you have inextricably tied them into the pump and radiator and overall system.

What you are proposing is fine if you want to test complete kits available on the web, but it will do nothing to improve the overall knowledge level of the reader or the advance the quality of waterblock design. Idle hands are the devil's work, however.

some comments from someone who has bought a lot of test equipment, done a bunch of testing, and then upgraded the equipment to start the cycle again, several times - and it is ongoing

I would suggest that the goal be clearly identified, and the experiment designed accordingly
the methodology used must yield data that answer the experimental goal
and the equipment used must have the capability of producing accurate and repeatable results

I state the obvious as it is apparent that some are not familiar with the 'design of experiments'
if one wishes to quantify a system's performance, then that is what is tested
if however one wishes to quantify the capability of a component, then the individual unit should be tested in isolation from the effects of limitations imposed by other parts of the system
-> this is the purpose of bench testing
(a simplistic analogy would be running an engine on a dyno vs. running the car on the track)

given that it is desired to quantify the capability of a component, what are the appropriate tests ?
since wbs are the topic of the day, and several are reported to be starting test programs, lets consider how such a program could be setup

the first step is to identify those use factors impacting wb performance
then to approximate the range of those conditions
then to identify the equipment necessary to create that range of conditions
then to identify the instrumentation necessary to quantify the actual conditions at that moment of testing
and finally to develop a written procedure (or checklist) so that all variables will be addressed in the same way each and every time
-> obviously detailed records must be kept both for data analysis, and more importantly, to be able to re-create a specific setup when some data 'won't fit' - to be able to retest

now this is a chore for component testing, it is a HUGE chore for systems testing

returning to wbs, what are the salient variables ?
specimen preparation (bp flatness and finish - how to quantify ?)
connection bore dia (stock, modified)
coolant flow (accuracy required)
pressure drop vs. flow rate
coolant temp (accuracy required)
heat source area
applied power (How Quantified ?, this is extremely crucial)
goop application (how thick ? - how do you know ? - did you measure ? -> big problem here)
clamping force (must be quantified, with numbers)
influence of ambient temp and air movement
die temp (the biggest problem of all)

quite a list
and any 'tester' who does not address each of these variables (one way or another) is ignorant, or a bull shitter
(feel free to quote me)

moving right along here
what equipment, with what degree of accuracy, is necessary to quantify these variables
different testers will have different 'standards', and budgets
but each of those variables listed needs to be quantified

pHaestus has commented on accuracy, and has discussed an article illustrating error propagation in calculations
I hope he does write this (soon !) as there are too many without the technical background to appreciate what constitutes a 'good' number
accuracy and resolution are not the same, and without some method of calibration its all make believe

do look at a tester's equipment list, and see what is said about calibration

all thermodynamic testing is 'touchy', and with the very small components WCers use the measurement increment is small as well
low resolution uncalibrated equipment can totally mask, or misrepresent, what we might think are 'significant' differences between units

almost forgot to mention that enough trials should be run to establish confidence in the results

testing is not so easy
BillA

EDIT: added pressure drop to wb variables

Thanks BillA. I was going to order what we've been talking about, but you did that for us!

The most important points that you've stated are related to documentation, methodology and consistency.

You didn't mention two more variables, and I've been meaning to bring them up for a while:

1-Nature of coolant: who ever said that water alone was the only coolant? I plan to try windshield wiper fluid (water+methanol)

2-Using a Peltier: this could turn into a discussion of its own. Does the waterblock need to be designed the same way?

Given the above, the number of combinations required for testing, under PowerHouse's specs, would be astounding.

As you also pointed out, thermodynamic testing is touchy. There are a great number of variables in such tests and so, everything should be controlled, measured and documented. I had this idea that in order to isolate the outside elements while testing a waterblock, that it should be isolated in a box, but then remembered that heat would accumulate within that box, and skew results. So I thought about venting the box but that too would skew the results. So I'm thinking that this needs a lot more thought.:D

My thoughts on the testing variables of a waterblock:

1-specimen preparation (bp flatness and finish - how to quantify ?)
TH:Usually, flatness is measured in microns, where 25 micron surface ripple is considered pretty well lapped. However, I believe that the block should only be lapped if the product is sold with instructions to lap it, or if it is in obvious need of lapping.

2-connection bore dia (stock, modified)
Needs to be measured and noted, but otherwise a wb would be tested at different flow rates, regardless of this.

3-coolant flow (accuracy required)
Measured and noted.

4-coolant temp (accuracy required)
Measured and noted.

5-heat source area
Measured and noted.

6-applied power (How Quantified ?, this is extremely crucial)
Measured in Watts, as a calculation of the voltage and current applied to a high power resistor, mounted to a plate to simulate a CPU (see www.dansdata.com)

7-goop application (how thick ? - how do you know ? - did you measure ? -> big problem here)
Using item in 6, I believe that it only needs to be noted.

8-clamping force (must be quantified, with numbers)
As I've mentionned in an earlier post, the mounting hardware that comes with a waterblock should probably be used, but if it is suspected that it may not be appropriate, then it could be substituted. This is really a subjective point, as waterblocks have different mounts, which may or may not be inter-changeable. Overall, the AMD specification should be considered.

9-influence of ambient temp and air movement
Measured and noted, along with humidity.

10-die temp (the biggest problem of all)
Using a test item (such as in 6), temp should be easily measured.

BigBen2k:

You trivialize understanding the true power (heat load) that is placed into the baseplate of the waterblock. Secondary losses, variation in the applied voltage (better use a line smoother of some sort), and flaws in the insulation become a huge issue. Trying to get the accuracy of the temperature measurements to where you want them is useless unless the input heat can be quantified equally well. Contrary to popular belief ("just use a really hot CPU"), the confidence in the numbers that come from a resistor or a CPU doesn't improve (IMO) with a bigger heat load. Instead, the likelihood that differences in temperature readings can be seen by your equipment may improve. This seems like a good thing, but if the temeprature difference is due to fluctuation in the house's power (maybe someone turned on the blow dryer in my apartment) then it can lead to very incorrect interpretation.

bigben2k
well, you're listening at least - but you'll find it hard to actually do

you are dead wrong about 'power'

try this thread
and DO READ Cathar's references
I repeat: you are dead wrong about 'power'

and I overlooked the pressure drop across the wb as a parameter that must be measured with the flow rate

Different versions of Radiate report different numbers. This in itself is troubling. To further complicate matters, some sites calculate a CPU's wattage with an equation that is a ratio of stock to overclocked speed and voltage. So for example a TBird 1.4 at 1.75V will have a different output than a TBird 1.0 with its bridges adjusted to run at 1.4 and 1.75V. Troubling again. Further complication is the assumption that fsb doesn't affect anything and only raw MHz matters in heat output. Is this valid? I dunno (any takers?). There is a pretty good read at benchtest.com on this topic:

http://www.benchtest.com/calc.html

Look at the Stress Test section. His estimate is around 88% of theoretical load (not so far off from Cathar's 81%). My new and improved diode reader should be here next week with 0.125C resolution. I will posthaste compare different "100% CPU load" programs.

AMD has programs that will actually load a CPU to 100% (their words not mine) but they aren't sharing. :(

Ok, after reading the last three posts, this looks more like a Doctorate Thesis, than it does a review on WaterBlocks.:eek:

Don't get me wrong, I'm sure that if and when you figure out how to test WaterBlocks (with the degree of accuracy you have listed) that no doubt you will have a great deal of information to provide for your readers. However, I believe the vast Majority of people will skip all of your hard work, and go straight to the conclusions as most people don't really care about Applied Power, Clamping Force, Thermodynamic testing, etc. This of course is my opinion, but I would say from the lack of others posts in this thread, that I'm probably correct.

If after all your testing, you are off by a less than a degree, who cares. It's not like the people reading your eventual review will have the same conditions in which you tested anyway. What I and others want to know is, which watercooling equipment would give me the best performance (to further overclock my system as that is really the point of watercooling isn't it).

Hope I'm not sounding too harsh, but these posts remind me too much of college lectures on stuff I could care less about. I'm a bottom line kind of guy, I don't mind reading information that I can use (which waterblock clamps onto the CPU the easiest, which has the best tube connectors, etc), but your methods of testing are far to rigid for my tastes.

lol

this from the fellow who posted pages on testing
but when push comes to shove, its all too difficult

there is a classic '50s English study which concluded that comprehension was inversly related to complexity
(in dumbspeak: few understand complex issues)

hence we have priests and politicians to tell us what is right and what we may do

ain't no Intel in this box

be cool

U Know power house is right about that.

U take an average "KID" who islooking to buy watercooling parts. Send him to a 30 page site on how the testing was done, and with what equpitment, and such, all he is going to do is keep hitting next, untill he gets to that graph that shows what block came out ontop, and not give a flying flip how the testing was done, or know that the testing may or maynot simulate the blocks true preformance on his overvolted duron 950. With DD cooling cube and 2 80 mm fans. Nor will he be able to comprehend most of the data supplided with the test. Alot of you guys are employed in this feild in one way or another, may it be working in a lab enviorment or, consumer testing, but the knowledge of most of the peopl on this thread, shows that you have a great knowledge of lab, and feild testing. Which is a good thing. maybe for some, to good of a thing. But please do me this one thing. Make your results ina format that is easy for the LAMAH to understand. Buecause if I am not mistaken, that is mostly your target group. (Fell that eplaining test data to a target group, should be included in the testing procedures as a MUST) If you complile all this data, and present it to say a 3rd grader, is he going to understand the data? No, but he will understand a Hot frying pan and a drop of water. See where I am going with this.

Keep it up guys, between your minds, we should see alot of good testing.

thanks
Fix

hahahahaha, i agree :p but accuracy and precision are impoortant. Fix also has a point, having all this data is good, but it would cretainly be easier to understand if you summarized whats going on instead of just giving us a graph and letting us come to our own (potentially wrong) conclusions..

methinks a CPU 'simulator' (kinda like what Frosty Tech uses to test HSFs) would be more consistent and more precise than using a MoBo and CPU.

i also think my eyes are about to fall out after reading all these essay-length posts

:drool:

ok, Fixittt does have a very valid observation

the presentation should be sufficiently straightforward that it can be understood also by those without a great deal of technical expertise

but such simplification needs to be done after the testing, not before

so if one wishes to discuss (good) testing, sorry but it will necessiate some rigor
(feelgood kinda stuff I'll leave to others)

Cyco-Dude (in case you've not seen them)
heat die simulators here and here

Well, I'm glad that someone picked up on my momentary lapse of reason, about power. I was thinking about it throughout the day, when I realized that what I wrote made no sense whatsoever.

But I don't think I'm very far off.

All in all, the lab condition should simulate, as closely (but reasonably) that of a PC. I think that we all agree there.

Using a high power resistor, and mounting it to some plate, the size of a CPU, is one way to achieve consistency in results, but it certainly is not a proposition that simulates a PC by recreating the power output of a CPU, and that power definitely cannot be simply calculated as I stated above (P=VI). (but I do believe that with the right power supply, the voltage spikes/drops would not be a factor).

Back to square one.

How do we simulate a CPU? How do we reproduce the power that a CPU generates? How do we measure that power? Should we consider max power, or average power? Should we test for max power +25% ? If all we're calculating is the heat transfer rate of a waterblock, is any of this relevant? What about water temp?

How do we deal with the other factors that dissipate power, beyond the waterblock (i.e. CPU to air, CPU to socket, ventilation, etc...)?

Can we do this cheaper than using a CPU and mobo? How does AMD do these tests?

And as unregistered pointed out (BillA?), there is even more info that we've mentionned (although, I will point out, I did refer to some of them, just not directly).:p
1-Total system heat capacity
as discussed here:
http://forums.overclockers.com.au/sh...threadid=60614
(Hello Aussies!)
This goes back to my point that some people believe that starting up a system, then noting data right away, or only after a few minutes, is acurate, when really, it's not. In short, heat accumulates in every single component, and is dissipated by each one of those components at different rates. (that's why I'm considering finned copper tubing).

2-Pressure drop across the waterblock: yes, it exists. yes it is a factor (and should be measured), but most importantly, it is a result of the flow rate used in the test. Knowing that info though, would help greatly in selecting a proper pump. I've always said that the waterblock is by far the biggest flow restrictor (unless the tubing is innapropriately sized, but I'm assuming otherwise).

Fixitt: you're right about the lengthy report not being fully read/understood, but the report doesn't have to have all that info in the main text, but should have references. The conclusion of the report should be able to satisfy just about anybody. As to why someone would take the time to write all that up, that's a personal issue.

PowerHouse is right, in that the objective of writing/testing anything is to be able to tell the average bloke which pieces he should get, and telling him what he needs to know. If PowerHouse (I'll use you as our "average bloke") doesn't understand the purpose of clamping, then he'll need to know, and how to do it.

The results to the tests could be posted in a table, in this form:
WATERBLOCKS
a)C/W
b) optimal gph
c) psi drop@that gph.

RADIATORS
a) heat dissipation rate (graph?)
b) optimal gph
c) psi drop at that gph
d) optimal air flow (graph?)

and so on. Knowing that info, any Joe, could mix and match, according to his budget. (The above is a proposition, is not complete, and is meant to spark the discussion about how to post results). Sorry PowerHouse, but telling you specifically which rig to get, could only bring on a number of liabilities: I'd rather give you the figures, and let you choose, that way, I'm not going to get sued!

PowerHouse: The purpose of water-cooling is not always for overclock. Some people just want to water cool to reduce the noise. If you are watercooling to overclock, then you would more than likely be using a thermo couple as well (otherwise, what's the point?).

If results posted are accurate to +/- 1 degree, then that may work, as long as the differences between the items tested are not within that range!

Better move C/W way down (off?) that list, as the whole power issue rears its head again (with a vengeance now). Bet you can guess I was going to say that after reading that thread? There are always tradeoffs; best plan is to simply address them.

There are some advantages and disadvantages of both the simulators you propose and actual CPUs:

Regarding testing with actual processors, the advantage is that you actually have a "realistic" load for the waterblock to deal with, with its dimensions and material and heat release pattern being consistent. Disadvantages are myriad: You don't actually know the heat load, you have difficulties in measuring the temperature with any accuracy, you have to contend with secondary heat losses and cooling pathways that complicate precise determination of a waterblock's performance.

Wait though that last one is REAL cooling that is relevant. Processor temperatures can be collected with a reasonable degree of accuracy with a carefully calibrated diode reader (no mobos), and the processor heat load can hopefully be ballparked better than currently done (how well? I hope AMD will respond to my e-mails). This is where I am headed right now, as there are issues with the simulators (and cost) that I can't surmount at the moment. Prevailing wisdom for testing is to crank the voltage and MHz, but if I can get more reliable heat load numbers from AMD for 1.75V and 1200MHz than I can for 2.0V and 1400MHz then shouldn't I go with the speed that improves certainty in heat input? Comments? I am also guessing that perhaps someone with a good simulator might be willing to help me check my W estimates by comparison with their known power inputs (pokes Bill with a stick).

Ok onto the simulators. I don't think that the W that is coming from most of these is any better than the W coming from radiate (or similar) for CPUs; it just looks more scientifically sound and so people don't question it. Look at that o/cers au thread for my diode reader comparison with JoeC's simulator. The only conclusion is that you really aren't any more certain of the power input with a resistor type simulator than you are with a CPU so the big advantage is the use of an embedded thermocouple over a CPU diode (not that this isn't a good thing).

Gotta think carefully about what you want to know and when to cut your losses. What were we even looking at again? Which block is "best"? I forget...

pHaestus

you're making it too complicated

use the wb heat throughput to cross-calibrate
you need only the inlet and outlet temps, and the flow rate (all quite accurately, eh)
since heat in = heat out, and assuming that the goop interface's thermal impedance is the same (???), using that Q equation from the rad article will enable the setting of 2 different systems at the same point wrt the wb

a typical CPU will have secondary losses, my heat die has close to none (HIGHLY insulated), JoeC's has a real problem due to the copper slug being exposed to the wb bp (addl heat transferred due to the close proximity radiation and convection outside of the die area)

I disagree re the inherent accuracy of a simulator
the AMD/Intel silicon chip heat dies are exact (but still do not replicate secondary path losses)
and my heat die is extremely accurate given the power supply, voltage and current measurements, and virtual elimination of secondary losses

EDIT: you and I could easily derive a correlation between 'true Watts' and any software generated 'Wattage'
-> so long as the CPU's heat generation was not limited by ANY ofher element of the couputer
(is this the case ?, I know nothing about such programs; more RAM, different chipsets, etc. change the 'heat' ??)
[I have heard they do]

Oops: generalizations
 

Bill:

Was referring to resistors on copper blocks (since that was mentioned by Bigben2k). Should have been more clear. Your simulator and test methodology seems to be quite good: I looked at your older flow rate vs heat load table at o/cers and noticed your %efficiency was around 98%. Not sure what you were meant by %efficiency; isn't that basically a heat balance? If you are accounting for 98% or more with your numbers then your simulator is working exceptionally well.

I would assume that the memory/fsb/settings dependency will vary from application to application. It may be necessary to combine several different operations at once to really get close to "maximum" load. I was also thinking about trying K7Burn in realtime mode from DOS, as high priority in windows already heats up the processor a good big more than Prime95 or foldiing or the others. I will have 2Hz sampling with the Maxim kit and I was planning on trying all of the common "100% CPU load" programs and see how they responded. I can easily try adjusting FSB as well with same MHz (Durons are easy to unlock).

I guess we'll keep this going 'till we get run off

as I mentioned to Les, don't go too much by the old data
I was initially going to title the rad article "A Poor Man's Calorimeter", until JoeC told me I had to define calorimeter.
that 98% business is fine until one of the calcs comes in at 102% and it is apparent that something is not properly calibrated;
but for sure, I can always compare the applied power to the cartridge heater to that heat being transported by the coolant

this is the point I was making; if you calculate the wb Btus what you have is the CPU generated heat LESS the secondary losses
- so if I adjust my power supply to the same wb Btu/Watt level
BINGO, we now know the secondary heat losses (for that board, conditions, etc)
-- nor would it be difficult to generate a curve related to the power level
and then to check that against other mobos

did you see saba's combined burn-in program ?
quote
Although it sounds rather drastic, try perhaps installing a simple web-server program and using Intel's IOMeter program to literally continuously hammer your system. Although it wont tell you specifically whats dead or not working, I find it really does put a huge amount of stress onto the system, which is always what stability tests are all about!

If I personally wish to stress-test my system, I normally install IIS (from your windows 2000 / XP cd's) and share some largish files (DivX movies work well). I set up IOMeter for 16 continuous WWW testing threads.

Starting a full recursive defragmentation within OO Defrag 2000 (www.oo-software.de, damn good defragger!) atop this is also very good. I also run two concurrent sessions of Quake II in a window, and let them demo cycle. Finally running 3dmark over the top of all this chaos.
unquote

I dunno, not my field

be cool

unregistered (I wish you'd just register!) has a point, in more ways than one.

I had to read unregistered's comments three times, to understand it. I think (correct me if I'm wrong) that you're proposing testing a CPU load, letting the system settle, and note the water temp. Then, replace the CPU by a plate mounted resistor, and adjusting the power applied to that resistor to match the water temp, in an effort to find out the ACTUAL power applied, as calculated by P=VI.

Then, calculating the power emitted by the CPU, using some software or calculation, you'd substract the power applied to the resistor, and you'd be left with the power lost by the other components of the rig.

Pfew... Ok, but I don't think that'll work.

The problem is multiple:
1-The CPu heat comes from a very small square (i.e. the die) and so, is concentrated in the center of the bottom of the waterblock. I'd like to be able to say that it's not important, but I believe that it may be. That's why I mentionned that cooling a Pelt may be performed better by a different block design, such as a Maze 1C, since (I'm assuming) a Pelt radiates heat in a uniform pattern.

2-I think that it is clear that the power generated by the CPU is the core of all tests, but I don't believe that knowing the actual CPU's power output is important, other than to know what the minimum, and maximum power outputs are. Here's why: the CPU will generate a certain amount of power, depending on the work that it is doing: running a word processor, calculating PI to the last decimal, etc... There is an infinite range of combinations.

3- (I'll get into that tmo, gotta go!)

Jesus Christ, these are the longest posts I've ever seen in any thread, anywhere. I was going to read them, but then I said f*** it. :)

bigben2k
nope
go to the rad article and look at the equation for Q (work done)

on a computer with wb; set 'Radiate" Watts of whatever, and calculate Q
then on my bench setup adjust power to achieve a Q, with the same wb, of the same value
since the CPU heat generated, less secondary losses, equals the 'true' bench measured Watts;
the difference will be the secondary losses

slick, eh ?
and yes, all measurements always at equilibrium (only)

ECUPirate
thanks for letting us know you're dead from the belly button in both directions
(hint: less effort with the 'Back' button)

be cool

Stumbling blocks on my end:

No laboratory chiller to control water inlet temperature. I can still get numbers but this will result in constantly changing boundary conditions. Possible to get around this?

Flow rate measurement. I have a relatively inexpensive flowmeter that is +/- 6%. This turns out to be the major stumbling block I think.

Still have to decide upon a way to assign watts. I'll most likely go with the Duron 1200MHz at 1.75V so that I can pull the typical max load number from the AMD spec and just go with that. Comments?

Good news: CPU die measurement possible with reasonably good accuracy and resolution, wb inlet and outlet temps are also I think ok (you tell me hehe).

"No laboratory chiller to control water inlet temperature. I can still get numbers but this will result in constantly changing boundary conditions. Possible to get around this?"

no problem so long as you can hold a temp long enough to reach equilibrium (say +/- 0.2^C, but NOT trending)

"Flow rate measurement. I have a relatively inexpensive flowmeter that is +/- 6%. This turns out to be the major stumbling block I think."

may have a soln, looking for a link

"Still have to decide upon a way to assign watts. I'll most likely go with the Duron 1200MHz at 1.75V so that I can pull the typical max load number from the AMD spec and just go with that. Comments?"

you do need to run that matrix described earlier, too much not known

you've got the coolant nailed

be cool

No concerns about getting run off (I have a set of keys). Getting better estimates of secondary heat losses (and bad numbers from calculators) turns out to be a little more interesting than testing waterblocks even (from my perspective); I will get right on this.

I can verify Prime95 heavily depends upon memory, as it will crunch all night with my Crucial DDR at 150 fsb conservative settings but crashes within a minute at "fastest" settings in bios. That is something I don't believe I have ever heard anyone point out before in relation to temperatures. RAM at CL3 means lower temps? Hrmmm.... Different stress programs and different FSB post haste! I will see if I can get i/ometer downloaded and installed as well.

Hrm a link, eh? I can just feel the cash leaving my wallet. Oh well; no turning back now (my interest level is too high).

HEY, goddamnit pHaestus
I was absolutly content as a hot stank, feel free to it change back

a prediction:
after jerking yourself around for several weeks you will find that this entire CPU-as-a-heat-source idea is worthless
(in terms of repeatable cross-platform results)

a suggestion:
start a list of each computer hardware variable that changes the heat output
then the same for software related items

gonna be a long list

still looking, but I'll find it