Your take on the Apogee

Erasmus354 is demonstrating the exact same line of questioning and concerns that I had, and still have.

The wb in question was tested multiple times, not just all in a row, but after much gnashing of teeth over the results, it was tested after some "known good" results with another block, then it was tested and produced anomolous results, then the same other "known good" block was retested - came back with the same "known good" results, then the anomolous block was retested, and it came back with the same anomolous results. Subsequent tests on different testbeds were unable to marry the results discovered on the TTV with anything else. As Bill pointed out, physical inspection of the anomolous block found it to be improperly clamped shut, with an easily flexible base plate, and a propensity to divert a significant (my estimate, almost 2/3's) of the water-flow directly between the inlet and outlet of the top tweo plates of the 3-plate design, totally bypassing the main cooling area.

Now given all of the above, when it comes to user error in testing a calibrated testbed, one does not normally expect to see results that as Bill described as "astoundingly good". Now to put that into context, we're not just talking "this is a good block", we're talking about results that totally demolish all known competition by a huge, huge margin. All this, from an improperly assembled, flow bypassing, base-plate flexing waterblock with a primary cooling patch even smaller than a White Water.

I do appreciate Bill's honest civil candour ~10 posts above. I was wondering for a while there where it had gone. Good to see it make a return. I do not enjoy the locking of horns.

The line of reasoning that Erasmus is following, and that I too follow, vs the differences in opinion between Bill and I ("going at it hammer and tongs") is perfectly highlighted in the above ~8 posts of interaction between Bill and Erasmus. IMO, one does not obtain a bad set of data with one design that should NEVER be performing as such, then recheck the testbed against "known good results" (such a "known good" status is now in question though), then recheck the anomolous block, achieve the same anomolous results, shrug the shoulders (metaphorically), discard/ignore the anomolous results and then continue on. Such is indication of a systemic error that MUST be understood, otherwise it brings every single result into question.

That is where the core difference in opinion lies, and is what has been driving much of this public display of attrition.
 
E3
suggest you consider what you are about
you wish to engage in a detailed discussion about a specific test series
why ?
I infer from your questions you would consider this data as material to the evaluation of the TTV as a test platform.

if so, your thoughts are foolish indeed

to evaluate the TTV, ignore me and all my data
ok ?
go back to post #68 and address, in detail, that first paragraph
you need to demonstrate your grasp of sink testing before you have much hope of grilling me about the design of experiments

to Stew also,
Do you understand that the validity of the Intel TTV has nothing to do with (is not dependant on) my work ?
You would both use my imperfect work on a wb known to have imperfect assembly to prove . . . . ?
If one is unable to accept that the TTV validation done by its mfgr is competent, then you guys need to start everything from scratch to establish the competency of the mfgr for every single thing.
??

My purpose in that testing was not to generate 'compatible data' for your calcs, and I had zero commitment to allocate testing time to another's product in which I had little interest, just curiosity.
 
It seems to me that some are expecting absolute values from the TTV, where it, like any other measuring tool, can only accurately describe the differences between two test sets- the deltas. My question to all is this:
What does the validity of the TTV have to do with the usefulness of the Swiftech Apogee? I have the block, and it performs beautifully, albeit nearly identically to my 6000, but the Apogee was cheaper, looks better, and has a MUCH better retention system.
 
Kil4Thril said:
It seems to me that some are expecting absolute values from the TTV, where it, like any other measuring tool, can only accurately describe the differences between two test sets- the deltas. My question to all is this:
What does the validity of the TTV have to do with the usefulness of the Swiftech Apogee? I have the block, and it performs beautifully, albeit nearly identically to my 6000, but the Apogee was cheaper, looks better, and has a MUCH better retention system.
Click to expand...
one of the things that is under question is the difference between swiftech's numbers and the numbers over at systemcooling, and which more accurately represents the way in which the block performs in a real system.

$45 US for the MCW6002 vs $50 US for the apogee at stores that i just checked, so the price is pretty close, in spite of you noting that in the apogee's favour. looks are subjective, and some people like me just don't give a damn how the block looks, as long as it performs. in terms of which block performs better, although swiftech's numbers are showing that the apogee is a better performing block, systemcooling is showing that the MCW6002 is a better performing block.

so the question is: who's testing is most reflective of what the waterblock can do in a real system? is the TTV a better representative or is the heat die that robotech used for his review?
 
BillA said:
to Stew also,
Do you understand that the validity of the Intel TTV has nothing to do with (is not dependant on) my work ?
Click to expand...

Of course. Your work, and the TTV's validity are two separate things.

You would both use my imperfect work on a wb known to have imperfect assembly to prove . . . . ?
Click to expand...

See Bill, this is where what I'm about to say may surprise. I have more faith in the end-user engineer's perception of reality than a manufacturer's. In all large industry what the original manufacturer and designers had in mind may not always account for what is seen at the other end. Hate to use the car anaology, but it is quite relevant. Often it is the end-user mechanics who have to work on the things day in and day out that start to pick up the faults and flaws that were not obvious in manufacture.

Do not take this as offense, but I see yourself in much the same light. You are the end-user "mechanic/engineer" who has to deal with the anomalies of what's been dealt to them as part of their job.

I have more faith in your (Bill Adams') ability to identify and isolate issues with a testing apparatus than that of taking the word of a manufacturer marketing engineer. Having worked in big companies of comparable size to Intel, the intent is never there to deceive by the manufacturers, but things do move slowly and nothing is ever perfect, regardless of high hopes and intentions. All too often it is the humble ground level engineer who is first to pick up on even quite glaring flaws. Having worked in a similar industry for a while, I have learned to trust the word of the ground level guy before I trust someone in marketing. Not because marketing lie, but because the ground level engineer sees the issues first and needs them fixed.

If one is unable to accept that the TTV validation done by its mfgr is competent, then you guys need to start everything from scratch to establish the competency of the mfgr for every single thing.
??
Click to expand...

Nothing is black and white in a manufacturing world Bill. Intel have made some spectacular blunders in the past. Everyone and every company makes mistakes. Big companies are machines and the left hand is often unaware of issues in the right hand's palm. Finding flaws and oddities in the TTV does not spell disaster for all of Intel. It is a single set of units that are exhibiting anomolous behavior, not altogether different from say a cam chain linkage that may misbehave on a certain type of engine from a car maker.

It is not black and white. What is true though is that if the anomolies and issues continue to be overlooked and dismissed, then they will never be addressed. It has to start with the humble ground level engineer in my experience. All too often the engineers in the "ivory tower" of home base are unaware of issues in the field. Nobody is perfect, but dismissing bad data is a sure fire way to make sure nothing gets fixed.

My purpose in that testing was not to generate 'compatible data' for your calcs, and I had zero commitment to allocate testing time to another's product in which I had little interest, just curiosity.
Click to expand...

It's not about compatible data. There is an issue somewhere. I have more faith in the tester and his selected equipment, because he had used them for a while and knew their quirks. The random variable here, in my mind, is the TTV. Why? Because it was the one thing in all of that testing that was producing results that could not be explained.

Don't sell yourself short and put Intel and their TTV on a pedestal. It doesn't work like that and it never has in my experience. The TTV is an air heatsink validation mechanism. I can't imagine that the original designers of the TTV ever anticipated that someone would mount some cooling apparatus with the physical peculiarities of the anomolous wb and then expect it to always work correctly in that instance. Intel themselves qualify their own level of faith in the "infallibility" of the TTV in that they strongly recommend that everything be cross-checked on a real world system.

I believe that Bill, the humble engineer, has stumbled across something even though he is unwilling to believe it and doubt himself first.
 
Well I had a nice eloquent response all typed out and then my machine crapped out on me :( Guess the memory wasn't entirely stable at 250. Anyways I am a little depressed losing all that I typed, but Cathar covered most of it. In fact I think he stole the words right out of my mouth when he said the TTV is by no means infallible.


BillA : The reason why I wanted to start a discussion about test data and what not is because I am a student and am incredibly intrigued and interested in all of the design and testing of waterblocks. Why not try to learn more from one of the most experienced in the field?
 
Stew
I've used a couple of generations of these things, burned some up too. Routinely pulled the pins on P4s, thought I would break the board with 775s.

Every variation I've encountered has been attributable to something other than the TTV, the notable exception being the wb under discussion. I am not perturbed, how many hundreds of tests were otherwise ? The wb and tests have known problems and the field/review reports have shown some variation as well, we discussed this at length.

It is the notion of a "flawed" thermal validation tool that is far fetched; 3 generations that I know of, worldwide distribution to channel for sink validation, and some rather large commitments made on the basis of such. A validation tool may burn up, that is ok; they cannot produce inconsistent results. As I have not found them to be a source of variation, I cannot support that hypothesis. And obviously I am sick unto death at seeing that hypothesis presented as a fact.

E3
"the TTV is by no means infallible"
is TTV a collective noun ?
which TTV ?
different ones had different 'issues', so ?
are you talking about variation ? or sources of variation ?
specifically which one(s) ?
-> how much variation is 'ok' ? (VERY simplified, eh ?)
or are you proposing that that particular TTV was bad ? (only on that wb ?)
please, give it a rest
 
BillA said:
It is the notion of a "flawed" thermal validation tool that is far fetched; 3 generations that I know of, worldwide distribution to channel for sink validation, and some rather large commitments made on the basis of such. A validation tool may burn up, that is ok; they cannot produce inconsistent results. As I have not found them to be a source of variation, I cannot support that hypothesis. And obviously I am sick unto death at seeing that hypothesis presented as a fact.
Click to expand...

...and as I've stated repeatedly, without Tjunction, along with Tcase, then one cannot conclude consistency or inconsistency. Intel's own internal testing uses Tjunction as part of their own consistency assessments, but such have ever only been done on "well behaved" cooling apparatus, or at least that is what is implied by the available documents. Sadly Tjunction is not available on a TTV to verify consistency for cooling apparatus outside of the original intended well-behaved air heastinks that the TTV is supplied for pass/fail validation purposes of said cooling apparatii.

I never stated that the hypothesis is fact. It is a hypothesis until it is proven otherwise. It is a hypothesis supported by the presence of inexplicable data.

I am unwilling (obviously) to take it on faith alone that there is not an inherent systemic issue when the TTV is used on cooling apparatus outside of both the publically disclosed statement of purpose, and on cooling apparatus exhibiting characteristics outside of the basic premises that are established (or at least presumed) the background Intel design documents on Tjunction and Tcase interactions with respect to assumed physical properties of that cooling apparatus.
 
I sense a little bit of "pro" mentality eeeking over into one of my beloved forums. Say it isnt so.

Be that as it may, I truely believe these TTV discussions be better served in a TTV thread... instead of EVERY WATERBLOCK thread concerning the apogee and storm. Without failiure, ever single block thread that mentions the apogee, the TTV eventually comes up. ... then all the "pro" people come out in force and the same childish debates contaminate yet another thread.

Tiresome, to say the least. How many times can the same things be said?
 
orkan said:
I sense a little bit of "pro" mentality eeeking over into one of my beloved forums. Say it isnt so.

Be that as it may, I truely believe these TTV discussions be better served in a TTV thread... instead of EVERY WATERBLOCK thread concerning the apogee and storm. Without failiure, ever single block thread that mentions the apogee, the TTV eventually comes up. ... then all the "pro" people come out in force and the same childish debates contaminate yet another thread.

Tiresome, to say the least. How many times can the same things be said?
Click to expand...
if it's not one thing, it's another. if it's not testing methods it's big bore vs small bore, and i personally find some of the details that i have read about how waterblocks are designed and tested to be fascinating.

to each their own: i see a discussion comparing the apogee to other blocks being a perfectly valid place to talk about which tests show what block having the edge, and more interesting still would be what could account for it.
 
very true.

I guess I've just been haunting too many forums. I've read this same argument no less than 20 times on 5-6 different forums. This is just the first time I've seen it on the [H], and I'd had hoped we were immune here. ;) :D
 
Sorry to come from left field here, but what is TTV? I see it everywhere but dont know what the acronym means. What does it do? How does it relate to the Apogee? Thanks
 
TTV= Thermal Test Vehicle.

Basically an artificial CPU testbed for evaluating the performance of heatsinks.

I think....
 
You must log in or register to reply here.
Back
Top