Are we being to anal about testing?

[bigben2k]

From my perspective...

There's 3 or 4 kinds of testing (depending on everyone's perspective):

1-system testing: that's the old PC test.

2-Comparative testing with PCs

3-Comparative testing with a testbench (heat die)

4-Analytical testing with a testbench (is there any other way?)


The issue then becomes about repeatability, aka the ability to reproduce the same results, wether it's in one day or one year from now.

Accuracy is important, mostly in analytical testing, because you want, as much as possible, to come up with a result that can be achieved with another completely different analytical test bench: that way, you have a true measure of a block.


When I decided to get in on testing, I knew what challenge I was facing, and I knew the extent of the work that it was going to involve, just to get to that one day when I can test that first block.

I also expect to make errors on the way, but I am confident that I can get it done, even though my education only consists of 4 years of college (2 in Applied Science, 1 in Electronics, and 1 in PCs). The only part I won't get into, is custom electronics, like that fan controller to maintain a specific water temp that I mentionned some time back, or that CPU ammeter (by Groth), because my confidence level there is low. (I would be more than likely left with a bunch of unused parts!)


The real issue here, is wether any testing out there is significant. "System testing" just doesn't have the accuracy to return meaningful results to anyone else but the tester, so it's really important to keep that in mind. Given that, there's still a need to reconcile testbench results with PCs, so pHaestus' work here is of the highest value (yes, I've been watching!). Those CPU temp results are really important.


On a personal note, I've come to accept ProCooling as a place to be criticized, so when I need that, I'll post something. Otherwise, I intend to keep a lower profile, and try to catch some of that "education" that's become so rare here. I'd love to stir things up again, to further advance all of our knowledge in all this, but I don't care to deal with the attitudes of some: seems to me like there's more interest in flaming/bashing, then an actual willingness to work together to learn/figure out things.

[BillA]

a far better approach Ben
stay at it

[Since87]

Originally posted by bigben2k
seems to me like there's more interest in flaming/bashing, then an actual willingness to work together to learn/figure out things.

Seems to me there is more serious work put into learning/figuring things out here, than at most WC forums.

I mostly follow oc-forums and procooling. There are probably 10 times as many WC posts at OC, and I am about 10 times more likely to link a PC post at OC than vice versa.

By my reckoning that means about a factor of 100 in signal to noise ratio. (40dB) If bashing comes with it, so be it. (IMO - I certainly don't speak for Joe and/or pH)

Actually, I don't see people, who post here to learn, getting bashed. What gets people bashed is presuming to teach without knowing the subject.

No one is better at being a teacher than they are at being a student. Those whose ego is wrapped up primarily in being seen as a teacher, don't have much to offer at the cutting edge.

[Player0]

Well, I'm only human. I get just as pissy as everyone else when I start getting rubbed the wrong way. If you find some of my remarks to be unproffesional and defensive, it's only because of being assulted on here.

When you take into account that his testing setup was an Abit in socket thermistor and a digidoc, that he mounted blocks 2 times, and that he generally has disdain for bench testing then I stand corrected JayDee.

The Abit sensor has been isolated with silicone and neorprene to limit external temperature stimuli, the digidoc was calibrated with a Fluke DMM with a probe attatchment, and none of the blocks in the review have only been mounted 2 times. As I continue my testing, I am revisiting the blocks, and each block has been mounted 3-4 times, with nearly no deviation in performance between mountings. The Swifty and RBX will be mounted more than half-a-dozen times when I am done. And I have no disdain for professional bench testing. I have nothing but respect for those results and the people willing to take the time to be so thorough. But Im not trying to compete with that, and I am doing the best I can with the tools at hand, and Ive never claimed that my my results are 100% gospel. Simply a rough guide for others to compare performance by.

The review is probably of little real use to anyone (well blocks performing within 1-2 C of one another with a 10C error bar).

There's a spread of about 8c degrees between the different blocks. Yeah, I do wish that I had more input wattage to test on. But the reality is, waterblocks on an Athlon do actually perform very similar to each other. Yes, on a P4 the results could be different, but that wasnt in the scope of my test. I would like to see the ranges a little higher, but I still see value in my testing. I dont beleive the error is anywhere near 10c. Within two degrees dT max, if something went very wrong. But my results are very consistant and im sure they are reproducable.

Regarding the PM quote, Im not familiar with the incident, but its not beyond the bounds of possibilit. Its all up on LN.com somewhere, since we don't delete posts. Like on any other site, sometimes people get in to debates, sometimes people leave. I dont think that if I 'drove' someone away that I was the only guilty party. I have my bad days just like everyone, and hell, I might have even been wrong in that case. I admit my mistakes and I try to learn from them. But there are always two sides to every story. The guy was probably being a jerk anyway.

Actually, I don't see people, who post here to learn, getting bashed. What gets people bashed is presuming to teach without knowing the subject.

I dont recall trying to teach here. I just posted some of my results because the RBX topic came up here. If my testing doesnt meet the standards some of you might have, then just ignore them I guess. All I ever commented on was that the RBX has near Cascade performance, and i think other comparisons will show that. I still think the Cascade is better though, especially on P4 systems where the differences will be more apparant.