Thermalright SP-97 Review - an example in Hype-Testing - Rant.

[Joe]

ok now normally I wouldnt post links to such shitty sites as the one I am about it, but... with all the talk about accuracy, and all the posts (and flame wars) recently about it here and other cooling sites, you would think a "cooling" site would take note.

Now since I consider the site I am going to link to, to be much more a wordy advertisement page, hell bent on the propigation of hype instead of actual results; I can understand that they arent really a cooling site and lack the interest in actually "testing". But Cummon, lets read this together:

"The Test System

We've made some changes to our standard test bed since the last heat sink review. Our Athlon XP2100+ has been replaced with an Athlon XP2700+, and the motherboard beneath it, while still an Epox 8RDA+, is now a rev 2.0 board. We also changed the mounting for the temperature probe slightly by gluing it in place with Arctic Alumina thermal epoxy, rather than using thermal tape. The net result of this change is that we achieved a much more responsive and more accurate temperature reading than we had previously...

...Also note that the average temperature readings run somewhat higher than with the previous setup, so direct comparisons against previous reviews are not valid. ..."

Umm WHAT?! So not only are they using temp reading methods that went out of style back 2 years ago, but they openly admit that their past works were wrong?

Heres some more goodness:

"Testing Methodology

All testing was done using a closed case, so as to more accurately reflect real-world usage conditions. CPU and ambient temperatures were monitored via thermal probes attached to an Enermax multifunction panel. Ambient temperature was maintained at 22ºC, +/- .5ºC throughout the testing."

I busted out laughing reading that. We all know the stunning accuracy and detail 30$ temp reading devices made for drive bay slots have. One thing I would like to point out is that the .5C +/- is a lie. Ive done enough testing myself that tells me getting a DeltaT in ambient air like that for any period longer than 3- 4 minutes in a normal house is incredibly hard to control. (Especially for a HSF which is heating the ambient air its inhale'n)

So check out the great work of imagination, and creativity, cause there sure as hell isnt any real testing going on at that site:

http://www.systemcooling.com/modules...ticle&sid=1464

[pHaestus]

Joe are you picking on the retarded kids again?

It IS funny though that when they revise the testbed that this is the best they can come up with. Kinda surprised that they did even this as if it was producing lower temperatures before then that would be mo' betta' from an advertising standpoint. We both know that most manufacturers would prefer to see a shitty quality review with CPU running 3C over room temp under load than a quality review showing 25C over intake air temp.

[bigben2k]

I'm more concerned with the weight of this unit (565 grams!) than with slamming SystemCooling, again!

At least they were honest, about their testing method.

The statement about the 0.5 deg C was meant, I believe, to state that the temperature was maintained at 22.0 degrees (as read by the unit) and held within 0.5 deg during the test. There's no claim of "0.5 deg accuracy" here. How this was achieved isn't specified though (ducting air from another room?!?).


One thing that bugs me, is the results table: once again, the error margins aren't specified, and the numbers are posted with one decimal, which (ideally) implies a 0.1 deg C precision, but this type of setup, as most of us know, will easily have a 2.0 deg C variance. That's one part that irritates me.

Still, a 2 deg C error is enough to state that the SLK-800 was beaten, but what's the actual margin of error?:shrug:

[Joe]

At least they were honest, about their testing method.

hu? maybe I am the only one that thinks the ambient temp gradient is fabricated. but thats just me.


Funny you should post so soon, we were just talking in chat about how they could use a cooling "expert" like you to whip them into shape

[pHaestus]

At least they were honest, about their testing method.

HUH? Isn't there at least the inference that this is a VALID way to test? Otherwise why would they do it? If they were honest, they'd say "we don't know what the **** we are doing and so we just glued a $5 temp probe on the side of a CPU and went with that." Or "Well we are aware of the AMD guidelines for testing, the possibility for correlating CPU under measurements with die temps, and with the work that people have been doing with diode readers. But we opted just to do things half assed and pretend like it's OK."

How is it honest to post expert opinions when you obviously are not?

[bigben2k]

Well, I should correct that... they "presented" their testing method honestly, but not the accuracy. They could have left out the testing method entirely, which would have been a big no-no.

[BillA]

Originally posted by pHaestus
. . . .
How is it honest to post expert opinions when you obviously are not?

HUH ?
wtf pHaestus, you would pose such a question to BEN ?

don't wish, or need, to say any more

[jaydee]

Originally posted by pHaestus
Joe are you picking on the retarded kids again?

Ouch. Little course there. I agree on the testing aspect. I defended Scomp's review at one time because I believed he was making progress on the testing methods, but I don't see any significant improvement since then from that site. I don't know how much cash they make with all the ads and fluffy reviews but one would think they would invest in some decent equipment by now. I know Pijo and respect him for what he has done at amdmb and his willingness to help people. He isn't retarded by any stretch but he shouldn't be doing cooling reviews IMO. About all I am going to say. Just glad I didn't start this thread!

[pHaestus]

hmm sorry I should have phrased that better. Not calling anyone a retard; meant to make an analogy. That calling systemcooling testing measurements to task was a little like picking on retarded kids. Or shooting fish in a barrel. Whatever analogy you like. Purpose wasn't to make personal attacks.

I pity anyone who buys into Bob Dyl and co's brand of crap as objective testing though.

And FYI ads on that page cost a LOT (I was told $1000 a month and up and that was over a year ago). You can do the math as there are lots of ads.

[bigben2k]

Actually, there's a number of other missing elements in the testing method.


What's the purpose of this thread again?

[Joe]

Ranting.

[BillA]

complaining about websites with higher Alexa ratings

[Joe]

Its all about Alexa envy for me

[BillA]

pssst, Joe
I think they have a patch now

[TerraMex]

I think Joe should concentrate on this Alexa.