www.overclockers.com tests Glaciator, SK6 and the Swiftech. wonder who wins??

[PlatinumGold]

check it out.

Link

 

[anime]

yeah i got sk-6 and it worked prety well on my 1.4ghz t-bird

 

[Martijnos]

I wonder how the SK6 performs with a less aggressive fan. I'm in the market for a new heatsink, but it's really hard to make a choice between these three. I've read somewhere that the noise of the delta is more tolerable on the SK6, is this true?

 

[Pabster]

ROFLMAO. ThermoEngine? Ok, I'm officially removing overclockers.com from my host lookup file. These guys are ridiculous.

 

[PlatinumGold]

martinos

i've used the SK6 w/ a 4800 rpm fan on it. decent performance. about equivalent to WBK38 w/ delta on it. but in order to achieve those results you really have to have good case cooling. copper hs are very ambient temp sensitive. much more that aluminum it seems to me.

 

[Bozo Galora]

ummmm.....
SK-6 made by THERMALRIGHT, not Thermoengine
SK-6 all copper
ThermoE all alum. (but copper cored model coming)

 

[Martijnos]

Thanks Platinumgold

I'm going to choose between a Swiftech 462A with a 80mm Panaflo H1A (39.6 CFM, 32 dB-A)and a Thermalright SK6 with a YS-Tech FD1260257B-2A fan (25,9 CFM, 36 dB-A). The Glaciator is not available here in Holland.
Which of the two would give me the best performance?

 

[PlatinumGold]

martinjos

have you verified that the Swiftech will fit on your mb?

i can pretty much guarantee that the sk6 will fit on your mb. but i must comment the clip on the sk6 is one of the worst i've seen.

but if your going to put it on once and forget about it not a big deal.

 

[Mikewarrior2]

Wow, very impressive.

Pabster,

You bring new meaning to zealotry. Anyone that says anything even remotely non-positive about Swiftech is automatically on your "hate list". That is ridiculous.

I cannot believe that you really, honestly, with a clear mind actually believe that the swiftech will never be beaten. Overclockers.com has not said that the swiftech is bad, just that the thermalriht is slightly better.

Your views are clearly blocked by that haze of "zealotry" you have. Its very funny to read.


Mike

 

[Pabster]

"You bring new meaning to zealotry. Anyone that says anything even remotely non-positive about Swiftech is automatically on your "hate list". That is ridiculous."

Not at all. It's just interesting that people are so hyped about supposed "results" from a site that just can't seem to make the numbers add up. Am I the only one who also noticed their "test" was done with another model using a Delta and the Swiftech only using the Sanyo-Denki? How in the world is that a fair comparison? I've also seen much better results with the same Sanyo-Denki fan the Rev1 units are shipping with. And, based on Overclockers.com previous performance ...

"I cannot believe that you really, honestly, with a clear mind actually believe that the swiftech will never be beaten. Overclockers.com has not said that the swiftech is bad, just that the thermalriht is slightly better."

And like I've said before, if you believe that, good for you. (I've also got some land for sale in a certain desert.) I'll stick with the 462-A.

"Your views are clearly blocked by that haze of "zealotry" you have. Its very funny to read."

You're right, I probably should just not bother recommending the top-of-the-line HS/F. I mean, those Orbs are good enough.

 

[dowxp]

Gorbs are good enough! i know! =) 55c on both p3's running rc5....

 

[Mikewarrior2]

The swiftech's now come with the quieter 53CFM fan. IF that's the way it comes from the manufacturer, then that is how it should be tested.

it is okay to recommend the swiftech, as it certainly is an excellent unit, but For you to constantly SAY that it is hte only choice is annoying to no end. You just can't believe that there are top "tier" performance heatsinks that include the Thermalright and Glaciator units, as well as the Swiftech.

And where have you seen "muhc better" performance with the swiftech? Socket-thermistor based tests? More heatsink-thermocouple tests?(which, lookign at this test, is still a flawed testing method. Internal is still, by far, the absolute best). Heatsink-thermocouple mounting area plays a huge role. Depending on where it is sitting over the core can drastically alter the readings. This is from personal experience and testing.


Mike

 

[Pabster]

Mikewarrior2 wrote:

"The swiftech's now come with the quieter 53CFM fan. IF that's the way it comes from the manufacturer, then that is how it should be tested."

Well, that invalidates 90% of the "heatsink reviews" floating around then, eh? Nearly every one uses a fan different from that which ships with the unit, and there's nothing wrong with that. But they shouldn't compare a unit with a Delta to a unit with a lower-volume Sanyo-Denki and then declare the former the "best". Doesn't cut it.

"it is okay to recommend the swiftech, as it certainly is an excellent unit, but For you to constantly SAY that it is hte only choice is annoying to no end. You just can't believe that there are top "tier" performance heatsinks that include the Thermalright and Glaciator units, as well as the Swiftech."

Sure I can. I'm not arguing that the SK-6 and Glaciator are "top tier" units. But declaring either of them superior to the MC462-A is laughable, and results floating in from people using these models confirms that.

"And where have you seen "muhc better" performance with the swiftech? Socket-thermistor based tests? More heatsink-thermocouple tests?(which, lookign at this test, is still a flawed testing method. Internal is still, by far, the absolute best). Heatsink-thermocouple mounting area plays a huge role. Depending on where it is sitting over the core can drastically alter the readings. This is from personal experience and testing."

I've observed far better cooling results with an MC462-A and the stock Sanyo-Denki. Whilst "Joe's" test shows 46C under load with a 98.1 watt heat dissipation, I personally observed 40C under full load with nearly 115 watts of heat dissipation. Same board, same fan. And this was done with a heatsink-thermocouple (as you put it), unlike "Joe's" test where he (for some reason) uses the in-socket thermistor. With the Delta 68.5cfm fan, that full load temp dropped to 36C.

You make 2 observations which I have a problem with - first, you are still discounting the accuracy of a heatsink-thermocouple. I think we can both agree that for now, it is THE most accurate way of measuring temps. You yourself heatedly debate any readings from socket thermistors, yet that is precisely what "Joe" used in his "review". This coming from a man who has previously admitted inaccuracies and flaws in his "reviews", yet hasn't changed his ways a bit. We can argue about placement of the probe forever, but, fact is, it doesn't make a hell of a difference. Obviously you want it as close to dead-center on the core as possible, but a slight variation isn't going to *drastically* change the numbers.

I'm just rather puzzled how you can be so adamant against socket thermistor readings and comparisons, and yet read "Joe's" latest "review" where he uses precisely that method and call the results "good". I beg to differ.

 

[NesuD]

I found the variances on the backside cpu thermocouple readings from apparently a cpu topside reading pretty interesting. Correct me if i am wrong but isn't that bottom cpu reading basically the onboard socket a thermistor. If that is the case than their results would indicate that the board reported temp is reporting a higher than actual temp.



<< What I found most interesting was the difference in CPU back temps - the Thermalright showing the smallest variance (7.1 C) with the Glaciator and Swiftech showing larger variances (11.9 and 15.2 C respectively) (CPU back Temp was measured by a thermocouple on the CPU's center back). Large variances from the front to back temps have been associated with less stability. >>

 

[Mikewarrior2]

Joe's tests use the heatsink-thermocouple. He has been using heatsink-thermocouple tests for a long time now.

Per the latest review, since you obviously can't decipher information properly: CPU Case temp is what JOE is calling Heatsink-Thermocouple temp. I guess from someone hwo doesn't read overclockers.com makes hte wrong assumption about their reviews.

Thermalright SK6 Heatsink-thermocouple temp: 42.2C. ALso known as CPU Case temp.
Thermalright SK6 Thermocouple Back-temp: 50.1C.
Thermaltright SK6 Socket-thermsitor Temp: 40C

Glaciator Heatsink-thermocouple temp: 46.3C
Glaciator Thermocuple back-temp: 58.2C
Glaciator Socket-thermistor Temp: 46C

Swiftech 462A Heatsink-Thermocouple Temp: 45.6C
Swiftech 462A thermocouple back-temp: 60.8C
Swiftech 462 Socket-thermistor Temp: 46C.

So JOe uses 3 methods of temp measurement: Heatsink-thermocouple to get a heatsink-surface temp; Backside thermocouple mounted directly behind the core; Socket-thermistor readings through Motherboard Monitor.

Yes, the CPU's are at different speeds/wattages, but for you to say that he is using socket-thermistor readings is incorrect. He is using Heatsink-Thermocouple first, and foremost.

And no, I don't agree that heatsink-thermocouple testing is the best. Shifting a thermocouple from center to 1/4inch toward the bridges on a t-bird will drop the reading several C as you move over the L2 Cache. It will become apparent as soon as windows programs that read the Athlon MP internal diode become available that internal diode readings are still, far and away, hte best readings possible.

As evidenced by the earlier Overclockers.com Heatsink-Thermocouple tests, the heatsinks were ranked: Swiftech, Glaciator, and Thermalright in that order. However, the &quot;worst&quot; performing heatsink-thermocouple unit overclocks the best, leading truth that even heatsink-thermocouple tests have flaws.

Maybe you should read some of their older reviews before discounting them.

Mike

 

[PlatinumGold]

pabster

youre just a troll.

it's not that your opinions are offensive, you are welcome to them and you have the right to them, but on a board like this, mb temper your opinions a bit. we are not 13 yr olds and you are not our father. you are not in any position to lecture.

a little tact would be in order.

also, when you go off on www.overclockers.com at least have your facts straight. Where in that whole article was the Thermoengine mentioned and in what context??

finally, why should we respect your opinion any more than what we're getting from overclockers.com??

you have to earn your respect before you go off like that.

 

[Pabster]

PlatinumGold wrote:

&quot;pabster

youre just a troll.&quot;

No, I'm far taller than those things

&quot;it's not that your opinions are offensive, you are welcome to them and you have the right to them, but on a board like this, mb temper your opinions a bit. we are not 13 yr olds and you are not our father. you are not in any position to lecture.&quot;

Yeah, I should just listen to those who call me a &quot;troll&quot; because they don't like what I write. Tough luck.

&quot;a little tact would be in order.&quot;

Practice what you preach, mother.

&quot;also, when you go off on www.overclockers.com at least have your facts straight. Where in that whole article was the Thermoengine mentioned and in what context??&quot;

Where are you getting ThermoEngine? Did I mistakenly use that in a previous post? The latest article was the SK-6, Glaciator, and 462-A, no ThermoEngine. If I did indeed use the TE, my mistake.

&quot;finally, why should we respect your opinion any more than what we're getting from overclockers.com??&quot;

Perhaps you shouldn't. I'm just pointing out that they've had serious problems with their &quot;tests&quot; before, and admitted to inaccuracies. Your mileage may vary.

&quot;you have to earn your respect before you go off like that.&quot;

<cough> Alright, father.

 

[Pabster]

Mikewarrior2 wrote:

&quot;And no, I don't agree that heatsink-thermocouple testing is the best. Shifting a thermocouple from center to 1/4inch toward the bridges on a t-bird will drop the reading several C as you move over the L2 Cache. It will become apparent as soon as windows programs that read the Athlon MP internal diode become available that internal diode readings are still, far and away, hte best readings possible.&quot;

1/4 inch is a vast movement on a die as small as the Bird. When I said a small deviation in probe placement wouldn't affect the numbers drastically, I wasn't talking a quarter inch. Perhaps when such a program is available (or much better, when consumer mobos are able to read the diode) we'll be able to settle the temperature game a lot easier. One can only hope.