Swifttech MC462A or GLACIATOR?

[heffe734]

got my swiftech...i'm happy!

 

[mikef208]

Pabster, so you say the Switech MC370 will beat out the Taisol copper bottom? I'm just asking cause in the couple of reviews I have seen it doesn't and thats with a crappy low profile delta on it. Have you tested them both? I am trying to decide on a HSF and I had already tossed out the MC370 cause of what I saw on it's performance, but if you say it's better I may include it again.

 

[Pabster]

mikef208:

Yes, the 370-0A will beat *any* Taisol, and I don't care what fan you couple with it. I like the Pabst ~4900rpm that ships with the retail units, but you could use another. If you'll notice, 9 out of 10 of these "Heatsink Roundups" conveniently forgets to include the MC370-0A, and often "forgets" to use the 462-A. I wonder why?

 

[clarkmo]

Fan clips and Delta screamers. Not an issue.
That is if you make a muffler for your fan.
I make mine out of plasticboard for the frame and a combo vibration damper/acoustical foam for the liner. I'm still trying to settle on 1/2" or 1/4" foam. I make the duct slightly undersized ao the Delta can be slid in about halfway and held securely. A guesstimate would be tha sound is halved in volume. Annoyance is decreased about 95% though, I haven't tried it with the case closed because it's 5" long and the case won't close. You know the fan is still there but the annoying whine is gone. It's like the bathroom door closing when someone's using a blow dryer, but better cause the highest pitches are trapped by the foam. And my temps improved, probably by the fan being off the heat sink by 2" but the airflow still being directed by the duct. My project will be complete when I curve or angle it enough to be able to close the case and not impede air flow. Velcro strips hanging from various case appurtenances hold it in place.
Anyone else trying something like this?

 

[Turin39789]

if you are spending swiftech type money, why dont you look at getting a noise control silverado. Very quiet, but on par with everything but a silverado/glaciator with their loud deltas

 

[Courtland]

the mc462 is the best hsf out there, no question, just, can you handle the noise and the price? that basicaly sums it up.

 

[LASHER555ZX]

I don't trust Overclocker's.com review of the GLACIATOR. I would buy the MC462A cause of its mounting setup too. 780 grams on ur socketclips is scary. Its already proven itself to me.

 

[Insane3D]

I think it's funny you guys are comparing reviews that use the socket a in socket thermistors. They are not even close to measureing actual core temps and are usually off by at least 5C. The measure the CPU temp from the bottom of the CPU ceramic. They are taking a measurement through quite a few layers of it. Also, you guys are not taking into account the superior mounting system of the Swiftech. I would never attach one of those Glaciators with a clip..what if that little plastic tab snaps..it would take out some of your hardware on the way down.

 

[mikef208]

Insane3D, overclockers.com used the Socket A thermistor, but they also used a temp probe at CPU/HS junction, so they have a little more accurate temps, but since they are reviewing all the sinks in question on the same board, it shouldn't matter anyway.

 

[Insane3D]

"but since they are reviewing all the sinks in question on the same board, it shouldn't matter anyway."

Actually, it will. The in socket thermistor is affected by airflow around the CPU socket. Each HSF unit will have different fans and/or different air flow characteristics. For instance, the super orb's directed air straight down towards the CPU socket, and because of this, they appear to perform very well in tests. However, this is not the case, as they actually do pretty bad on socket A chips. I have witnessed the airflow factor myself on my personal system. I also use a "Compunurse" to measure temps at the core/HSF junction as well, and my in socket thermistor on my KK266-R used to always read an average of 5C cooler than the core. I had a low speed Sunon 120mm (~60CFM) blowing on the CPU area in the side of my case. I replaced the Sunon with a 130CFM YS Tech..now my temps are almost 10C cooler in the socket. I am impressed however that they used a thermal probe at the core in their tests..thats the first one I have seen to date. I have no doubt the Glaciator performs well, I just feel that a HSF that is that heavy should not be mounted via a clip where you have all the weight of the HSF on one small plastic tab on the CPU socket. Another thing to consider is that when you stand the motherboard up...like in any tower case, heavier heatsinks that are attached with a clip will actually pull away from the core surface just a bit due to gravity. My Hedgehog will keep my CPU around 2C cooler if I lay the case on it's side. I would imagine the heavier the HSF is, the worse this would be. That's why the Swiftech mounting sysytem is so much superior.

 

[MGTasco]

<< I don't trust Overclocker's.com review of the GLACIATOR. I would buy the MC462A cause of its mounting setup too. 780 grams on ur socketclips is scary. Its already proven itself to me. >>



Overclockers.com method of testing heatsinks is one of the most accurate I have seen. In my own test using digital thermometers and with thermocouple to the core, most of my calculated thermal efficiency C/W matches very closely with Joe Citarella's (overclockers.com). For example for Gladiator, he got 0.23 C/W, where else I got 0.22 C/W. So performance wise, I have no doubt that the Glaciator is a pretty good heatsink.

The only thing that have to worry about is weight. As someone just said earlier, the temp. can have a difference whether is the case lying horizontally or vertically. But a good design of the clip can eradicate most of these problems.

I have ordered the Glaciator, will keep you guys posted on its performance. I don't use on board thermistors for test, which means I will verify the results on overclockers.com and see if it is correct or not.

 

[Insane3D]

&quot;The only thing that have to worry about is weight. As someone just said earlier, the temp. can have a difference whether is the case lying horizontally or vertically. But a good design of the clip can eradicate most of these problems.&quot;


Actually, even the best clips, like the Taisol's will still be vulnerable to this. A single clip pressing down on a small area in the middle of the heatsink will never completely avoid this problem, it will just be less of an issue with a good clip. The four spring-loaded screw mounting system is just a better solution with the heavier HSF's. It will give you better contact with the core since the clamping force is even across the surface of the HSF instead of just the middle with a clip. Another thing is over time, a clip will &quot;stretch&quot; and slowly lose it's clamping force little by little, where four screws are static and will not be affected by this problem. I am really surprised more HSF are not using this system..it is just superior to a clip mounting system. The one downside is, of course, you must remove the motherboard from the case to mount it the first time, but I think the extra effort is worth it.

 

[frankroh]

Who sells the Silverado? I know it's a great HSF, but I can't find a single vendor....

 

[MGTasco]

<< Who sells the Silverado? I know it's a great HSF, but I can't find a single vendor.... >>



You can check it out here :

https://secure.webstrikesolutions.com/chillblast/docs/start.htm

You have to buy it from UK though....not really a big deal considering I bought it all the way from Singapore.

 

[MGTasco]

<<

Actually, even the best clips, like the Taisol's will still be vulnerable to this. A single clip pressing down on a small area in the middle of the heatsink will never completely avoid this problem, it will just be less of an issue with a good clip. The four spring-loaded screw mounting system is just a better solution with the heavier HSF's. It will give you better contact with the core since the clamping force is even across the surface of the HSF instead of just the middle with a clip. Another thing is over time, a clip will &quot;stretch&quot; and slowly lose it's clamping force little by little, where four screws are static and will not be affected by this problem. I am really surprised more HSF are not using this system..it is just superior to a clip mounting system. The one downside is, of course, you must remove the motherboard from the case to mount it the first time, but I think the extra effort is worth it. >>



Yeah..I do agree with you. A good clip design can avoid some problems but certainly not everything. The screw mount design would be the best..but again it will come with a price.

 

[Renob]

Well i just installed my Glaciator and it kicks a$$ here is a link

 

[Mikewarrior2]

MGTasco,

Overclockers.com testing is not as accurate as most people think, nor is it as accurate as they think. NOt only do they call their temp measurement CPU temp and even call it a true core reading, when it clearly is not.

What it really is a heatsink-junction C/W. And it is severely undercutting manufacturer C/W. It is accurate within the scope of their reviews, but using their C/W measurements to determine DIE temp is impossible. USing their C/W to determine Heatsink-Junction Temp, fine.

Yes, there are other problems. When you're talking about an external measurement, maintaining accurate placement of the thermocouple is absolutely critical. When most of their heatsinks are in teh same .3-.4 range of numbers, making sure they are over the exact same part of the core will definately have an effect. yes, different portions of the core run hotter. And heatsinks do have some &quot;shift&quot; available in them, and some of them have enough leeway to shift 1/4inch either way away from the &quot;center&quot;.

And regarding the swiftech, you can't compare swiftech readings to the other heatsinks. They drilled the thermocouple in from the side. It is farther away from the core than the other readings. And even more of a heatsink-temp and less cpu temp.

NO, i'm not saying the Swiftech is a bad heatsink. It is the damn best clip mechanism and damn good performance. Its just you can't compare that review to the others they do.

Thermocouples mounted to the core will also result in lower than full die temps, since you're not solely in contact with core. Again, if they match heatsink-thermocouple, they are roughly 50-70% of the core temp rise over ambient.

Going by AMD's Athlon MP pdf file, they suggest using the Palomino internal diode in conjunction with a heatsink thermocouple to determine DIE temp. This makes total sense since the internal diode is in the core, and in between it and heatsink thermocouple is the DIE temp. So guessing from their current testing, you're going to be looking at, wtih most heatsinks, DIE TEMPerature being 50% higher than even heatsink thermocouple readings.


Mike

 

[clarkmo]

Nice reply Mike,
Makes a lot of sense. It does appear that you can take a consensus of reviews and look for some consistency in the results to try to determine which hsf is better. I have not found a bad review of the sk6 anywhere. It always seems to pull out on top. I haven't seen a straight up comparison with the mc42, though. I think the one early comparison I read shows the Swiftec's results from an earlier test. Not a lot of good comparisons out there with the mc42 Maybe Hardocp's next shootout will have this. At any rate the sk6 from Millisec is next for me. I don't think I can go wrong with it,whether it's the best or not I'll leave to the reviewers.

 

[MGTasco]

<< MGTasco,

Overclockers.com testing is not as accurate as most people think, nor is it as accurate as they think. NOt only do they call their temp measurement CPU temp and even call it a true core reading, when it clearly is not.

What it really is a heatsink-junction C/W. And it is severely undercutting manufacturer C/W. It is accurate within the scope of their reviews, but using their C/W measurements to determine DIE temp is impossible. USing their C/W to determine Heatsink-Junction Temp, fine.

Yes, there are other problems. When you're talking about an external measurement, maintaining accurate placement of the thermocouple is absolutely critical. When most of their heatsinks are in teh same .3-.4 range of numbers, making sure they are over the exact same part of the core will definately have an effect. yes, different portions of the core run hotter. And heatsinks do have some &quot;shift&quot; available in them, and some of them have enough leeway to shift 1/4inch either way away from the &quot;center&quot;.

And regarding the swiftech, you can't compare swiftech readings to the other heatsinks. They drilled the thermocouple in from the side. It is farther away from the core than the other readings. And even more of a heatsink-temp and less cpu temp.

NO, i'm not saying the Swiftech is a bad heatsink. It is the damn best clip mechanism and damn good performance. Its just you can't compare that review to the others they do.

Thermocouples mounted to the core will also result in lower than full die temps, since you're not solely in contact with core. Again, if they match heatsink-thermocouple, they are roughly 50-70% of the core temp rise over ambient.

Going by AMD's Athlon MP pdf file, they suggest using the Palomino internal diode in conjunction with a heatsink thermocouple to determine DIE temp. This makes total sense since the internal diode is in the core, and in between it and heatsink thermocouple is the DIE temp. So guessing from their current testing, you're going to be looking at, wtih most heatsinks, DIE TEMPerature being 50% higher than even heatsink thermocouple readings.


Mike >>



Thanks Mike for the explanation. Just wanna clarify a few things. In overclockers.com, they say that they drill a hole through the centre of the heatsink and place the thermocouple there touching the top of the core. Is this what you consider heatsink-junction temperature?

But looking at the current situation now without any internal diode in AMD chips, wouldn't you consider that overclockers.com method would the most accurate compared to most other websites which basically just uses the mobo thermistor? I mean..yes..overclockers.com may not be accurate enough, but isn't it the closes you could get looking at the condition of AMD chips now?

 

[Insane3D]

&quot;But looking at the current situation now without any internal diode in AMD chips&quot;


Don't forget the new Palomino core in the mobile Athlon 4, the desktop MP, the upcoming offical desktop Athlon 4, now have an internal thermal diode like Intel's chips. I've got a 1.2ghz Palomino/MP on the way...