dean's musings...



My post to an "advocate forum" regarding the discussion and debate about "premium" thermal compounds (the "goop" you put between a CPU and it's fan).Written for no other reason than the heat of inspiration from hearing what seemed silly claims about pastes. the post Garnered a lot of response...

Before I post this, let me just say that I'm dousing myself in gas to make the flaming easier for you. Just a spark will get the job done.

Arctic Silver III is fine to use, I have a tube of it here. I also have some silver compound from Antec, and that's fine too, as is my Generic Thermal Compound I bought once when in need. I use them carefully -- when I need to mount my HSF I open the drawer where I keep this stuff and pull out the first one that catches my eye. But I'm careful not to get my fingers caught in the drawer when I close it. I also have some blue Gel Elmer's Glue that I'd use if needed, but it's not a favorite because it's water-based, and would dry out in short time. Until it dried out, though, the cooling effect would be the same as ASIII. (Perhaps even better, water's an excellent thermal conductor in general).

My results, using any of the above (ignore the Elmer's though) are completely consistent. Using paste makes my CPU run cooler, but in a double-blind test I dare say I wouldn't be able to measure the difference using any of them.

The purported claim that ASIII, or any "silver-based" compounds work better is supported largely by the manufacturer. No surprise there, they're selling the stuff to make a profit after all, and they don't mind charging a hefty premium when people are willing to pay for it. I remember seeing a blister pack for one compound that actually said (from memory) it performed "200%" better than other compounds, and had a graph showing just that. Not to matter the graph wasn't zero-based, and the columns started at some arbitrary reference like "41 deg C". The performance of that silver compound showed a max CPU temp of something like 42.5 deg C, whereas that same CPU when using the "other" goop, jumped to a "startling" 44 deg C! Work the math and you see you can find a value of 200%, if you're creative. Of course, you have to assume the 41 dC baseline was based on something real, and I suppose they COULD have done their testing in a warm oven. But if they did it in a freezer the difference would come out to less than 4%. They didn't pester us with those nitty details, however.

(I don't actually remember the numbers they used. But the "spirit" of the comparison is accurate, and I'm not being nearly as, uh, creative here as the blister-packs claims were).

So who wouldn't pay the extra premium for a product that's "200% better"?! I sure would, especially when you read that is doesn't just use silver, an expensive and excellent conductor on its own, but that it's "75-80% silver by weight". Obviously you're getting almost pure silver for your money! Of course, if you wrapped a silver ingot in 100 yds of tissue paper it, too, would probably be at least 75% silver "by weight." And it would be a terrible conductor of heat, but I don't think you'd read that on the fine print . Oh, and it's not just plain silver you're buying, it's "99% pure micronized silver". Apparently they shape the little pieces of silver (99% of them) so they fit better together and transfer heat better as a result. Sounds impressive, without a doubt. It's also "negligibly conductive" which is a good thing; I doubt anyone here would willingling stuff their mobo with steel wool to wick heat and expect a "positive experience". Sparks, maybe. Of course, from the figures you assume you're getting essentially "all silver" in a paste form (75%), so you're practically "guaranteed" a high-content paste that relies on silver to improve the thermal conductivity. On top of that, the suspending "paste" isn't just something they found lying in the garage, it's a "controlled triple phase viscosity" paste. Who wouldn't flock to that? The old "uncontrolled double phase" stuff must pale in comparison.

But how that mostly-silver paste can be non-conductive eludes me. A metal that selectively avoids electrons but can still wrestle heat from molecules (while being only "slightly capacitive", maybe that's the hook).

Based on all that, it's a pretty compelling motive to pay the extra $$ for "200%" improvement because of that "micronized" (99%, no less) "pure silver". You'd be a fool not to!

And there are plenty of reviews to show the premium compounds are superior to "plain" compounds (you don't need to send me the links, I've seen most of them, thanks!) Of course, you generally need to ignore the test methodology used in determining how effective ASIII, or Antec, or whatever-they-tested, is. Usually they run some CPU-intensive program to get the CPU "hot", and measure the resultant temps using various componds. They put those in a graph and declare a "winner!" They don't actually measure the true CPU temp in a consistent fashion (they can't, the chip is safely "squashed" under the HSF with no room for a thermocouple), and it's a rare review indeed that includes the ambient temp while doing this (some do). A room temperature can change 5 dC if the heat should kick on, someone opens a door, or the sun starts angling itself in through the window. But there you go, and you still get to see the graphs produced from the test. What you really need to do, of course, is ensure a constant heat-source (a CPU isn't effective in acting in that way), and a known and controlled ambient temperature. It's probably wise to ensure the application of each compound is identical (same thickness, same positioning and pressure from the HSF, no lint or hairs trapped underneath, so on), and that's difficult to do hunched over a PC and using a credit card to slather "goop" on a chunk of metal only 10's of mm on each side. Without these controls you wind up "measuring what you measured," and if variables exist, they're reflected in those measurements.

So you no-doubt get my point: I'm, uh, skeptical about the efficacy of "premium" compounds vs. the run-of-the-mill ones in reducing temperatures. More than that, some of the false marketing ("200%" better) just irks me, not to mention the questionable "science" they're sold and evaluated under (in that light, you can buy mysterious and sealed boxes that plug into your home AC to "tune" the "harmful resonances" from the house-wide AC emanations and improve everything from your health to the speed of your computer. And there's "data" to back that up too, although I was disappointed to see they didn't see if wearing quartz crystals around your neck, or wrapping magnetic bands around your head improved the results).

On the other hand, if someone just feels better buying the premium paste, or feels a measure of security in knowing they're not buying something from the bargin-bin at CompUSA that's completely unknown, that's fine. Heck, I've done it myself! The additional cost isn't really a deterent, and the premium stuff usually comes in a better package, and some come with applicators and guides. So that's a bonus. The premium stuff may also hold up better and not dry out as easily, which would ruin most of its effectiveness.

BTW, I didn't just come up with this rant in a vacuum, and one review I thought was wonderfully done, as well as fun to read can be found here, just in case you didn't get enough already!

Better hurry, the gasoline's starting to evaporate...!