HardOCP Thermal Paste Test

[akugami]

IC7 seems like an interesting product. I don't know if I'll use it, I still have a good sized tube of Ceramique and for my modest overclocking uses (most taxing is video encoding/transcoding), it's not critical or even necessary to eke out another 100mhz.

 

[BonzaiDuck]

Originally posted by: akugami
IC7 seems like an interesting product. I don't know if I'll use it, I still have a good sized tube of Ceramique and for my modest overclocking uses (most taxing is video encoding/transcoding), it's not critical or even necessary to eke out another 100mhz.

Sure. I chose the diamond paste because I initially wanted to "go after the last grain of rice" for air-cooling. After that, I just chose to stick with it.

 

[chizow]

Ease of application is why I've stuck with AS throughout all the different versions as well. That and the fact one of their 3 oz tubes or whatever last me a year or two. I've tried some of the other pastes, but they're much harder to spread evenly, like a mix between peanut butter and gum. Even though results often show 1-3C difference, I know from first-hand experience a bad application of TIM can and will often result in a much bigger temp delta than that potential 1-3C gained. With AS5, the only way you can really mess it up is if you use *too much* of it, especially if you've lapped both the IHS and heat sink.

 

[Idontcare]

Yeah no kidding on that. AS5 is easy to do right and easy to know when you've done it wrong.

I used TX2 just to check it out once and it was like you say, peanut butter, and when I pulled my HSF because of temps I saw that the stuff never squashed out flat on the IHS to cover the whole IHS, instead it was like a pancake and real thick too. (even with my tuniq screws turned down to their stops)

 

[nyker96]

everything looks pretty decent and similar. i like the one that cost just $10 for 2oz. probably can be used forever.

 

[OCGuy]

Let me post what Aigo is going to say before it is even said:




"Who the hell uses air cooling????"

 

[Idontcare]

Originally posted by: OCguy
Let me post what Aigo is going to say before it is even said:




"Who the hell uses air cooling????"

Aigo does, all his rads are air-cooled :roll:

(couldn't resist, you get a free "hit IDC with a wet-fish" card, redeem at any time, never expires )

 

[corei7965]

IC Diamond is the way to go. I have an 80w pletier and use stock fan. my processor is just about taped out and under full load I,m at 64c with ic vs 73 with as5

 

[Zap]

Originally posted by: TidusZ
I replaced the TIM on my BFG Gtx 295 with OCZ freeze and load temps dropped ~10c. The TIM on my gpu's was caked on really thick, certainly not the thin layer on those water cooled cards.

The water cooled cards are hand built here in Lake Forest, IL. The air cooled cards have the compound applied by machine.

 

[DrMrLordX]

Hmm, methinks they need better machines if the hand-built cards have superior TIM applications.*

*: so long as Skynet/Cylons are not involved.

 

[Idontcare]

Originally posted by: DrMrLordX
Hmm, methinks they need better machines if the hand-built cards have superior TIM applications.*

*: so long as Skynet/Cylons are not involved.

Better machines = higher cost

 

[DrMrLordX]

True, at least at first. Mechanized labor allows them to reduce/eliminate human labor which can be a money-saver down the road, but only once better-quality automation is available at the right price-point.

 

[TidusZ]

Or they could just do it properly by hand on all videocards over $600. I could take the card apart, replace tim, and put it back together in an hour. Much less if I didn't have to clean off the old stuff or take the card apart. Maybe 15 minutes to apply the tim and assemble the card. When the price is $650, it's not unreasonable to add this 15min-1hr job that makes a huge difference to the end user (whose obviously buying a top tier card because they want the best, and know hardware to a degree). Even at $25/hr it's not gonna make a huge difference in the end. I would of paid a $20 premium rather than have to do it myself, which I did.

 

[DrMrLordX]

If you're talking about a low-volume part, perhaps so. Still seems kind of sad that the machines can't do a better job than that. And the general trend here is that mechanizing the process and doing it right would cost more than human laborers would. If anything, paying a guy $25/hr or what have you is generally considered to be a cost-cutting measure.

It would really be interesting to see what sort of machines are used to apply TIM etc. on the cheaper parts.

 

[Blain]

Originally posted by: BladeVenom
After looking at the results, I guess I'll have to stop using cheese as a thermal interface material.

You could switch to Toothpaste or Vegemite.

 

[Idontcare]

Originally posted by: DrMrLordX
Still seems kind of sad that the machines can't do a better job than that. And the general trend here is that mechanizing the process and doing it right would cost more than human laborers would. If anything, paying a guy $25/hr or what have you is generally considered to be a cost-cutting measure.

No one is disagreeing with you, but we are saying the existing (already paid for and depreciating) machines simply don't. Replacing them would merely add cost (the cost of the machines) at a questionable probability of being able to raise prices to recapture the investment.

If consumers were shunning the parts because of the TIM issue then it would be sales loss issue, and the investment would be billed as a necessary enhancement to their existing production line.

Or if someone in marketing could convince a capital expenditure decision maker that upgrading the equipment would result in the creation of a more desirable (i.e. shifts the demand pricepoint to the right) end-product from which the estimated increase in ASP is expected to pay for the equipment upgrade plus deliver an ROI which is competitive with the other existing/known ROI opportunities for the same capex allocation.

A new company that might come onto the scene would not burden their decision makers with the task of proving ROI for upgrading existing antiquated equipment (because they wouldn't have any to begin with) and this is actually a competive advantage to new comers in the industry segment.

No one is saying equipment doesn't exist to do the job, nor are we saying that the equipment would cost more than humans given a threshold minimum volume, but there are standard GAAP metrics of capex replacement that get in the way when the consumers aren't providing the incentive (i.e. reducing demand) to the company's decision makers to make investments in themselves.

Sidenote: BTW this is exactly how/why the Japanese steel industry upended and routed the established american steel industry in the 1980's.

Japan's primary disadvantage, its lack of raw materials, was overcome through the building of huge vessels designed to carry large amounts of ore for low cost, building deep harbors and steel mills on the coast, and utilizing the most advanced technology. For example, 86 percent of Japanese steel is continuously cast-a process that saves energy and raises productivity-compared to only 26 percent in tile U.S.

A Political History of Steel - published 1983

 

[Gillbot]

I use Ceramique. It's cheap, stable and works very well.

 

[DrMrLordX]

Yeah, that all makes sense. Having a vested interest in older technology can make upgrading a difficult expense to justify.

It would be interesting to see what machinery is used in the fabrication of the physical cards themselves.

 

[Zap]

The other thing with the machine-applied thermal interface is that it is "enough" for the product to function properly. I know as enthusiasts we want to improve on stuff, but for the gamer just tossing it in a system and running it stock, the TIM works just fine.

With the GTX 295 in particular (referencing an earlier post) the thickness of the TIM and those white pads on memory and grey pads elsewhere is a part of the assembly. If you take it all away and put a super thin layer on, your card will actually be "thinner" (closer together). The problem is that then the PCB will have to bend a hair for the bracket.