Thermal Paste Shootout

A thermal paste shootout is something that takes a lot of time and effort to do correctly. I've tried my best to obtain the most accurate measurements possible given the timeframe. Being a bit conservative, I estimated the testing in this review to take 2 weeks, compiling the results to take a few days, and the whole project to be done in less then 3 weeks. The reality was quite different. For example, I tried to keep the temperature within +/- 1C of the initial room temperature for all tests. Every time, the temperature deviated by more then 1C I redid the test. So each time the temperature changed I had to wait till it was back to normal before I could test again. It's too bad I don't have an air conditioner. I also obtained a few questionable results, so I reapplied the goo. This resulted in better readings. All in all, I've learned from my mistakes along the way. The final results are, I hope, quite accurate and I believe they will be of much use to readers.

For this shootout, I tested each goop for 3 days recording the load temperature each day. I used seti@home to keep the load constant at 100%. At the end of the test, I recorded the idle temperature after a cool-down period of 30 minutes at no load. It didn't seem necessary to me to record more idle temps because I'm simply more interested in the load temps. For the testing I used MBM 5.

I spread all the pastes in a very thin layer and used the same heatsink each time, a Global Win TAK 68. I chose this heatsink because of the ease of installation. Some heatsinks take a long time to remove and reinstall, the TAK 68 took less then a minute total. All load measurements were taken between 6-8pm. The cpu was a 1700+ DLT3C AIUGA running at 166x12 or 2Ghz.

Note: This review will stick to the pastes' basic composition and the properties I observed in testing. For more detailed information about the paste, read the PR on the manufacturer's sites.

The first paste I tested was the white paste that came with my sk6+. I was hoping it was just generic paste and would serve as a control group. It actually managed to hold its ground with the other pastes quite well which surprised me. It looked like regular silicon paste and spread quite well. Application was easy and so was removal. The paste came in a syringe, something uncommon for regular white paste, and thus was easily kept for later use. This was helpful, since there was enough for many applications. I would guess that it is nonconductive, since it appears to be silicon paste with no metal in it, but I can't be sure. I would be a bit concerned about long-term use without reapplying, since silicon-based pastes tend to dry out over time.

The second paste on the test bench was Geil Thermal paste. It is a copper/silver mixture, which I expected to do well. Both copper and silver are good conductors of heat, though silver is the superior conductor. If the paste was composed of only silver, it likely would have done better. This brand also came in a syringe, though it was not as big as the other syringes. There was easily enough paste to satisfy most users (10+ applications), but people who regularly change their cpu may want to opt for a product with more volume. It was a bit thicker then the rest of the pastes and quite hard to spread. It has a gold-like color, which is kind of cool, not that you see it once its under the heatsink anyways. If you choose this brand, I hope you do like the color; if you ever try to take it off, your fingers will end up colored gold. This paste was an absolute pain in the butt to remove, and it got everywhere when doing so.

The third paste on the bench was OCZ Ultra I, which is now discontinued. OCZ Ultra I is also a silver-based paste, and was quite easy to spread on. The syringe in which it comes holds more paste then the previous two, and made for easy application. It's a very fluid-like paste, and was very easy to remove. However, I wasn't too fond of this paste; it seems not to be mixed correctly, as I could see it separating a bit. Judging by its appearance, its oil-based, and the oils weren't staying mixed.

The fourth paste I tested was Arctic Silver Ceramique. It was also a pretty small tube; about the same size as the Geil tube. It was also a syringe, but by adding a round rubber cap on the end, it seems they tried to transform it into a phallic symbol... Maybe this will tempt people to buy it... Anyway, Ceramique was a polysynthetic compound, ie. no silver or other metals. This means that it is electrically non-conductive, which is an advantage. Another benefit of Ceramique is that it is not silicon-based, so it won't dry out. This brand was quite gooey, so it was easy to spread on and to remove.

The fifth paste is OCZ Ultra II, a micronized silver/polysynthetic compound. It also did not contain silicon, so it wouldn't dry up like plain white paste. Similar to Ceramique, it is not electrically conductive, which is a definite bonus. It was a little thicker, though not as thick as the Geil paste. Like the Geil paste, it was very hard to get off your fingers or anything else it comes into contact with. Luckily, I had a few weeks for the Geil to wear off before I tested this. My fingers look better when they are all one color. ;) Application was easy, and the best part was that the syringe was huge. You get a bit more compound, though not as much as the size of the syringe would suggest. According to weight, it contains approximately 3g, compared to Ceramique's 2.7g, but there was quite a bit more in this tube than there was in Ceramique's.

And now for the results, (Pardon the text format, a chart with this same info will be added soon):

Sk6+ Paste

Day 1 57C
Day 2 57C
Day 3 57C
Day 3 Idle 51C

Geil Paste

Day 1 59C
Day 2 57C
Day 3 56C
Day 3 Idle 51C

OCZ Ultra I

Day 1 58C
Day 2 57C
Day 3 56c
Day 3 Idle 52c

Ceramique

Day 1 57C
Day 2 54C
Day 3 54C
Day 3 Idle 48C

OCZ Ultra II

Day 1 58C
Day 2 56C
Day 3 54C
Day 4 Idle 47

Conclusion

The results basically speak for themselves. The biggest difference in load temperatures was 3C; idle was 5C. Many people care enough about shedding a couple of degrees to go out and buy the pastes. If you do, I would recommend Ceramique or OCZ ultra II. Ceramiuque and OCZ Ultra II had the lowest temperatures in both load and idle. But it is due to the lack of conductivity, and because neither one will dry up, that I would really recommend them. A tube of good thermal paste will last you for quite a few cpus and maybe a few video cards too. I would rather spend a bit more to get a non-conductive paste then fry my cpu for minimal savings. As to which one is the best, it's pretty much a tie. OCZ Ultra II had a slightly lower idle temperature, but Ceramique was easier to remove. So when making your choice, base it on which is more readily available to you, and at what price. The last time I looked into it, when you buy OCZ ram from ramstore.ca, they include a free tube of OCZ Ultra II. That's how I got my OCZ Ultra original! I'd like to thank OCZ, Arctic Silver, and Geil for supplying me with samples. I purchased the sk6+ paste and the OCZ Ultra I.

Added: September 27th 2003
Reviewer: Mixam
Score:
Hits: 21869
Language: