swaaye wrote:

Thermal paste effectiveness varies less than internet sentiment leads one to believe.

Great review! I share your opinion. The Russian reviews show that the Soviet КПТ-8 shows only slightly worse results than the top offerings, even though it is old as hills. By the way, I bought my tube in 2006 for 0.5$. It's only half empty.

This is why I just go by volume per cost anymore. Arctic Ceramique is the last stuff I bought. $7 for a huge 26ish gram syringe of it. The huge tubs of Cooler Master stuff available are really temptng

I saw the post about lapping on the first page, and wanted to comment on it based on my own lapping adventures over the years. In short, it's worth it some times and not others. Works like this...You can assume your cooler is going to be surfaced checked if its a name brand, quality cooler. This means it will either be pretty much dead flat, or will have an intentional (but oh so slight) concave surface. Either way you need to know and the manufacturer will say. It really comes down to the CPU and how well the IHS is bonded to the chip. If the chip is naked (eg old athlons, or you popped the IHS off) then you should avoid a cooler with a concave surface or you should lapp it. Sometimes the IHS can have a slight convex shape from the bonding process. This has been a on and off again thing with Intel more than AMD the past 7-8 years. Intel likes to solder/braze the IHS to the chip and it can cause this shape to form. Again, it happens with some versions and then gets ironed out in a revision or two. The slight concave shape of a cooler can help compensate for the sometimes convex shape of a poorly stuck on IHS.

I've found that if you have one of "those" chips, its best to lapp it lap the cooler on the same surface. The only exception I've made in the past are for Danger Den and Swiftech waterblocks, as their added concave shape is very minor and often gives a bit of an edge in getting extra clamping pressure. Zalman on the other hand tends to be a bit to extreme on the concave shape, and it often doesn't make full contact with the entire surface of flat IHS. This only seems to be an issue with their copper coolers. Their aluminum coolers are pretty much dead flat, but why bother with those on a CPU? They are only good for chipsets and whatnot. Same thing goes for Thermeltake coolers.

Anyway, as long as you know your cooler and do a TIM spread check to see if good contact is made then lapping is not needed. If you see that only about 2/3rd of your TIM is diffused across the surface of the IHS and cooler and then lapping will help if its done right.

As for the Arctic line of products, some are worth it and some are way past the product end of life cycle. Artic Silver was fantastic back in the late 90's till about 2004-5. By then other brands offered equal or superior options. The fact they they still sell it is kinda sad. There really is no reason to sell a TIM that needs a full break in cycle that can take a couple of weeks for the average user in this day and age. But Artic Ceramique is unique among most TIM's and can be considered one of the best value TIM's on the market if you know how to use it. When it comes to phase change coolers or LN and dry ice coolers its hard to beat it. Its one of the few TIM's that doesn't freeze under those conditions. I never would have bothered with dry ice cooling (quit the hobby a couple of years ago), if it wasn't for Ceramique. Every mount requires a pretty huge glob of TIM (which is the complete opposite of normal air and water cooling), and it needs to be easy to clean and cheap and remounts are common on the extreme side of things. Its also a good TIM for average joe installs as its cheap and works ok under normal cooling as long as you work it into both the IHS surface and the cooler's surface, wipe off excess and then mount with the standard grain of rice method. Not great mind you, but sufficient for non overclocking needs.

Nowdays I just stick with liquid metal TIM's and water cooling my modern rigs. I'm considering lapping one of my K6-2's and using a liquid metal TIM with a 1U copper cooler I have for it. Kinda curious if I can drop the temp a full 10 degrees. Its been a long time since I've seen a nice drop like that on air, by just "modding" the parts at hand. I think it might work though.

I played around with trying to get decent socket A cooling a while back. As I recall:

- a big Akasa copper-cored heatsink [786?] had poor contact with the cpu
- the cooler master aerocool - copper - was noisy and heavy but didn't cool better than a normal cooler
- a small, solid copper Qtec with quite short and widely spaced but thick fins, and a 50 or 60mm fan, did as well as anything
- dynatron copper [server?] coolers are ok if you can stand the noise [but ultimately I couldn't]
- a basic Speeze cooler was the best - nothing performed better in terms of temperature and it was quieter than the Qtec

I have bought a couple of slks now but have not tried to use them yet.

Nice work.

In my pursuit of lowering temps on my old 4670k I had spent quite a bit of time working on temps figuring what worked best.

My build at the time was..
Gigabyte GA-Z87X-UD4H
Intel i5 4670K
CoolerMaster Hyper 212 Evo
16GB DDR3 1600 Samsung mem
GTX570

My experience:
Hyper 212 Evo + Included CM paste = 94c
Hyper 212 Evo + AC MX-2 paste = 88c
Hyper 212 Evo Lapped (1k grit) + AC MX-2 paste = 84c
Hyper 212 Evo Lapped + AC MX-2 paste + 1.080v = 68c
Hyper 212 Evo Lapped + AC MX-2 paste + 1.080v + 140mm fan in 5.25" bay (intake) = 65c
Hyper 212 Evo Lapped + AC MX-2 paste + 1.080v + 2nd Blade Master (pull) = 65c

What I gained from this that worked best
-undervolting
-increase airflow channel front to rear
-use quality paste

Lapping helped some but only around 4c which isn't anything to sneeze at. Pressure modding a heatsink can help marginally (usually 2~3 at best if the mechanism isn't the best).

I've lapped more in the past and wouldn't do it for an old S478 build or anything at this point.