I have read quite a bit about this and it seems that there are a lot of different opinions on this

What do you recommend?

The CPU fan blowing towards or away from the heatsink?

ditto, draw it away, and make sure you get a nice feed of clean air in there.
You want the coldest air closest to the cpu, so if it draws it bottom up, it'll work better than top down.

Aree, it should blow down to the processor letting sucking the case air down and over CPU circulating cooling chipset and mem also.
My LianLi PC60plus has an intake at rear, with a duct pulling in cool outside air and blowing it towards top of CPU fan, so it can then pull cool air down to the CPU.
So down it is.

Thats interesting. Thats the reverse of the way it is normally done. However, either will work, depending on your setup. on my rig, I have seperate VRM cooling, and the best thing I can do is pull the air away from the Chip to the rear exhaust fan.

(I believe most case circulate air in through the front and out the back. in that case you want to pull away, so you are not pulling hot air across the CPU).

The most important thing to remember is to move air from the coolest source you can get to the nearset exhaust you can get. On different boards/cases, you will get different results, though I still generally recommend pulling air away.

one thing to note, as rodney_ws stated, it that you are probably only going to save a degree or two by switching direction, which in most cases won't matter. It may play a little into OCing, depending in the rest of your setup.

LOL some people crack me up. Your stock CPU fan will not work well you reverse the rotation. PLEASE notice the fan blades are CURVED NOT FLAT. This increases air volume so that lower rotation speed can be used. If you reverse direction alone you are hurting performance of the fan. CPU coolers always blow across the the heat sink to maximize coverage area. If you suck air away from the heat sink you lose a lot of area for reason that air always take the least line of resistance and will only pull through the outer most fins thus overheating issues.

PC-60plus also has a 120mm fan blowing out rear, along with 80mm fan blowing duct to CPU for cool air.
Damn good cooling with LianLi PC-60plus case.

Well, it's obvious that the first two guys to jump in know nothing about what they're talking about. Of course I have experience and training, yeh, I went to school for this stuff. Sorry!

OK, so the first problem with fans is that they pull air from the sides. That means a fan pulling air through the cooler works less effectively than a fan blowing in. Many of the guys here would say that a pulling fan has a larger "dead spot" in the center.

The second problem relates to the first, since the fan pulls air from the side the only part of the cooler getting good airflow in the "pull" configuration is the top. Not the bottom, where the CPU is. Bad idea.

It's such an incredibly bad idea that Intel, that's the company with all the funky engineers, specifies their own designs, all of which blow into a cooler rather than sucking out of it.

Now there was this cooler maker nobody remembers, I think they're still around, called Alpha. They thought pulling was better, but realized that fans pulled air from the sides. So they installed a shroud around the top half of the heatsink, so that air had to be pulled from the bottom. A funny thing happened: Even though the sinks were well designed and had a lot of surface area which optimized cooling, further optimization was found by REVIEW SITES who reversed the fan, to blow inward.

This fan shroud is really obvious to dumb mechanics, as many hot rods have been found to overheat when you remove the fan shroud. Yet as previously stated, you get better results from pushing air.

One reason it's better to push air of course is the directional thing previously mentioned. But another reason is...air doesn't pull very well. It's a gas. It likes to expand. Pulling just lowers the air pressure.

So it's easier to reach the base of the cooler from a push configuration not only because it's directional, but because the pressure also helps it reach the bottom of the cooler.

All this has been discuseed numerous times of course. And there's other reasons why push works better, such as the air being cooler as you move away from the board. But when it comes down to it, you could simply search the forums and find this all explained in more detail in old posts.

And then you'd also find guys who were really good at arguing for the pull configuration...loosing. These modern geeks are too soft for that kind of debate.

Of course you're right about configurations changing results, but I've always seen a measurable difference no matter how slight.

This is very interesting. I always assumed the fan was pulling air away from the heat sink, toward the exhaust fan toward the top and back of the computer case... and new cooler air would seep in through the sides of the heat sink. I was always under the impression that the general goal was to keep the air flow in the computer case streamlined, like in those smoke chambers used to test airplane wings. Thus the heat from the CPU would be caried out of the case in a smooth efficient current of air.

Blowing the fan in toward the CPU seems like pointing a fan toward a wall. It will bounce the warmed air in all directions causing turbulence in the system, not necessarily toward the top/back of the case.

I'm not saying this is wrong (or right), I'm saying it is counter-intuitive... at least from my mindset.

Hi,

Coming to this post a little late but have been looking into this myself and want to chuck in my tuppence-worth for anyone who comes after. My first post here too! I finally decided to register as I need to ask for assistance with some other stuff but that's another story.

I have an old ATX case. the only fans are the PSU, GFX Card and CPU fans. My processor is an Athlon Thunderbird (1.1 GHz) with a basic aluminium (?) heatsink. The heatsink used to have a generic crappy fan, which blew outward, until I replaced it a while back with a quiet Acoustifan job to reduce noise levels. Without thinking I put it on the same way, i.e. blowing away from the CPU.

I'm a little more versed in PCs these days and wanted to try a bit of OC'ing, so I upped the FSB in stages and then ran CPU Stability test until it crashed. The most I found I could get was 109MHz without errors - pretty low OC really! Anyway I ran MBM5 whilst testing the chip using the CPU Warming test and noticed it was up at 58C - pretty hot but seemed stable enough. Then I had a bright idea and turned the PCU fan around. I ran the test again and MBM5 showed 54C - a 4C difference!

I belive the fact that I have not been able to raise the FSB speed any further is due to hardware limitations rather than the CPU temp (mobo is the main suspect - has bad caps, chipset seems to have IDE trouble too) but this experiment certainly shows that, for low end computers at least, a simple change like this can have a big effect. At the very least this should extend the life of the chip a little so worthwhile doing!