CPU coolers - Is this really the best technology available?

The copper tubes are heatpipes, there's a liquid inside. The liquid evaporates at the cpu end, condense at the heatsink end. the condensed liquid flow back to the cpu end via gravity and/or capilary action in a special material inside. it moves heat very very efficiently from one end of the pipe to the other

-Lasse

I've had good luck with Zalman coolers. Effective and quiet.

In case you were wondering, looks for a CPU cooler is a selling point these days. Many enthusiasts have at least partially transparent cases, so stuff even inside the computer has to look "cool".

It is just so much dorkage. It's like the trunk lid spoilers on FWD cars that do nothing at any speed the car is likely to operate and do it on the wrong set of wheels. Or the turtle wheels that make the car unstable 30mph and can't slow down as well as a yugo without turning sideways.

Those cpu fans are for people who have too much money and no clue what to do with it. Don't do anything useful with your computer. Just use it to run benchmarks and the defrag utility.

I always get a laugh out of overclockers. Get 15% more performance, and reduce uptime to less than a couple of hours. And then watch it be blown away by 50% in a few months when the next cpus are released.

Anybody wanna buy an 1802 overclocked to 3mhz?

(snip)

The solution is evident to me and has been overlooked by computer makers from the very beginning of PC-Time. The CPU is simply being mounted on the wrong side of the motherboard. Yes it is that simple. If the CPU were socketed on the "foil" side of the mother board, then the CPU could be pressed against a heatsink that is an integral part of the computer case. Conduction from the CPU to a heatsink having a large mass and fin surface area (for convection) would be all that would be needed. Mobile radio transmitters have used this design for decades. The heat generating component is bolted to a large exterior heatsink. No fans are generally needed. There would be less chance for failure and it would be a noise free solution.

--
Joe Leikhim K4SAT
"The RFI-EMI-GUY"©

"Use only Genuine Interocitor Parts" Tom Servo  ;-P

Modern procesors can dissipate more then 100 W. I think you would need an incredibly thick alu or copper case wall to pull that of. Also you need to guaranteee sufficient airflow on the side of that PC case if you do it like that. That can never be guaranteed, as a user may put it agains the wall. Normal convection, even with a big heastink, is not sufficient for

100W, you need a forced airflow. Remember, the fans in the PC keep the CPU temp very low, maybe 40 to 50 Celcius, at least in my PC. The thermal design is perhaps specified at about 75 degrees C. So for 100 W , 40 C outside, you have a play of 35 C, makes .3 degrees C per wat max, before the CPU dies. But for normal use (with 45 C CPU) and 40 C outside, you have 5 degrees play max at 100W, requires a .05 C/W heatsink....

Who cares what it looks like so long as it can move ~100W from the CPU? Heat pipes are the best technology for the job. Nothing else comes close. Some of them have "designer" look and price tags to match but there is plenty of choice in the works well but isn't pretty category.

Not quite. The better designs will beat the stock Intel coolers hollow and do not cost that much more. I finally gave up on one of mine on an older machine after it developed a nasty case of bearing whine that made it intolerable. The slightly more expensive Katana2 replacement is essentially silent and the fan barely has to spin at all to keep the CPU cool even when running flat out.

Some of the high end ones are capable of handling enough power that up to 90W or so they don't really need a fan at all. This is useful in a system intended for silent running as a media PC for classical music. They are ugly and physically big heatsinks but that is a small price to pay for very quiet operation. >

The rate of improvement in the real world is starting to slow. Unless you have a problem that splits nicely across multiple cores the newer CPUs are only a marginal improvement.

Regards, Martin Brown

rs

t

The lower theta jc heatsink allow for lower speed fans and thus a quieter box. If you buy a CPU in a retail box, you get a fan that will do the job, but with very annoying (loud) operation. The Zalman heat sinks are really good. A bit of a pain to install. I just replaced a failing gigabyte fan with a Zalman and had to pull the entire mobo. Still, I believe it was worth the effort.

Really not so difficult. It is commonly done with land mobile radios with 100 watt transmitter having 35% to 45% efficient design, dissipating 90 to 140 watts of heat. Duty cycle is a design concern of course. These mobile radios must work -30C to +60C per EIA/TIA-603, to meet vehicular installations, often a hot trunk. I doubt a home office user will require 100% duty cycle nor have the environmental conditions that would stress the design.

If you design the heatsink fins and their orientation properly, putting it against a wall is not a concern. Keeping dust out of the tiny fins in the heatsinks used on CPU's today is a much bigger problem.

The base plate for the motherboard used in my Lian Li tower is over 250 square inches. Between the base plate and the side cover is 1/2 inch of dead space. Imagine if that were all finned aluminum.

--
Joe Leikhim K4SAT
"The RFI-EMI-GUY"©

"Use only Genuine Interocitor Parts" Tom Servo  ;-P

I agree with the OP they are "Rube Goldberg and Dorky". They get jammed up with dust and the motors fail. Then the aftermarket sellers get add on business selling yet another overpriced contraption. The attachment to the CPU is dodgy at best and they cannot take mishandling, if you drop the cabinet, the attachment point gets stressed. I had a Gigabit MOBO and AMD processor at a previous job, that our "IT Guy" built out of overclocker hardware. It seemed to work ok unless I tried to use a bunch of CPU cycles on a propagation simulation, then I would get BSOD. It escaped us until one day he offered to install yet another huge CPU fan. He found the old fan heatsink was not contacting the CPU adequately due to the alignment.

--
Joe Leikhim K4SAT
"The RFI-EMI-GUY"©

"Use only Genuine Interocitor Parts" Tom Servo  ;-P

On a sunny day (Sun, 19 Jul 2009 17:18:18 -0400) it happened RFI-EMI-GUY wrote in :

Yes I have such a transmitter, with 150W pep on SSB... on my desk now. The left one in this picture:

But it has a large heatsink bolted to the bottom, with fins, and still it gets way too hot in my view, so I have a 12 V old PC fan next to it.

I doubt 'average user' will like the big fins sticking out, I also doubt you can get away without fins on that side plate. This transmitter is supposed to be mounted under the dashboard, there it would have space for air circulation. The real thermal test was a piece of plywood of 1 square meter with the test object in the middle, Very little convection, your PC would be dead in no time.

Yes, that would help.

A lot of games max the CPU. Older ones will max a single core, newer ones will max all cores. High-resolution h.264 video is rather CPU-intensive as well.

t to it.

Yeah, Integrating a large chunk of heat sink aluminum into the side of the case doesn't seem farfetched.

A few years ago when I first saw a press release for a water jacket cooler for a CPU thought it was a GAG press release! I laughed even harder when I found out that there was an entire specialty market for people doing such stuff.

Did they get individual Wintel processors to work beyond 3.73 GHz?

Is that where they exhausted the technology?

Peltier junctions inside the CPU?

-- Nick, KI6VAV. Support severely wounded and disabled Veterans and their families:

Thank a Veteran! Support Our Troops: You are not forgotten. Thanks ! ! ~Semper Fi~ USMC 1365061

Would simply make things worse. That heat still has to go somewhere and the Peltier adds that much more heat to the problem.

Your Rube Goldberg is I believe our Heath Robinson contraption.

I am not so sure though. The basic designs seem sound enough - a set of heatpipes in very good thermal contact with a lump of copper on top of the die leading to a set of fins over a wide area with a big varispeed fan moving the air. Some designs can even make use of the cabinet airflow in a big case. I will grant you that the ones with insane surface finishes and LEDs are very H-R/R-G.

I wasn't impressed that hardly any of them specify their capabilities in what numerical terms that I would recognise as standard heatsink units. In the end I chose one price performance and it has not disappointed.

Not half as quickly as the stock Intel coolers which seem to clog with dust and gum up their bearings from the moment they are switched on and whine like hell whenever the machine is asked to work hard.

If the heatsink isn't in good thermal contact then all bets are off. I did worry about the large physical size and leverage, but I do not move my boxes around very often. YMMV

Regards, Martin Brown

On a sunny day (Mon, 20 Jul 2009 00:40:40 +0100) it happened Nobody wrote in :

I run a lot of programs at 'nice -xxx' in Linux. Doing that makes sure the CPU is always fully loaded. It has no influence on the normal usage. ftp://panteltje.com/pub/filemanager26.gif

Pretty much anything will be quieter too. I have one of the HSFs in the test mentioned, Noctua NH-C12P. But the installation requires removing the motherboard, so I haven't gotten around to it yet. And to think I'm in this situation because I believed a salesguy, he said the Intel stock cooler isn't noisy!

Anyway, the reason for the other HSF is simply that the stock intel thing is annoying. It's noisy already when the CPU is idle and revs up to very noisy as soon as the CPU is loaded. I have a fairly quiet system otherwise, so the noisy CPU fan really makes itself heard. And anything I can hear is noisy...

I know some super computers have used strange cooling. Isn't there one that has the electronics emerised in coolant, or was that a dream or movie.

I got a Luxman audio power amp with cooling pipes. The only advantage is transistor mounting considerations.

greg