|
Originally I had not really planned a Part II, of this post. However, after reading a couple of press reviews of our LCLC recently, I got inspired to follow up on the heat pipe VS liquid cooling thing again. If you have not read Part I yet, you should do that here
Last week I got a link to an LCLC review at Hothardware.com. I was not even aware they had a unit for review, but I have to say that I was very impressed with the professionalism and soberness of the article when I read it. I feel it is seldom that any site reviews anything this thoroughly nowadays. The funny part is that they basically came to the same conclusion as for example Anandtech, when stating that liquid cooling has exactly same performance as air cooling (or rather the opposite). However, if you start to scratch a bit under the surface, you will find that there is significant difference anyway. For example when looking at the pure CPU temperatures, they are about equal. However as the review unit includes a GPU cooler as well, and as the 8800 GTX probably consumes 60W in idle mode, the LCLC is actually removing 60W more of heat than the air cooler, and still capable of keeping the CPU at the same temperature. These 60W probably increases the CPU temperature with 5'C. Just to do a santity check I looked at this review, in Danish (English conclusion on the last page) they test without a GPU in the loop, and see a 5'C LCLC lead over their air cooler. So my "back of an envelope" estimate is probably pretty accurate. Also the air cooler and LCLC is not tested with the same air flow in the Hothardware review, since the fan on the air cooler is running 2640rpm, and the fan on the LCLC is only running 800rpm. There is a test where the air cooler is also only running at low rpm, and this case the temperature is 5'C higher. So it it very fair to assume that the LCLC would have performed 3-5'C better at the same airflow although I have not studied it in detail. Finally all tests are done in an open bench. As the air cooler would pump all the CPU heat in to the chassis, the CPU temperature would definitely be much higher, whereas the LCLC would transfer the heat out of the chassis. Again this could easily have a 1-3'C gain on the LCLC. All things equal it actually means that the LCLC actually performs significantly better than the air cooler in this test. I have suggested my good friend Dave, to do an update to the article, but no matter what I still like the review a lot. The reason for liking this review a lot, although my above-mentioned points are not explicitly addressed is that Hothardware, although implicitly, addresses them systematically by mentioning the lower noise etc. in the article.
Very good read indeed! ASE
» 5 Comments
1Comment
at Tuesday, 03 June 2008 10:23
Just reasently I thought of "best cooling system" for overclocker/ Followa are the results:
Overclocked computer cannot be cooled be the air cooler - that is known. But what other cooler to choose?
There are three basic types of coolers to select from: water, thermoelectric (Peltier) and two kinds of phase change, namely: thermal pipes and direct evaporation (in principle, they differ only by cooling substance, but in result differ dramatically).
If to compare by the temperature the CPU might be cooled to, they will stand as follow (better first): direct evaporation, thermoelectric, thermal pipes, water.
So, my first thought was to use direct evaporation. But after second thought I understood that it (and Peltier) is good either for a short run, or for the work in the humidity-controlled environment in 24x7 operating mode. Without the humidity control there sooner or later will be water condensation that will ruin the system. If computer is turned on/off (at least for the night), thermal fatigue will do it naughty job.
On the other hand, neither thermal pipes, no water may keep processor cool at high clocks. Thermal pipes are good to take the heat off only is they are substantially chilled themselves - with air cooling the appropriate vents combination must be about half a meter high! Water is a good cooler, but somewhat latent and just cannot be effective on the small chip surface. On the other hand it is not easy cooled. So, there seem to be a dead end.
Bu
2"Where Are We Really At? 1"
at Tuesday, 28 July 2009 19:23
[Request to remove the 1500 char limit]
I wholeheartedly agree with the above comment, We do indeed arrive at dead ends with almost every marketable solution. AseTek seems focused almost exclusively on the OEM perspective and relationship. I don't blame them. We all need our bread and butter. But upon examination of where the cooling alternative market originates from in the first place we will always be at dead ends until chip and MB manufacturers consider cooling as an integral part of their design.
Most of us turn to alternative cooling solutions because we do not believe chip manufactures. They tell us that the theoretical clock limits of their current process is around 3 gigahertz. Balderdash! Most of the people reading this probably already know that just from garage hacking we can achieve twice that. Yet Intel and AMD do not change their process and continue on with PoP (Package on Package) designs which are at best, difficult to cool. There is die level corrugation to consider and WB (Wire Bondning) or FCB (Flip Chip Bonding) on FPC (Flexible Printed Circuit) SMT (Surface-Mount Tech) are all better alternatives (but certainly not the only alternatives available) to PoP mounting, when it comes to the employment of cooling systems. Why AP (Advanced Packaging) designs do not include such innovations as encapsulation, flow chambers, and etc. for the PC market is beyond me. The technology exists and has been in use for years elsewhere!
Is it a conspiracy?
3"Where Are We Really At? 2"
at Tuesday, 28 July 2009 19:56
One can only guess.
But as a result of such neglect users are now forced to result to such drastic measures as total submersion with VARs and OEMs such as Armari (http://www.armari.co.uk/xcp-gallery.asp) and HCC (http://www.hardcorecomputer.com/) following right along. The situation has become desperate simply because chip manufacturers do not innovate in this area at all. Their big contribution has been the HIS. :(
But why should they? No one is asking them to. Companies like AseTek in tandem with MB manufacturers should in coalition, approach Intel and AMD and begin an R&D think-tank of sorts in order to remedy this.
Intel is currently telling developers that they need to bare the brunt of their future product maps by saying "Get ready for application designs that can deal with thousands of cores in future processor models" [paraphrased]. This is all well and good but only 30% ~ 40% of applications can (at all) benefit from multi-threaded architecting.
No, we're right not to believe chip manufacturers when they tell us that they're reached some magic limit, have hit a wall, and that as a result we will be stuck at around 3 gigahertz. We've ourselves, proven this to be untrue beyond any doubt and have shown that it's only a lack of innovation on their part.
So, where are we really at? Or maybe the better question is: Where should we really be?
Should we be following along at the OEM level with different but same solutions or should we b
4"Where Are We Really At? 3"
at Tuesday, 28 July 2009 20:01
[Sorry that last sentence got cut off - the char-counter wasn't working properly]
...
Should we be following along at the OEM level with different but same solutions or should we be at Intel and AMD's doorstep making requests and placing orders?
-END
5Comment
at Tuesday, 17 November 2009 05:40
Ok.........you are preaching to the choir on the concept of liquid cooling/low noise........I'm getting old and it is a race between low noise tolerance and going deaf. :)
I have been an overclocker for 20 years..........I actually had a vaporchill at one point. I am no longer going extreme.......just moderate tweaking and low noise is now the goal.
I built my current system in 2003 and it has run perfectly for all of these years........a twin Opteron. So, I am finally building a new game system and will use your Corsair type LCLC Cpu cooler. What do you all have for twin ATI 5870's?
Thanks man!
John
» Post Comment
Only registered users can write a comment.
Please login or register.
|
Blog 
