Swiftech Apogee GT vs the D-Tek FuZion CPU blocks @ CES

[Barb]

It is by no means possible that a warped base on a lapped IHS would perform better that a truly flat base on a lapped, truly flat IHS.

I'm going to go ahead a completely agree with you.

I would be willing to bet that using a stiffer mounting plate would result in a better contact pressure distribution and better heat transfer for either gasket.

Personally I'd rather not have a block that is so finicky. I don't want to lap my cpu or have to try out various configurations to achieve the best performance.

[Philly_Boy]

It is by no means possible that a warped base on a lapped IHS would perform better that a truly flat base on a lapped, truly flat IHS. One of two things are happening here......

1). IHS is not lapped truly flat and the warped base fits it nicely.
2). The base is actually not being warped.

How are you measuring the severity of the "warp" factor. You would need some dial indicators and two truly flat surfaces to measure against.

Think about the true rules of physics you are negating here..... less surface contact (bowed based- assuming a flat lapped IHS) allowing more heat transfer, resulting in better temps??? I don't think so.

Although I have a FuZion, I don't who we conclude has the "best" block, but let's not negate some basic physics theory's.

You can't argue with the results. Flat was 8c worse than bowed. The bowed base has also helped the FuZion have better performance as well. It works better with lapped CPU's as long as you're careful to get a good even mount.

I'm going to go ahead a completely agree with you.

I would be willing to bet that using a stiffer mounting plate would result in a better contact pressure distribution and better heat transfer for either gasket.

Personally I'd rather not have a block that is so finicky. I don't want to lap my cpu or have to try out various configurations to achieve the best performance.

The mounting plate will remain true to the point of bowing the mobo if you tighten the mount too much. That's why we were so finicky with the mounting for the testing. Good even contact pressure just short of bowing the mobo worked best for us. YMMV. The Apogee comes congigured witha flat base. If you don't want to experiment with getting better performence, then you don't have to use the gasket that bows the base. The choice is yours. Either way both blocks are very good performers.

[Pete]

Umm duno what to belive now. Dual core is more comen, better to of tested with quad why bother, lapping not all do it. I'e had 5 doa chips new now so i wont be lapping at all

[Sparky]

I guess what I am more interested in is which one performs better on a single die CPU, with and without an IHS? I don't care much about the dual die CPUs since I'm not going to have one anyway - I want single die CPU test results!

[johnanderson]

I guess what I am more interested in is which one performs better on a single die CPU, with and without an IHS? I don't care much about the dual die CPUs since I'm not going to have one anyway - I want single die CPU test results!

I agree. Think Ill skip this year and wait for integrated one die quad.

[Turtle 1]

Very good read nice work guys. I am a little surprised that you guys are finding clamping loads making a differance in results. As torque specs. are usually very important .

I keep thinking of the stock heatsinks on the northies when you clamped these babies down I thought they would break the M/B or the cpu. I mean the force I had to apply to clamp these things scared me. But I never Had a problem and you talk about bowing the M/B it was massive.I suspect this pressure forces better contact between the Core and the IHS. Which would promote better heat transfer. Very nice work guys.

[migueld]

This bowed thing seems fishy, either that or there's some great unknown science about it.

[Sparky]

I agree. Think Ill skip this year and wait for integrated one die quad.

AMD's K8L - or KL8 or whatever it is called - is gonna be single die, quad core, released sometime in the second half of this year IIRC. So you might not have to wait another year unless you want an intel chip instead.

[Barb]

You can't argue with the results. Flat was 8c worse than bowed. The bowed base has also helped the FuZion have better performance as well. It works better with lapped CPU's as long as you're careful to get a good even mount.

If it works then it works. I am very interested in why it works. These results should seriously shift the focus of good water block design. If a simple gasket can yield a 8°C temperature drop in the apogee gt, then maybe all blocks would benefit from bigger gaskets.
What if you use an even bigger gasket?
Are the results going to be consistent for various cpu's and mounting brackets?
What if you make a water block entirely out of gasket material?

Further testing needs to be done. Testing that should have already been done by the manufacture.
Waiting till a few days before to official release of the product is a bad time to be doing this kind of testing. If the thicker gasket works better, then it should probably only ship with a thick gasket, or perhaps the thiner gasket will work better on a bare core.
Either way R&D should not be the job of the consumer.

[Turtle 1]

This bowed thing seems fishy, either that or there's some great unknown science about it.

No real mystery here clamping force between objects has been used for years .exspecially were heat transferr is desired. It insures even contact between the heatsource and the trasferr medium. I suspect that higher forces are required to get better contact between the ihs and the core that would explain the better results when bowing.

[TheJollyFellow]

I guess what I am more interested in is which one performs better on a single die CPU, with and without an IHS? I don't care much about the dual die CPUs since I'm not going to have one anyway - I want single die CPU test results!

me too

[nikhsub1]

Hmm, my Apogee GT did not come with an additionally o-ring. Neither has anyone else's that has already received the retail packaging. This makes me believe that this bowed base was an afterthought... If Swiftech knew the bowed base was so good, the block would have shown up for the challenge (the first challenge) with a bowed base. Mention was made that the Fuzion has a convex base - how coincidental! What this boils down to is just changing the mount geometry, nothing more. ANY block with a bowed base will perform better on a current gen quad core. Give ANY block a thicker o-ring, hint hint, and performance should go up. Of course, all bets are off on this technique with a single piece of silicone in the center of the cpu.

[WeStSiDePLaYa]

my theory as to why the bowed base gives better numbers,

when you have high pressure on a flat metal base, the copper will bend slightly, and lose some contact pressure around the center of the block.

and a bowed block compensates for that.

as well as my included diagram, there is an easy way to show what i mean.

take a CD, place it on the bottom of a glass, you will see that it sits flat. but then apply pressure to the outer parts of the cd, like how blocks are mounted, and you will see the middle of the cd raise away from the glass.

i think the same thing is happening with the bases of blocks, except on a less obvious because its happening with metal.

EDIT: this means that ANY cpu, no matter what core configuration, will benefit from a bowed base. Also, no point of testing on bare cores anymore since ALL IHS are soldered on now, for amd and intel.

[RyderOCZ]

Totally agree with WSP....my thoughts exactly as to why bowed showed lower temps.

[Praxis1452]

Totally agree with WSP....my thoughts exactly as to why bowed showed lower temps.

yup

[WeStSiDePLaYa]

Totally agree with WSP....my thoughts exactly as to why bowed showed lower temps.

so would i be correct in my thinking that the ocz block will have a bowed base

i dont think this was any secret to ocz.

[darcon]

But does a solid piece of copper really bend that much to create such a dramatic difference in temperatures?

[nikhsub1]

But does a solid piece of copper really bend that much to create such a dramatic difference in temperatures?

Absolutely. You are concentrating the mounting force to the CENTER of the IHS rather than having it deflected to the edges, which in theory could pull the the center of the base AWAY from the IHS as in WSP's illustration.

[darcon]

I just thought of something else...what about the thermal paste factor?

[serialk11r]

Not too much thermal paste but just enough so that there is very good contact.

[IanY]

I find this whole convex water block base thing very questionable. It would be helpful if there were flat IHSes in the first place. What if the next batch of cpus indeed have perfectly flat IHSes. Then what? FuZion and Apogee GT go into the trash ? Anyway, I'm tired of all this bickering over tiny differences in blocks.

[cdelong]

You can't argue with the results. Flat was 8c worse than bowed. The bowed base has also helped the FuZion have better performance as well. It works better with lapped CPU's as long as you're careful to get a good even mount.

The mounting plate will remain true to the point of bowing the mobo if you tighten the mount too much. That's why we were so finicky with the mounting for the testing. Good even contact pressure just short of bowing the mobo worked best for us. YMMV. The Apogee comes congigured witha flat base. If you don't want to experiment with getting better performence, then you don't have to use the gasket that bows the base. The choice is yours. Either way both blocks are very good performers.

first off.... no inflammatory comments intended here

I'm not disagreeing with your results at all.... numbers are numbers I guess all I'm trying to make clear is that more of a contact patch vs. less of a contact patch between the WB and the IHS, will yield better results.

The only way a warped base performs better (assuming every other varaible is constant) is if it yields more of a contact patch... period!

I have done a many a "hot rock" problems in heat transfer. If I need to explain the hot rock theory, I guess I'm speaking to the wrong audience here. You can't tell me that dipping a small portion of a hot rock (IHS) into a pool of liquid (WB), would cool the hot rock better than dipping the entire hot rock (IHS) in the pool of liquid (WB). This is all asssuming the specific heat capacity of the liquid is constant- I'm guessing you used the same mixture of coolant after refilling for each test.

I think we all the see the results of the test's... let's just be sure WHY we are seeing them Less of a contact patch is not the reason for sure

[Cathar]

Bowed bases? Convex bases? Warped IHS's?

What is this? Send in a hunchback to do a hunchback's job?

What happens when the next batch of CPU's come out with a revised IHS design that is flatter?

This is exactly why I always preferred bare-die and/or die-sim testing. No such dependencies upon manufacturing variances to obtain solid repeatable results. Building blocks is already "horses for courses" enough, without then tweaking to perform the best on an singular individual CPU (or batch of CPU's) with their own inidividual quirks.

What is greater truth here? Shall we start making individual CPU specific water-blocks? Buy a CPU and get the matching w/b?

[ziddey]

IanY, the theory (as WSP put out) is that with a flat base of the block, when you bolt it down, the force on the perimeter causes the base to become concave. It's when you start the base off concave that after you bolt it down, it pulls it such that it ends up roughly flat. So to have a flat starting base of block, IHS would have to be convex in the same fashion the block curves.

[SiGfever]

Bowed bases? Convex bases? Warped IHS's?

What is this? Send in a hunchback to do a hunchback's job?

What happens when the next batch of CPU's come out with a revised IHS design that is flatter?

This is exactly why I always preferred bare-die and/or die-sim testing. No such dependencies upon manufacturing variances to obtain solid repeatable results. Building blocks is already "horses for courses" enough, without then tweaking to perform the best on an singular individual CPU (or batch of CPU's) with their own inidividual quirks.

What is greater truth here? Shall we start making individual CPU specific water-blocks? Buy a CPU and get the matching w/b?

Now wouldn't that make the block retailers happy? We would be changing blocks like we change socks.