Swiftech Apogee Discussion ( The Cat is Out of the Bag )

[nikhsub1]

http://www.swiftnets.com/products/Apogee.asp

[moonlightcheese]

better than the storm!?

ZOMGWOW. let's see some XS testing ^_^

[nikhsub1]

better than the storm!?

ZOMGWOW. let's see some XS testing ^_^

Don't be so sure...

[Nettwerk]

what was it? links dead

[MRBIGSHOT]

page cannot be displayed

[moonlightcheese]

Don't be so sure...

that's why i requested the testing... it doesn't seem like it would outperform the storm at all. that "diamond grid" seems like so many other channel style blocks. it seems unlikely that it would outperform jet impingement. but hey, who knows.

[nikhsub1]

Haha, they pulled the page, good thing i got some pics...

[moonlightcheese]

pics of graphs

[moonlightcheese]

pics of block

[Bloody_Sorcerer]

actually, i recall seeing a graph a while ago with the STORM versus the MCW55 on the same die in a C/W versus hydraulic power comparison and it had the diamond pin of the MCW55 beating the storm as it was. I also remember seeing a picture a few years ago of Jason AKA MickeyMouse running an MCW50 on a CPU. There's definitely a lot more to the diamond pin grid than most people give credit. *pokes cathar* wanna shed some light?

[moonlightcheese]

BasePlate

The CNC machined C110 copper base plate is at the heart of Apogee's cooling efficiency. The Patent Pending Diamond Pin Matrix was optimized using Computational Fluid Dynamics analysis to yield the remarkable results recorded by the water-block. In the process, thickness of the base was reduced to 3mm. This resulted in a higher compliance of the base with its mating surface (i.e. the CPU heat spreader) thanks to the base plate added flexing ability.

PD vs FR

It can be seen that the pressure drop of the Apogee is lower than that of the MCW6002 at equal flow rates, and substantially lower than that of the Storm. In the real world where the range of flow rates is limited by the power of the pump, this will also translate into noticeable gains when multiple water-blocks are connected in series. A simple example illustrates this point quite clearly: using the same MCP350 pump, it will take three Apogee water-blocks in series to drop the flow rate to that of single Storm.

TR vs FR

Here it is apparent that the Apogee has the lowest thermal resistance at all flow rates and more importantly features a wider performance advantage at "real life" flow rates between .3 and 1.5 GPM.

The above two graphs can then be combined below to illustrate the pressure drop / thermal resistance relationship

TR vs PD

From the above curve, it can be seen that the Apogee water-block does not necessitate a high pressure pump to be extremely efficient. It yields in fact lower (better) thermal resistance values than the MCW6000 and Storm water-blocks at all pressure drops.

Another way to consider the relationship between the water-block's thermal resistance and the pump capability is to plot the hydraulic resistance, which is literally the work that the pump must do.

TR vs HP

As in the preceding graph, the Apogee outperforms all previous solutions whether a small aquarium pump or a high pressure industrial pump are used.

It should be added that under certain circumstances, the Storm water-block may perform better than the Apogee water-block. We cite for example earlier generations AMD Athlon XP, MP and Duron processors where the die size is smaller in surface area (100 to 140mm2) than current microprocessors. In such instances, the difference in temperature was found to be 1.4°C at 100 Watts, and at the maximum flow rate allowed by our test equipment (about 3.3 GPM for Apogee, and 2 GPM for Storm). Such test were conducted using the alternate testing procedure described in the link below.

[recons]

Looks interesting. I shall await a procooling.com review and the review on OC.com

Did BillA do the testing on this before he left Swiftech?

[D4nnyOcean]

Impressive but I hope I don't need a firehose type pump to get slightly better cooling.

Any word on the cost? Maybe why the Storm block has been seen recently at a reduced price? Although I did nab one for $60 while the gettin was good

[Cathar]

There's definitely a lot more to the diamond pin grid than most people give credit. *pokes cathar* wanna shed some light?

No, there is no more to diamond pin. The issue is with the testbed.

Independent tests will reveal the truth of the matter.

[situman]

Wow block releases within months of each other. I bet they jsut don't wanna give their profits to Cathar anymore. THe less they sell of the storm, the less they would have to give up.

[Cathar]

Specifically this bit:

It should be added that under certain circumstances, the Storm water-block may perform better than the Apogee water-block. We cite for example earlier generations AMD Athlon XP, MP and Duron processors where the die size is smaller in surface area (100 to 140mm2) than current microprocessors. In such instances, the difference in temperature was found to be 1.4°C at 100 Watts, and at the maximum flow rate allowed by our test equipment

I'm guessing that the results described in the above paragraph would've been done on BillA's old small-die testbed (10x10mm). The Storm was better by 1.4°C.

Now the results in the graphs were all done on a testbed that does not measure the actual die temperature, but rather the IHS surface temperature. Irregularities in how the waterblock is applying contact between the heat die and the IHS can vastly alter results (by +/- 4C) when measuring at the die instead.

To date, when measuring on-die temps, as opposed to IHS surface temps, I have not been able to reproduce anything that could describe the results presented in the graphs.

Once again, independent testing will reveal all.

[Edit: I am not saying that Swiftech's results are incorrect. I am saying that I have legitimate concerns over their testbed's ability to predict what the actual CPU die temperature will be, and I await independent testing to verify the results shown.]

[Gimmpy224]

jet imp. on a diamond pin

[Overconfidence]

Harrr.

I'm sad because I was going to do the same design with a little saw and some thick plastic.

=/

[WeStSiDePLaYa]

Cathar, you must admit, this WILL be the block to have simply because of the affect on flow versus the storm. anyone with a multi block loop will want this block over the storm.

offtopic a bit:
Any wonder how dangerden is going to go with this? they really need to get a new CPU block out.

[Cathar]

Cathar, you must admit, this WILL be the block to have simply because of the affect on flow versus the storm. anyone with a multi block loop will want this block over the storm.

No, I don't accept that at all.

If you have a multi-block loop, the flow rate differences will be MUCH smaller because with the added blocks, the correspondent effect of the CPU block on the overall system resistance is much lower.

Then again, for those who think that more flow rate is more important than greater performance, then their minds will already be made up.

[CrimeDog]

Cathar, you must admit, this WILL be the block to have simply because of the affect on flow versus the storm. anyone with a multi block loop will want this block over the storm.

offtopic a bit:
Any wonder how dangerden is going to go with this? they really need to get a new CPU block out.

then people would be using the maze4 instead of a g4/storm or even a tdx... just looking at that design you should know the storm would outperform it.

[WeStSiDePLaYa]

No, I don't accept that at all.

If you have a multi-block loop, the flow rate differences will be MUCH smaller because with the added blocks, the correspondent effect of the CPU block on the overall system resistance is much lower.

Then again, for those who think that more flow rate is more important than greater performance, then their minds will already be made up.

But with performance being so close, but yet this new block being less restrictive. and the most common secondary block is most often a maze4 gpu which is low restriction, this would be the more ideal block.

Single block/no ihs: Storm>apogee
single block/with ihs: storm<apogee
mutliple block w/with out IHS: storm<apogee

so yeah storm is still better idealy, but with common multi block set ups, i think the slighlty less effeciency is much offset by lower restriction, that would be much needed when running gpu or sli blocks.

[Cathar]

i think the slighlty less effeciency is much offset by lower restriction, that would be much needed when running gpu or sli blocks.

I prefer good data to speculation myself.

[Overconfidence]

If this new design was that much better I think we'd have heard of it before .

[situman]

I say we support Cathar all the way. A lot of the design of the new block was inspired by the storm and Cathar. Of course this is just all speculation. I do feel that Swiftech is trying to make a statement that "we can make a block that's better or matches what Cathar can make...and mass produce it at an affordable price."