from memory the gemini has more channels, but of course each channel is a bit smaller.

Peter at Gemini is working on having a P4 block soon. I talked to him yesterday and he brought it to my attention.

Do you know what the ID of the tubing and fittings in Joe's test system are? (maybe get Joe to do a quick and dirty schematic of his set up in this regard, including all fittings, etc.) I didn't see the info anywhere in his wb roundup and hate to guess from pictures ;)

I am pretty sure that Joe uses 3/8" tubing and fittings for his testing.

I highly doubt that the gemini blocks need something like 5 GPM for optimum performance. I also am skeptical that any watercooling setup that is in common use will generate 5 GPM. My system (a mix of 5/8" and 1/2" fittings and a heatercore and an Eheim 1250) generates around 1.2-1.4 GPM depending on the block chosen. That is always past the sweet spot of the waterblocks I have tested, so higher flows than that will only lower the efficiency of the radiator. The gemini looks like another spiral channel block to me, so it should perform similarly to the Maze2 or the spiral with respect to flow I would think. Super high flow rates might be useful if you have a block like the swiftech and are trying to move to the turbulent regime, but for a block like the gemini truly turbulent flow isn't possible anyway.

You should always pay close attention to the previous work of someone before deciding whether to take their results for water cooling tests as gospel. Go look at some of Hoot's recent heatsink reviews at overclockers.

I am very impressed. I called Peter on Friday at Gemini Cooling and received the block today. It looks totally assume. Can not wait to get this system up and running. Came with a really nice set of nylon bolts, nuts, standoffs, and washers. You can adapted this block to any AMD motherboard. It's small....

I will be using a Danner 500GPH MAG 5 pump. So this will answer a few of your questions. I will also change out the 3/8" to 1/2" fittings.

I have no idea whether the Gemini block out-performs the MAZE2, but I did notice one design difference. From the CAD drawings on the site, the area that sits on top of the core is the comma-shaped beginning of the channel wall. The MAZE2 has the beginning of the channel over the core while the Gemini has 12mm thick copper over the core. Is this correct?

--Rob

BTW: Wasn't this a thread about Swiftech? :D

Phaestus - we get back to the "depends on the system" issue ;) For example in Joe's case, he's over 2.5gpm based on the test figures he's posted, and if you're correct about his setup, that was with 3/8" tubing (and 1/2" tubing is a larger change than most think in regards to back pressure, it just isn't noticed when using a restrictive block. 1/2" tubing has a 77% larger cross section than 3/8").

Regarding rad issues, we get back to systems again - size/type of rad and single or dual rads.

As far as blocks, I will agree that small channel single path blocks won't benefit much past the 1 to 1.5gpm range (the improvements become fractions of a degree). But, for larger apparent cross section blocks (whether that is accomplish with mutliple parallel channels or larger channels), it can and does make a difference.

And who said anything about "common use"? (and define the scope of "common" - all users or all high end users or...) :p

BTW - nice review on the CrystalFontz :D (and how much would software be worth to you?)