Cathar su Xterme systems oggi:
Hmmm, need some clarifications here obviously.
G7 is as far as the design can go physically with CNC machines and cutters that I have access to. Could perhaps do a G8 given aerospace quality mills but to do so would also require custom building most of the tools and cutters to make the block because we're now dealing with incredibly small dimensions.
The other problem with going to higher than a G7 is that the flow resistance starts to climb quite aggressively due to the nature of the design. Also, some of the principles in the block begin to break down and all I'd be left with was some sort of modified micro-channel type block.
To put it into perspective for you guys for a G7 level design's intricacy, consider a US dime (10c piece) size area, the G7 would pack well over 60 (sixty) jets/cups pairs into that area. Go grab a dime and picture it.
So G7 is the limit, unless I find a spare US$5K and get a G8 made up, which is just truly silly.
Cost for me to get say 10 of them made? This is not known yet but if I had to estimate it, ~$300US each. Yes. It's that kind of silliness.
Giving serious thought to stopping making CPU blocks completely after this. I WILL do another G5 batch, which will include technology explored in the G7, and run them until I run out of the silver base-plates I have here in reserve (about 30), and after that I'll stop. I still have two more ideas that I want to explore for waterblock design but my confidence level is rather neutral at present for purposes of imagining that they will improve upon the Storm design. Have begun prototyping of one of the ideas already.
Seriously though guys, hitting very, very hard physical limits here now. There are no BIG gains left. There's only so much you can do with 0.5-3W of hydraulic pumping power, PC-case limited radiator sizes, and so on. Even with much stronger pumping powers the gains are small.
I wrote some years back that I didn't think that more than 3.5°C better than a White Water on a 100W 100mm² CPU would be realistically achievable, and that 4.5°C better is about an absolute theoretical limit (and anything between 3.5-4.5°C is not realistically achievable or practical). I believe that the G7 will pretty much take me right up to around the 3°C better mark. Now by 3°C, I'm talking about relative to BillA's old CPU test die, not Procooling's which compresses temperature differences.
Oh, and the G7 is also designed to really kick-butt at low-flow too, not that any of my designs suffered comparitively at low flow (<0.5gpm) but I'm aiming to set a new benchmark at down to 0.3gpm here too.