megatron-uk wrote on 2024-04-20, 07:28:

An FPGA design may be the way to go here.

Yes I am lately intrigued by the Next186 core which does 90-110 "Landmark Mhz" which would be about a 66Mhz 286.

Oldest...

Surprised no one posted this yet:

Cray-2 with Fluorinert liquid cooling in 1985.

Wasn't the one that looks like the middle of a Tardis also fluid cooled and 5+ years older?

Water cooling any chip before the complete move to "flip chip" technology is kind of pointless, since the limits on the cooling are between the chip die and the package in which it is encapsulated in older chips. Also, it's not until we get to 40-50+ Watts TDP and smaller process nodes that water cooling can become cost-effective.

As for the earliest water cooling I've seen... I think that takes me back to socket 462/A systems. I'm sure someone tried it even long before then... but s462 was when many people really started to do "silly" things with their systems to get it to the highest possible clock. I remember running into random websites that showed off custom builds with completely hand-made water blocks and systems. Now those were the truly fun days IMO. Nowadays, you just buy everything and slap it together, save for hardline water cooling loops, of course - only those are a bit more of a challenge and require some DIY.

Not water cooling but there was some interest in essentially building fridge cases for overclocking - eg. Kryotech's K6-2 system in '98 (https://www.anandtech.com/show/210). Asetek also did some epic Vapochill cases a few years later.

JidaiGeki wrote on 2024-04-20, 16:17:

Not water cooling but there was some interest in essentially building fridge cases for overclocking - eg. Kryotech's K6-2 system in '98 (https://www.anandtech.com/show/210). Asetek also did some epic Vapochill cases a few years later.

i actually almost purchased a vapochill pc running a p3 or athlon a few years ago for $400. the deal fell through though.