On 16 February 2010 14:50, Ken Porowski <[email protected]> wrote: > IIRC this is not cooling of the CPU/chip itself but an add on cooling > 'door' on the rack to help with the heat dissipation. This has been out > for a couple of years at least.
That's about what the first IBM water cooled mainframes did. The 360/85 and similar 370/165 and /168 had water circulating to all the CPU frames from a central power and coolant distributing unit. But the only chips directly water cooled were in small distributed power supplies, where the power transistors' heat sinks had water channels through the block. The actual logic chips were mounted on conventional circuit boards, and each internal rack/stack of such vertically mounted boards had a finned heat exchanger like a small car radiator (should be called a "conductor", but it's too late to change car terminology) with fans blowing air through it and past the boards. With the 3081, IBM mounted multiple chips in a TCM (Thermal Conduction Module), and each chip had a helium surrounded piston (aluminum, iirc) pressing on it, and conducting heat into the surrounding aluminum block. Then each block had water channels hooked up to the central PCDU. It makes some sense to extract heat from the CPU frame(s) with water, rather then sending it into the room air only to be extracted less efficiently into circulating water or other fluid. Whether it's chip->metal->water or chip->air->metal->water or even chip->metal->air->metal->water probably doesn't make that much difference. Tony H. ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
