Tony Cole's Cray-1

From: William Donzelli <>
Date: Tue Apr 28 12:06:24 1998

> The Cray-1 was ECL-10k fast for it's time but low density and rams for
> that technology were 1 or 4k ECL bipolar.

I do not think the Cray-1s used 10K ECL - I believe they were custom parts
and were faster (10K gates have a delay around 2 nS). Only four types of
chips were used in the whole beast - I think two were OR/NOR gates, one
was a flip-flop, and the other RAM.

The F10470 is indeed a 4K device, with a 15nS access time - quite fast for
its day. 1981 date codes imply that this was from a Cray-1S (-1Ms used MOS
memory, and straight-1s were out of production by then).

> The
> copper plate worked with a cooling system to conduct the heat away as
> that machine was impossible to air cool and remain that small.

Now you know why older Crays tend to be so darn heavy!

The copper plate simply conducted the heat from the board to the tall
aluminum uprights (visible on the -1s, hidden on the X/MP and old Y/MPS).
These uprights carried the cooling fluid.

Although some may be saddened by the fact that Mr. Cole is chopping up a
Cray, keep in mind that the survival rate of the machines is phenominal
- probably 10 to 20 percent of the things tend to get decommisioned
straight to the lobbies of the computing centers, to serve as gate guards.

As a side note, anyone want to buy some 10G logic, for the ultimate in

William Donzelli
Received on Tue Apr 28 1998 - 12:06:24 BST

This archive was generated by hypermail 2.3.0 : Fri Oct 10 2014 - 23:30:42 BST