ok this just kick's ass (was RE: [rescue] The onyx godness
Francisco Javier Mesa-Martinez
lefa at ucsc.edu
Thu Jan 16 00:32:34 CST 2003
On Thu, 16 Jan 2003, dave wrote:
> me at dansikorski.com writes:
> >On Thu, 2003-01-16 at 00:00, dave wrote:
> >> me at dansikorski.com writes:
> >> >Control Data Cyber205 S/N 515. Let no more suffer the same fate.
> >> S/N *. As far as I know, no 205 survives.
> >Wow, i just realized that the machine was installed 20 years ago last
> >week. (January 4-9, 1983). I was three years old.
> Any details?
> All I know about the CDC-205 was that it was huge (largest non-tube
> CPU ever?) and that it did memory-to-memory vector ops on really
> long vectors (64K?).
> Its memory bandwidth must have been terrifying.
I think that was what killed the machine, the bandwith was actually quite
poor. And since it was based on the star architecture, you could have
instruction that literally took almost a second (if not over) to complete!
A single instruction... Plus scalar performance was also sucky. CDC never
recovered after that series of machines.
More information about the rescue