Historical caution ...
•Jack
Schwartz wrote a paper called Ultracomputer in which he
observed
•The
ideal parallel computer would be a PRAM (he called it a
paracomputer), but it can’t work
•The realistic alterative would be an “ultracomputer”
which was
a specific (CTA-type architecture) using a shuffle-ex network, vN
processors, etc.
•Allen
Gottlieb started a project to build an ultracomputer, but after getting
started, i.e. after the computer was already named, decided they’d build a
paracomputer
Although the original ultracomputer
doesn’t have a
shared memory the machine
that was built did … confusing