The Power of the NORC 3754

That *is* a valid point. On a smaller scale, one could imagine, in the old days, people running a cross-compiler for the PDP-10 under OS-8 on a PDP-8 system (using a matching floating-point format, of course) which would then submit batch jobs to the PDP-10. Of course, that means you have to remember that source files go on the PDP-8's disks, and data files go on the PDP-10's disks.

Since both supercomputers and desktop PCs now use the IEEE-754 floating-point format, cross-compiling these days should be a breeze; you do that in your "desktop terminal", not in a "front-end processor", both terms being somewhat misnomers in such a case.

Obviously, machines should be designed to efficiently serve their intended application. The poster to which I was replying, somewhere up the thread, raised the point that I was hasty in dismissing the AN-FSQ-7 as irrelevant in declining to compare its performance to that of the NORC: he was raising the point that a really powerful computer designed and used for *some other purpose than numerical analysis* has as much right to be called a supercomputer as a supercomputer addressing that particular problem domain.

I was admitting his basic point was correct, although I didn't know of too many cases to which it applied.

It may indeed be a waste to *directly* work on this now, at the current limited state of our technology. But I do consider this sort of thing a desirable goal, when its time comes.

John Savard Usenet Zone Free Binaries Usenet Server More than 140,000 groups Unlimited download