How the Pentium Fell Short of a 360195 4036

With Nick, I am not either.

I would say "Not enough to satisify" the general programming populice.

APL is the common typical example given in past decades. I've only seen a little of it. I've heard of its influence from the CDC STAR-100 and the Convex line of machines to various IBMish incarnations. But I have also heard excuses from slow interpretation to left-right evaluation and obscure notation (sure I've seen the old APL keyboards).

Now, we mostly have message pbutting (lots of copying).

The problem comes with the semantics, not the syntax, of parallel languages. Operating system programmers are familiar with some of the problem, but less so with most application programmers.

Ease of use is a another related but somewhat separate issue. I've yet to see an easy to use sequential language (programming or natural). This is all decades old. Where Dijkstra's GoTo Harmful letter is cited lots, lesser known is Habermann's short article on global variables.

depends on version: cft, cft77, cft90...

Beware automatic claims.

Loop unrolling Strength reduction If-branch processing, etc. Compiler directives (versions of C$DIR) which for vectorization despite shady dependence analysis. ...

There is a presumption (not valid) that all these techniques get tested. Loop syntax is deceptively simple looking. It's not general purpose. The right groups exist to discuss this.

--