Re: Parallelizing (WAS: Death by pointers.)

pardo@cs.washington.edu (David Keppel) writes:


>Amdahl makes the point that using two processors more than doubles the
>processor cost -- because you also have to connect the processors --
>and less than doubles the performance -- because shared resources
>become bottlenecks. Moral (not quite, but I'll fudge it): If you *are*
>going to build a parallel machine, make sure that you minimize the raw
>cost/performance, even at the expense of making it harder to program.


I'm not sure how you got from Amdahl's point to the moral, but surely
we've already seen plenty of relatively cheap but
much-too-hard-to-program parallel machines. For that matter, I think
the uniprocessor machines are becoming significantly more difficult
to program -- very hard to achieve good performance in the presence
of a deep memory hierarchy. Parallelism offers a way around that
particular bottleneck.


Preston Briggs
--

Return to the comp.compilers page.
Search the comp.compilers archives again.