space vs. speed optimization

There was a paper that suggested dynamically adjusting space vs. time
tradeoffs by assuming a model:


cost = A * space + B * time + C * space * time


then, based on the measured execution frequencies of various parts of
the program, the compiler could choose the least-cost code. This is a
real win relative to the usual strategy, because frequently-called
code was optimized for speed, and infrequently-called code was
optimized for size. And, as the author pointed out, programmers don't
generally know which is the better tradeoff in every case.


Dale
[From Dale Worley <think!compass!worley@eddie.mit.edu.uucp>]
--

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