Re: Code quality

drw@zermelo.mit.edu (Dale R. Worley) writes:
>Is there much of a market for another 10% in speed of generated code?


henry@zoo.toronto.edu (Henry Spencer) writes:
> There is no shortage of applications which are hard up against processor
> speed limits, where a modest improvement in code quality can save a lot
> of people a lot of pain trying to squeeze out a bit more performance.
> "Just switch to a faster processor" doesn't work when you've got a large
> installed base to worry about, or you're already using the fastest
> available, or the CPU box is orbiting Jupiter...


You don't have to get nearly so exotic as this to find a market for an
extra 10% in code performance. To the user, it is still much less
expensive to upgrade compilers than it is to upgrade a CPU. So if a user
can upgrade his compiler to get an extra few % in performance, then he can
add extra functionality to his application without hurting its
performance. This is especially important in real-time applications,
which often have to do a given amount of work in a specified amount of
time. As we all know, however, the "given amount of work" often increases
over time.


Most users are not going to upgrade their CPUs every year, nor even every
2 years, but they are highly likely to upgrade their compilers every year
(if nothing else, for bug fixes).
--
Bill Leonard
Harris Computer Systems Division
2101 W. Cypress Creek Road
Fort Lauderdale, FL 33309
bill@ssd.csd.harris.com
[Funny, I can upgrade the CPU on my system to a double-speed version for
about $400, but a super-duper optimizing compiler costs more like $700.
-John]
--

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