home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!paladin.american.edu!howland.reston.ans.net!zaphod.mps.ohio-state.edu!cs.utexas.edu!uwm.edu!ogicse!das-news.harvard.edu!spdcc!iecc!compilers-sender
- From: preston@dawn.cs.rice.edu (Preston Briggs)
- Newsgroups: comp.compilers
- Subject: Re: Code quality
- Keywords: optimize, architecture
- Message-ID: <93-01-019@comp.compilers>
- Date: 6 Jan 93 19:25:39 GMT
- Article-I.D.: comp.93-01-019
- References: <93-01-017@comp.compilers>
- Sender: compilers-sender@iecc.cambridge.ma.us
- Reply-To: preston@dawn.cs.rice.edu (Preston Briggs)
- Organization: Rice University, Houston
- Lines: 26
- Approved: compilers@iecc.cambridge.ma.us
-
- drw@zermelo.mit.edu (Dale R. Worley) writes:
-
- >How important is generated code quality these days? There are a lot of
- >good optimization techniques that seem to be adequate for ordinary
- >programming. But they still are at least 10% or 20% worse than the ideal.
- >Is there much of a market for another 10% in speed of generated code?
-
- I think high-quality code generation is still important. However, people
- aren't worried so much about the final 10%; rather, they are concerned
- about the factors of 5 and 10 and 20 that current compilers lose to
- (knowledgable and hardworking) humans when faced with an exotic
- architecture. Typically, compilers still lose big on various kinds of
- parallel machines and on machines with interesting memory hierachies. Of
- course, most high-performance machines have many levels of parallelism and
- a multi-level memory hierarchy, so... Lots of problems to work on!
-
- Note that I'm not advocating hand coding for these machines. I believe
- that even simple compilers are easily good enough to economically justify
- their use in almost every circumstance. Instead, I'm saying that the
- demands (and money) of high-performance users justify quite a lot more
- research effort.
-
- Preston Briggs
- --
- Send compilers articles to compilers@iecc.cambridge.ma.us or
- {ima | spdcc | world}!iecc!compilers. Meta-mail to compilers-request.
-