Re: Death by pointers.

graham.matthews@pell.anu.edu.au
Tue, 12 Sep 1995 02:37:36 GMT

          From comp.compilers

Related articles
Order of argument evaluation in C++, etc. hbaker@netcom.com (1995-07-08)
Death by pointers. (Was: order of argument evaluation in C++, etc.) johnston@imec.be (1995-08-30)
Re: Death by pointers. jhallen@world.std.com (1995-09-05)
Re: Death by pointers. whitten@netcom.com (1995-09-05)
Re: Death by pointers. chase@centerline.com (1995-09-06)
Re: Death by pointers. hbaker@netcom.com (1995-09-11)
Re: Death by pointers. graham.matthews@pell.anu.edu.au (1995-09-12)
Re: Death by pointers. chase@centerline.com (1995-09-12)
Re: Death by pointers. preston@tera.com (1995-09-13)
Re: Death by pointers. ECE@dwaf-hri.pwv.gov.za (John Carter) (1995-09-23)
Re: Death by pointers. stefan.monnier@epfl.ch (Stefan Monnier) (1995-09-25)
| List of all articles for this month |

Newsgroups: comp.compilers
From: graham.matthews@pell.anu.edu.au
Keywords: C, functional, performance
Organization: Australian National University
References: 95-07-068 95-09-030 95-09-074
Date: Tue, 12 Sep 1995 02:37:36 GMT

chase@centerline.com (David Chase) writes:
>I think you're also overestimating the efficiency of languages that
>completely lack side-effects -- a naive translation of in-place pivoting
>matrix factorization into a side-effect-free-language might run O(N^2)
>times slower than the original, if your compiler fails to figure out that
>the storage can be reused (I added the pivoting because that's what
>people do, and because it will make access patterns harder to analyze).


I think you need to look at Sisal. It's totally side effect free, and its
performance on a large range of tasks (eg. the Lawrence Livermore Loops)
is as good (and often better) than Fortran. By being side effect free
(strictly single assignment), Sisal swolves the aliasing problem, but
introduces the copying problem (lots of programs appear to require lots
of copying to ensure single assignment semantics). The latter problem
appears to be easier to solve than the former problem however -- the Sisal
compiler does a great job copy avoidance.


>> The most pressing question, of course, is what the new, side-effect
>> reduced, language should be called.
>I think the answer remains "Fortran".


How about Sisal.


graham
--


Post a followup to this message

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