Subject: Re: Floating Point speed in Common Lisp
From: Erik Naggum <clerik@naggum.no>
Date: 1998/03/16
Newsgroups: comp.lang.lisp
Message-ID: <3099027379447143@naggum.no>


* Matthew McDonald
| Do you have any examples to substantiate this claim?

  not any good set of particular examples, but 15 years of experience with
  C on lots of platforms and lots of compilers has alerted me to the fact
  that C compilers and environments differ _very_ much in quality, more
  than my experience with (native-compiled) Common Lisp implementations and
  environments do.  Common Lisp implementations are generally quite good,
  but the same is not necessarily true for C compilers -- naive compilation
  of C is often a viable option for small systems.

  if you need a specific example, consider the bundled C compiler with
  SunOS 4.X with the commercial compiler for Solaris 2.X.  GCC 2.8.1 beats
  the former by a factor of 1.7, but the Solaris 2.X commercial C compiler
  appears to be slightly better than GCC.

  I find it interesting that you jump up to require substantiation of the
  claim that C implementations differ, but do not challenge that Common
  Lisp implementations do.  this is the general sentiment out there, as
  sort of a cultural requirement on programmers.

  C is not fast.  C is primitive, which means it has a potential of being
  fast if the programmer is a genius and the compiler is on his side.  if
  the programmer is an idiot, and the compiler expects smarter programmers,
  you will get abysmal performance.  e.g., a system currently written in C
  but which I am replacing with one rewritten in Common Lisp shows a
  promise of reducing system requirements by a factor of between 5 and 6,
  after months of work to decipher the incredibly bad C code to figure out
  what it actually does.

#:Erik
-- 
  religious cult update in light of new scientific discoveries:
  "when we cannot go to the comet, the comet must come to us."