Author: Peter Fendrich
Date: 12:31:12 08/13/04
Go up one level in this thread
On August 13, 2004 at 15:06:00, Russell Reagan wrote: >On August 13, 2004 at 11:50:13, Anthony Cozzie wrote: > >>There are Fortran -> C compilers. There are C++ -> C compilers. I'm fairly >>sure there are ML -> C compilers. So worst case, you can generate C code that >>is just as fast as ML. > >Not necessarily. Suppose language X makes certain guarantees that C does not. >The language X optimizer may produce faster code. Using an X-to-C compiler may >even produce slower code since there are different guarantees. > >It's the same concept that explains why Fortran is faster than C for some >purposes. Sometimes the C compiler can't guarantee that two pointers don't point >to the same data. If another language makes such a guarantee, it can do further >optimization that the C optimizer might not be able to do. > >My only point is: It's possible. Claiming "always", "impossible", "never", and >so on is making a significant claim, and so far I haven't seen any arguments >that would convince me that nothing can ever be faster than C because C is the >closest to assembly. Maybe it's true 99% of the time, and I wouldn't dispute >that, but that wasn't the initial claim. Just to support that, here is an interesting shoutout between 30 languages. The difference from C to highlevel languages like ocaml, SML, CLisp and Scheme isn't that big. In fact C++ isn't number 2 after C... http://www.bagley.org/~doug/shootout/craps.shtml /Peter
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.