Newsgroups: comp.lang.lisp
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!news2.near.net!news.mathworks.com!news.kei.com!sol.ctr.columbia.edu!howland.reston.ans.net!EU.net!uknet!festival!edcogsci!jeff
From: jeff@aiai.ed.ac.uk (Jeff Dalton)
Subject: Re: Squeezing more speed out of LISP.
Message-ID: <Cy1FF0.4tz@cogsci.ed.ac.uk>
Sender: usenet@cogsci.ed.ac.uk (C News Software)
Nntp-Posting-Host: bute-alter.aiai.ed.ac.uk
Organization: AIAI, University of Edinburgh, Scotland
References: <37s318$pr0@cantaloupe.srv.cs.cmu.edu> <HJSTEIN.94Oct18135742@sunset.huji.ac.il> <380msr$s53@cantaloupe.srv.cs.cmu.edu>
Date: Fri, 21 Oct 1994 19:28:11 GMT
Lines: 23

In article <380msr$s53@cantaloupe.srv.cs.cmu.edu> sef@CS.CMU.EDU (Scott Fahlman) writes:
>
>In article <HJSTEIN.94Oct18135742@sunset.huji.ac.il> hjstein@sunset.huji.ac.il (Harvey J. Stein) writes:
>
>   ...  But, in that
>   CMU CL is considered to be so hot, I'd like to know what sort of
>   penalty is involved in using another lisp instead.  For example, as a
>   general, rough rule of thumb, would GCL code take 1.5 times as long to
>   run, or 3 times as long, or what.
>
>I think it's probably more than that for well-tuned single-float
>code.  When our Python compiler first came up on the MIPS-based
>Descstations, I tested a float-intensive neural-net simulator on it.
>This was well-tuned code, though Ken Anderson has recently been able
>to tweak it even further.  Anyway, the speed in CMU CL was very close
>to the performance of the same program translated into C by one of our
>grad students.  It was 5-6 times faster than the Allegro of the day,
>which I think was considerably faster (maybe 2x) than AKCL (now GCL).

Ah, but faster at what?  I don't know about floating point, but
for some things AKCL is not particularly slow.

-- jeff
