Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP
Path: utzoo!utgpu!water!watnot!watmath!clyde!rutgers!im4u!milano!mcc-pp!patrick
From: patrick@mcc-pp.UUCP
Newsgroups: comp.lang.lisp
Subject: Re: LISP floating point performance
Message-ID: <2786@mcc-pp.UUCP>
Date: Fri, 13-Mar-87 20:59:07 EST
Article-I.D.: mcc-pp.2786
Posted: Fri Mar 13 20:59:07 1987
Date-Received: Sat, 14-Mar-87 15:07:15 EST
References: <1368@hplabsc.UUCP> <31800002@ccvaxa> <150@utah-orion.UUCP> <151@utah-orion.UUCP>
Organization: MCC, Austin, TX
Lines: 51
Summary: Re: benchmarks

In article <151@utah-orion.UUCP>, shebs@utah-orion.UUCP (Stanley T. Shebs)
 writes:
> ... some valid comments about why writing a quality Lisp compiler
> is harder than writing a quality C compiler.
I agree the flexibility of Lisp makes it harder to compile for,
but once done, many benefit.

> >Refering to defstructs:
> >This feature may be overlooked by compiler writers since
> >the Gabriel benchmarks do not include structure accesses.
> 
> Check out TRAVERSE.  Granted, there could be more and better benchmarks,
> but after all the abuse RPG took for the incredible work he did,
> I doubt anybody else is going to stick their necks out for quite a while!

Oops.  My oversight.
Also, my apologies to RPG if my comment seems to imply critisicm
for his landmark work in getting Lisp implementers to be more aware
of many performance related issues, yielding faster Lisps for all of us.

> ... Unfortunately, much of current Lisp
> Machine designs are based on performance studies by Clark and Green about
> ten years ago, ... 
My comment was actually intended as a hope that the RPG benchmarks
would not be taken as the final word since current patterns of Lisp
usage are continuing to evolve.  We need benchmarks of more, and varied
programs to continue to challenge compiler writers to continue to improve.
Unfortunately, funding mechanisms are weak as stan points out.
Maybe increased usage of Lisp can create positive feedback here.

I am being vocal about more performance because my pre-Lisp background
is C.  Thus, Lisp offers much that I did not have before in many areas
such as dynamic storage management, data structures and much more.
If it also had the performance of C programs, then its usage would
expand even more rapidly.

> The sensitive implementor gets gray
> hairs trying to find a compromise, while many others get truculent and
> suggest that you should be using specialized hardware, or that people
> doing floating point deserve to use Fortran, or even that you should never
> care about execution speed.  I've heard everything, at one time or another.
Just adding my voice to the chorus of demands :-)

Perhaps the frequent comments of "let the optimizer take care of that"
during the Common Lisp standardization efforts were said too frequently.
Then again, maybe not...

Seriously, Stan Shebs and the PSL bunch are working hard in the
right direction as are a number of other Lisp compiler groups.
My original comments were triggered by the "let them use Fortran"
comments.  I've been there and do not want to go back.