Re: [Sbcl-help] cl-bench Windows results SBCL vs. Clozure

[Brandon V. E. <bva...@gm...>]

On Fri, Apr 15, 2011 at 5:02 PM, Nikodemus Siivola
<nik...@ra...> wrote:
>> as we established earlier, piggish in heap footprint.  Clozure does
>> better on most of the CLOS tests.
>
> CL-BENCH CLOS benchmarks are mostly notorious for not actually
> benchmarking CLOS performance. (Many of them actually measure compiler
> speed, not speed of compiled code. Several of those that measure CLOS
> speed measure atypical things. Several typical and interesting
> use-cases are not benchmarked at all.)
>
> Just FYI.

Yeah, well, the point still stands, SBCL and Clozure "are comparable."
 You can't very well distinguish their performance based on cl-bench.
I don't think cl-bench is a likeable benchmark, going into the future.
 I can think of lots of maintenance and deployment issues by which it
could be improved.  To do all of that, I don't think I'd call the
results "cl-bench" anymore.  It accomplished my present purpose, which
was to establish the rough speed of Clozure compared to other compiled
CLs.  And by extension, if you look at
http://shootout.alioth.debian.org , I'd say the better compiled CLs
are in roughly the same performance class as Java and C#.  They're not
at the level of C/C++.  They may obtain that performance in specific
cases, as might Java or C#.  But in general they won't, and they'll
also use far more memory to do it.  Also worth noting, is the better
compiled CLs are as good or better than any of the other functional
programming languages.  Compiled CLs perform way better than scripting
languages, and to me the possibility of doing all aspects of
development with the same language is a major selling point.


Cheers,
Brandon Van every