[plt-scheme] DrScheme Faster on Powerset than MzScheme - Why?

On Mar 15, Kyle Smith wrote:
> [...]
> I was really surprised that DrScheme beat out MzScheme.  Do I have
> something setup wrong on the command line of MzScheme, as I don't
> often use the stand-alone interpreter, so I could be missing
> something?

Your results show that running the code in MzScheme is consistently
faster -- if you ignore the GC time -- you have:

  (- 469 265) -> 204
  (- 469 266) -> 203

in DrScheme, and

  (- 563 405) -> 158
  (- 563 376) -> 187

in MzScheme.


On Mar 15, Jon Rafkind wrote:
> 
> [...]
> But mzscheme seemed to do much better the 2nd or 3rd time I ran
> (go):
> [...]

Yes -- it's common to have less GC time when there is a larger initial
footprint.  Looking at the CPU time minus GC time is something I
learned from Matthew after noticing strange things a few times.  For
example, some code was running slower when it was compiled to .zo, and
the reason is that compiling the file takes more resources which makes
a larger heap before running the code.  (The code that is getting to
run is identical in both cases.)

-- 
          ((lambda (x) (x x)) (lambda (x) (x x)))          Eli Barzilay:
                  http://www.barzilay.org/                 Maze is Life!