Path: utzoo!utgpu!watserv1!watmath!att!rutgers!cs.utexas.edu!sdd.hp.com!ucsd!ucbvax!CS.ARIZONA.EDU!kwalker
From: kwalker@CS.ARIZONA.EDU ("Kenneth Walker")
Newsgroups: comp.lang.icon
Subject: RE: UUXXCODE
Message-ID: <9009192222.AA24483@gacham.cs.arizona.edu>
Date: 19 Sep 90 22:22:40 GMT
Sender: daemon@ucbvax.BERKELEY.EDU
Distribution: inet
Organization: The Internet
Lines: 41


	Date: 18 Sep 90 16:13:58 GMT
	From: esquire!yost@nyu.edu
	
	In article <0093CBF0039A4D60.20400E83@mis.mcw.edu> tenaglia@mis.mcw.edu
        ("Chris Tenaglia - 257-8765") writes:
	>
	>I've attempted a port to Icon of UUENCODE and UUDECODE.
	
	Now there's a couple of programs I bet are really slow in Icon.
	I'd love to see how fast they run with the Icon compiler,
	compared to the interpreter and compared to the original
	C version.
	
	If the compiler can really do a good job on this sort of repetitive
	low-level stuff that kills Icon performance, it would be the dawning
	of a new day!  Imagine: we could use Icon instead of C for everything!
	
	Ken, can we see a benchmark on this?
	

I tried 3 runs of each program using the Unix time command and got a range
of results. The uuencode.icn I used has a fix that is not in the version
Chris posted. I also removed the write() expressions that print the number
of bytes in each group processed.

 uuencode:
     compiled is 4.0 - 5.8 times faster than interpreted
     system version is 21 - 40 times faster than compiled

 uudecode
      compiled is 2.4 - 2.9 times faster than interpreted
      system version is 22 - 36 times faster than compiled

While the time command is clearly not a very accurate measure of program
speed (I used the same data on all 3 runs), it does give a feeling
for how much the compiler improves speed and how much work is left to
do to get programs like these to run as fast as those coded in C.

  Ken Walker / Computer Science Dept / Univ of Arizona / Tucson, AZ 85721
  +1 602 621-4324  kwalker@cs.arizona.edu {uunet|allegra|noao}!arizona!kwalker