Path: utzoo!utgpu!news-server.csri.toronto.edu!mailrus!uunet!mcsun!ukc!mucs!liv-cs!liv!qq11
From: QQ11@LIVERPOOL.AC.UK
Newsgroups: comp.lang.rexx
Subject: Re: Rexx compiler(s)
Message-ID: <90235.095902QQ11@LIVERPOOL.AC.UK>
Date: 23 Aug 90 08:59:02 GMT
References: <90230.092947BOYDJ@QUCDN.BITNET> <90231.095318BOYDJ@QUCDN.BITNET> 
  <26d00a0a-3d.4comp.lang.rexx-1@tronsbox.xei.com> <MWM.90Aug21113015@raven.pa.dec.com> <90234.100626RMCCU@CUNYVM.BITNET>
Organization: University of Liverpool
Lines: 25

In article <90234.100626RMCCU@CUNYVM.BITNET>, RMCCU@CUNYVM (Robert Chambers)
says:
>
>There are some applications where compiling a REXX exec will not affect
>performance much at all, one of these is an exec that does mostly system
>calls....
True.
..
>I think people yearning for speed will have to break down and learn
>assembler, You won't get any faster than that.  Compiled REXX won't come
>close - Unless the REXX compiler is really really intelligent (which I
>doubt)
Assembler isn't always necessary. I and others have found that CMS PIPELINES
(un*x hackers can stop laughing :-)) can give the performance boost required
without having to learn IBM assembler (yuk :-)). The learning curve is easy
and speedups of the order of 10 times have been found.

REXX and PIPELINES were almost made for each other and this solution is also
bit cheaper all round :-)

Alan Thew
University of Liverpool Computer Laboratory
Bitnet/Earn: QQ11@LIVERPOOL.AC.UK or QQ11%UK.AC.LIVERPOOL @ UKACRL
UUCP       : ....!mcsun!ukc!liv!qq11        Voice: +44 51 794 3735
Internet   : QQ11@LIVERPOOL.AC.UK or QQ11%LIVERPOOL.AC.UK @ NSFNET-RELAY.AC.UK