malczews@castor.usc.edu (Frank Malczewski) (03/07/89)
Except for my vague awareness of an item by Richard Gabriel called Performance and Evaluation of LISP Systems (anyone have more detail on this?), I am looking for information pertaining to this topic in terms of overview articles of the types of benchmarks generally performed, perhaps some comparisons of actual implementations, what the benchmarks try to show, etc. Also, might the Gabriel benchmark set be available somewhere (readily accessible)? Can someone be of assistance? -- Frank Malczewski (malczews@castor.usc.edu) -- Frank Malczewski (malczews@castor.usc.edu)
preston@felix.UUCP (Preston Bannister) (03/09/89)
From article <2960@nunki.usc.edu>, by malczews@castor.usc.edu (Frank Malczewski): > Except for my vague awareness of an item by Richard Gabriel called > Performance and Evaluation of LISP Systems (anyone have more detail on > this?), I am looking for information pertaining to this topic in terms of > overview articles of the types of benchmarks generally performed, perhaps > some comparisons of actual implementations, what the benchmarks try to > show, etc. I picked up a copy of Gabriel's book from the local book store. Gabriel's book represents a fair amount of work, is probably one of the first things you should look at. He collected a representative set Lisp programs and ran them on a large variety of systems. Interesting reading. -- Preston -- Preston L. Bannister USENET : hplabs!felix!preston BIX : plb CompuServe : 71350,3505 GEnie : p.bannister
fischer@arisia.Xerox.COM (Ronald A. Fischer) (03/10/89)
Be aware that all of the Lisps described in Gabriels' book have improved or changed performance. His real contribution was to get a standard set of routines that could be used to performance test systems. However, the book can't really be used to directly compare the implementations tested any longer. (ron)
malczews@castor.usc.edu (Frank Malczewski) (03/12/89)
In article <645@arisia.Xerox.COM> fischer@arisia.Xerox.COM (Ronald A. Fischer) writes: > > >Be aware that all of the Lisps described in Gabriels' book have improved or >changed performance. His real contribution was to get a standard set of >routines that could be used to performance test systems. However, the book >can't really be used to directly compare the implementations tested any longer. > >(ron) You are not the only one to say something to this effect; I am interpreting this to mean that the LISPs themselves have changed, and not the value of the benchmarks themselves. Others have implied that the benchmarks themselves are no longer quite as comprehensive, that newer sets may be more effective, but perhaps proprietary, and thus less publicized. Any further comments, as I await procurement of the Gabriel text? -- Frank Malczewski (malczews@castor.usc.edu)