CMU Artificial Intelligence Repository
Prolog Benchmarking Suites
lang/prolog/code/bench/
This directory contains code for benchmarking Prolog implementation
compatibility and performance.
Naive reverse is the classical benchmark for computing LIPS (Logical
Inferences Per Second). However, the utility of small benchmark
programs like naive reverse (NREV) is questionable, since it may not
necessarily reflect overall system performance. For example, NREV
doesn't exercise the two main features of Prolog -- backtracking and
unification. In practice, performance on large programs is much more
important. Some of the benchmarks use larger programs, such as
cryptarithms, theorem provers, and chart parsing, to give a better
picture of system performance on real problems. Thus it is better
to use the Pereira or Holmer benchmarks than the NREV benchmark.
CD-ROM: Prime Time Freeware for AI, Issue 1-1
Keywords:
8-Queens, Benchmarks!Prolog, Chart Parsing, Cryptarithms,
Database Querying, Parsing, Prolog!Benchmarks, Prolog!Code,
Quicksort, Symbolic Differentiation, Theorem Proving
Contains:
(WGSYM of Information Processing Society of Japan)
wgsym.tar The 1985 WGSYM Lisp/Prolog contest benchmarks
portland.tar The Tektronix/Portland State Univ benchmarks
plbench.txt A short post of benchmarks by Brahme
the costs of tail calls vs. nontail calls.
term matching and term construction costs and
benchmark suites, such as procedure call cost,
separate costs normally conflated in other
a Prolog system. This benchmark tries to
strengths and weaknesses in the basic engine of
These benchmarks are designed to identify
pereira.tar Fernando Pereira's benchmarks (26-DEC-86)
pereira.txt
Zurich University
Norbert Fuchs, Department of Computer Science,
nrev.pl Prolog "Naive List Reverse" Benchmark
parsing, among others.
theorem provers, database querying, and chart
quicksort, 8-queens puzzle, cryptarithms,
benchmarks. Includes symbolic differentiation,
holmer.tar Bruce Holmer, Peter Van Roy, and Alvin Despain's
ecrc.tar The ECRC benchmarks
dobry.txt Dobry's benchmarks
References:
holmer.tar contains the benchmarks that were used in the papers:
Bruce Holmer et al (holmer@ernie.berkeley.edu), "Fast Prolog with an
Extended General Purpose Architecture", 17th International Symposium
on Computer Architecture, May 1990.
Peter Van Roy (vanroy@ernie.berkeley.edu) and Alvin Despain, "The
Benefits of Global Dataflow Analysis for an Optimizing Prolog
Compiler", 1990 North American Conference on Logic Programming,
October 1990.
Last Web update on Mon Feb 13 10:33:25 1995
AI.Repository@cs.cmu.edu