[PPL-devel] A comparison with CDD

[Roberto Bagnara]

Hi there,

here is how the PPL compares with CDD
(http://www.cs.mcgill.ca/~fukuda/soft/cdd_home/cdd.html)
on the benchmark suite distributed with version 0.93c of cddlib,
on a PC  equipped with an AMD Athlon 2800+ with 1 GB of RAM and
(of course) running GNU/Linux.  The benchmarks for which both
CDD and PPL took less than 0.03 seconds have been excluded from
the following table because they contribute nothing but noise;
the 7 benchmarks meant to test the CDD simplex solver have been
excluded too; all the other benchmarks in the distribution
appear below.

Benchmark		CDD time (s)	PPL time (s)
----------------------------------------------------
ccc6.ext		0.80		0.07
ccp6.ext		1.12		0.11
cross10.ine		11.89		0.17
cross12.ine		213.41		2.46
cross6.ine		0.05		0.01
cross8.ine		0.70		0.04
cube10.ine		0.33		0.09
cube12.ine		1.74		1.64
cube8.ine		0.06		0.02
cyclic14-8.ext		0.09		0.02
cyclic16-10.ext		0.29		0.07
hexocta.ine		0.04		0.01
icododeca_m.ine		0.05		0.03
irbox200-4.ext		1.24		0.05
integralpoints.ine	0.06		0.01
kkd27_5.ine		0.10		0.09
kkd38_6.ine		0.37		4.30
prodst62.ext		>2h		117.23
project1.ine		0.05		0.01
project2.ine		0.57		0.03
project2res.ine		0.11		0.02
reg600-5_m.ext		159.26		8.02
rhomtria_m.ine		0.06		0.02
sampleh8.ine		>2h		111.57

Cheers,

     Roberto

-- 
Prof. Roberto Bagnara
Computer Science Group
Department of Mathematics, University of Parma, Italy
http://www.cs.unipr.it/~bagnara/
mailto:bagnara at cs.unipr.it