Page 1 of 1

A good benchmark for a garbage collector

Posted: Mon Feb 25, 2013 6:31 am
by schoppenhauer
Hello.

I am currently experimenting with a few ideas on garbage collection, and I am looking for good benchmarks for them - algorithms that exhaust it, etc. One problem I see is that it is probably pointless to compare them to the widespread GCs. On the one hand, it is probably easy to top something like Boehm-Weiser, because it is optimized for working in a non-cooperative environment. On the other hand, the GCs of Lisp- and Haskell-Implementations are probably very sophisticated (generational, with a lot of heuristics and code optimization), while I just want to try out some ideas.

Re: A good benchmark for a garbage collector

Posted: Mon Feb 25, 2013 10:32 am
by nuntius
"The right" benchmark depends a lot on what you are trying to investigate. Where do you expect your algorithm to excel or fall behind -- those areas may need a special test. They may even need a particular programming style.

See for example the old CL FAQ discussion on cdr-coding.
http://www.faqs.org/faqs/lisp-faq/part2/section-9.html


Here are some lispy benchmarks to consider.

http://common-lisp.net/project/cl-bench/
http://www.hpl.hp.com/personal/Hans_Boe ... bench.html
http://www.cesura17.net/~will/Professio ... ecode.html

Many more can be found by searching for things like "gc benchmark".

Re: A good benchmark for a garbage collector

Posted: Wed Mar 13, 2013 12:35 pm
by nuntius
Here are a couple Java benchmarks that may be useful.

http://dacapobench.org/
http://www.spec.org/jvm2008/