[e-lang] E Benchmarks
kpreid at mac.com
Wed Nov 29 23:30:06 CST 2006
The E implementation benchmarking tool I've been working on is now in
my Subversion repository.
Things that need improvement:
- Measure CPU time as well as real time.
(This requires support from the E implementation.)
- Separating into emakers.
- Better handling for the multiple-data-set-size tests.
- Skip benchmarks that throw, instead of dying. There's currently
one which fails on E-on-Java because of a use of .diverge(any).
- Tweakable repetition count control scheme.
- Less wasted time in the repetition selection and empty-routine
- More benchmarks.
(Please help me think of them!)
The current tests included are very much just the things I wanted to
test for a particular change I was considering in E-on-CL; they are
not broad at all.
Kevin Reid <http://homepage.mac.com/kpreid/>
More information about the e-lang