I'm sure you didn't mean to be dismissive, but I don't understand your
attitude there. If Python is in the "pretty slow runtime", then NekoVM
also belongs here.
Hopes that the Neko/Lua comparison clear the misunderstanding.
- fannkuch is IMHO impossible to benchmark, with < 10ms running time
I may be doing something wrong but I got the following results with N=10
ocaml 1.18s
java 2.95s
lua 47.21s
neko 1m43s
python 2m48s
I'll have a look at it, the tests I did were not correct.
Are you planing to implement all shootout benchmarks ? Could you send
the ones you did for code review and inclusion on CVS ? Once I get
enough of them, we might be able to do a proper submission for Neko to
be listed in the shootout.
- binary-trees where 3.5 times faster in Neko than in Lua. This
benchmark mesure integer calculs, function call overhead, and
allocation of small objects. It's IMHO the most "generic" benchmark
among these 4.
I got the same results and Neko is really fast here.
I think this is important.
You might want to add a part for defining PPC registers. For example :
Note that I don't know anything about this, is it worth testing for
eventual inclusion in the default Neko distribution?
Yes it is, there might be a significant speed increase by doing so. Or
at least, could you send me the output of :
gcc -O3 -S vm/interp.c
This should produce a interp.s file.
Try also modifying vm/interp.c with the registers optimizations and tell
me how it goes for the benchmarks. If you have time for it of course.
Best,
Nicolas
--
Neko : One VM to run them all
(http://nekovm.org)
