lua-users home
lua-l archive

[Date Prev][Date Next][Thread Prev][Thread Next] [Date Index] [Thread Index]


Hi Mike, all,

probably this is not related to this problem but I've seen a problem
with LuaJIT2-beta6 when running with the option '-O1'. I've seen that
with the rkf45.lua.in ODE integrator, when running with -O1 it does
not complete the calculations and it does gives a "nan" almost
immediately. Here a sample of the output:

franko@vaio:~/sviluppo/gsl-shell$ time luajit-2.0.0-beta6 -O3
benchmarks/odeiv/odeiv-benchmark.lua
20000	-1.6463252205064	0.095568735238432
20000	-1.6463251027025	0.095568649533797
20000	-1.6463251027025	0.095568649533797
20000	-1.6463251027025	0.095568649533797
20000	-1.6463251027025	0.095568649533797
20000	-1.6463251027025	0.095568649533797
20000	-1.6463251027025	0.095568649533797
20000	-1.6463251027025	0.095568649533797
20000	-1.6463251027025	0.095568649533797
20000	-1.6463251027025	0.095568649533797

real	0m0.515s
user	0m0.504s
sys	0m0.004s

and

franko@vaio:~/sviluppo/gsl-shell$ time luajit-2.0.0-beta6 -O1
benchmarks/odeiv/odeiv-benchmark.lua
20000	nan	nan
20000	nan	nan
20000	nan	nan
20000	nan	nan
20000	nan	nan
20000	nan	nan
20000	nan	nan
20000	nan	nan
20000	nan	nan
20000	nan	nan

real	0m0.019s
user	0m0.012s
sys	0m0.008s

For me this is not a big problem but it does probably reveal an
underlying bug related to the "-O1" option. I guess that you may tend
to mostly debug the options "-O2" or "-O3".

I guess you already have the code to reproduce the problem but if you
want I can send you a simple "test case" code.

-- 
Francesco