This is the mail archive of the
mailing list for the Java project.
Re: Using GMP for BigIntegers.
"Mark J. Roberts" wrote:
> I have some problems with the BigInteger support in GCJ.
> It's broken, and it's slow. In one test case:
> GMP: 14 milliseconds per modPow. // compiled with GCJ using CNI
> IBM: 38 milliseconds per modPow.
> GCJ: 212 milliseconds per modPow. // wrong result due to bug
Hmm. Can you post the test case?
> So, is GMP in your future?
GMP was actually the original plan. Then Per donated his pure-Java
implementation, so we decided to use that instead. So basically: Yes. If you
want to implement it, I for one would be keen to see it in libgcj.
[ bryce ]