On 19/09/15 21:25, Steve Dougherty wrote: > Does anyone have benchmarks that demonstrate native acceleration having > significant performance improvements? Keeping it around makes for > maintenance, and anecdotes suggest native Java performance is > sufficient. If it's no longer a clear benefit to have native > acceleration I'd like to look into removing it to lower maintenance > load. Thoughts? >
For my memory many years ago, which may be inaccurate, I think the performance benefit of libjbigi was basically wiped out by Java 7's reimplementation of BigInteger. I wrote some tests, which might still work today: https://github.com/freenet/fred/tree/bigint7 There is also my library which uses GCJ's BigInteger (which indirectly uses GMP), with some more tests: https://github.com/infinity0/gmp-java And I seem to remember that my library has better performance than libjbigi as well. However, all of this is probably insignificant compared to the developer cost of maintaining this code. X -- GPG: 4096R/1318EFAC5FBBDBCE git://github.com/infinity0/pubkeys.git _______________________________________________ Devl mailing list Devl@freenetproject.org https://emu.freenetproject.org/cgi-bin/mailman/listinfo/devl
