gcj performance on Solaris 2.6

Thu Jul 26 11:24:00 GMT 2001

On 2001年7月26日, Andrew Haley wrote:
> I thought hotspot only optimized if you called th same function many
> times. That's not happening here.

I don't know about hotspot, but for ExactVM (thanks to Kresten for the
other name I forgot) I tried ten iterations of runSieve from main. It
didn't make a significant difference.
I'm not too surprised that gcj can win this benchmark since the main
loops contains no method calls. On many benchmarks I've tried, gcj is
slightly handicapped by dynamic linking alone.
Jeff