floating point problems in current EGCS?

[fnf@toadfish.ninemoons.com]

I see some previous discussion of problems found running "paranoia"
and indeed, I seem to be able to reproduce them on RedHat Linux 6.0
using the current EGCS sources updated just a couple hours ago.

Is anyone actively working on resolving the problems?  Are these
definitely problems with EGCS or something else, like some subtle
problem with the paranoia program itself?

Thanks for any info...

-Fred