Richard Zidlicky wrote:
>> Depends on whether the underlying FPU of the host processor has bugs
>> or not ;-)
> do you have the pentium division bug workaround?
No, I don't employ it. I've read Dr. Nicely's view of the bug and
agree that unless somebody wants to study chaos theory or similar
number crunching with QPC (*on a very old Pentium I system*) it
doesn't really pose a problem.
> x86 FPUs are very innovative so I would expect a lot of interesting
> bugs.
I believe the x86 FPUs are, if anything, much more bug-free than the
SMSQ/QDOS routines, the former being used by millions of people. As
for the FDIV bug, Dr. Nicely's PCs worked on his calculations for
several months 24 hours per day before the bug showed up. And that was
the biggest bug in the "recent" x86 FPU history I know of.
> To make it more interesting, there are probably many more x86
> "compatible" CPU types around than there are QPC users.
I don't think so.
Marcel
