When I first heard about the pentium bug, I thought the i387 assembly
versions of the math functions were safe, because the fdiv instruction
is not used.  I have recently been told that the bug effects more than
just fdiv.

I will be implementing the official Intel workarounds in the math
library as soon as can get my hands on them.  It should also be fairly
easy to integrate the patch that will fix gcc.

Until then, don't do numerics work on pentium processors.

	--jtc