Danny,

Thanks for this information. It looks really strange but it raises a
question. Do you have by any chance some AMD or any other non-Intel
processor in your machine?! Also, do you know *exactly* what processor
you're running on?! I have the impression that MSVC might generate code that
could be flaky with some hardware.

> Aside from noticing that each error occurs on a floating 
> point instruction with an operand of [esp+18h],

Which looks extremely weird. The instructions you're showing should not
generate any such exception - here's what the documentation says about these
exceptions:

"EXCEPTION_FLT_STACK_CHECK 
The stack overflowed or underflowed as the result of a floating-point
operation."

Quite unlikely in the code that's been shown...

"EXCEPTION_FLT_INVALID_OPERATION 
This exception represents any floating-point exception not included in this
list."

How instructive... Nevertheless what floating point error could eventually
occur when just trying to store a value from the FPU?!

  Andreas