Re: Precision problem

[George Russell]

Julian Assange wrote:
> Microsoft VCC once (still?) suffers from this problem. Whether
> it is because it accesses random, unassigned memory locations
> or because the optimiser has time thesholds, is unknown.
Optimisers for Intel can produce different results on floating point
because floating point registers have greater precision than floating point
in memory.  Therefore an optimiser which makes more use of registers can
lead to different (hopefully more precise) results, unless you go to extra 
trouble to enforce IEEE arithmetic.