Fabio,
>
>FPU have no influence to this bug.

This is your opinion. This is caused by the limitation of the number of significant digits. Simply shortening either variable by one digit from the sample shown in the KB article eliminates the problem.

>All new Intel and all AMD processor donnot have the pentium bug.

This is not a bug but a limitation of the storage capacity of 8 byte double precision floating point numbers.

>THIS IS THE MOST SERIOUS BUG ON VFP.

Again, your opinion. Anyone with the knowledge of the limitations of 8 byte double precision floats can easily avoid the problem.