Precision problem with matrix inversion

[Werner]

Cyrille:

The 12/15 digit algorithm as used by the venerable 42S has no problems with this system of equations, so I don't quite see why the Prime would.

Everyone else:
B A / is not the same as inv(A)*B.. The former is faster and more accurate, as it does not compute the inverse. Or at least that's how it has been in the 71B, 42S, 48/49/50.

Cheers, Werner