Hello, just to add somethink that wasn’t mentioned.
Some applications that require numeric precision “need” to be ran at 64bits machine. I run numeric Ocean models at both 32 and 64 machines and the result differences are sometimes scary. I’m not talking about performance, but the actual results are different due to the precision!!! Anyone that understand numerics and a little bit of caos knows that it can even lead to wrong conclusions/interpretation of the results due to 32bit truncation.