Sounds like IBM FortranH. IBM's FortranG was good, but often slow. It had
different optimization settings but none of the settings violated either the
properties of Fortran (or that Fortran programmers counted on--definitely not
the same thing).
FortranH on the other hand was willing to make lots of
assumptions to generate faster code. Optimization level 0 was expected to be
the same as FortranG's best optimization setting, and levels 1 and 2 were quite
willing to make assumptions that had no basis in the Fortran language--any
particular standard version, not that IBM cared about such things then.
So the
usual was to develop with FortranG. If the code ran fine, you could try taking
out the debugging statements and see if it still ran, or you could try to use
FortranH. Fixed all the "bugs" at level 0? Try level 1. As for level 2, you
had to be a glutton for punishment, or developing code that would consume days
of IBM mainframe time to even try setting it. [ Reply to This | Parent | # ]
|