[tools-compilers] Studio 12: ellided rand() calls in f95 w/ -xOn for n >= 1

[Reginald Beardsley]

I've discovered that turning on any optimization causes calls to rand() to be 
almost completely ellided.  Out of 20,000 calls, only ~4 actually get made.

google & sun solve both fail to turn up any mention of the problem or mention 
of a pragma that would prevent this.

What am I missing?  

Thanks,
Reg
--
This message posted from opensolaris.org