Unix on 386 machines vs. 3B2 machines; stat. pack S

[Richard T Carson]

=-=-=--=-=-=-==-=-=--=-=-=-==-=-=--=-=-=-==-=-=--=-=-=-==-=-=--=-=-=-==-=-=--=

I'm sending this from a friend's account.  Please reply to
keynes!shankar at ucsd.edu  ...!ucsd!keynes!shankar

=-=-=--=-=-=-==-=-=--=-=-=-==-=-=--=-=-=-==-=-=--=-=-=-==-=-=--=-=-=-==-=-=--=

I'm thinking about getting a Unix box.  Many people have suggested a
386-based machine.  Others have inclined towards the AT & T 3B2 series.
I plan to use the machine for statistical computations.  One package that
I have in mind is AT & T's S.  S is written in C and fortran and requires
Unix and C and f77 compilers.  It will run on the 3B2 series and on BSD
machines. 
So far I have been unable to find out whether the 386-based machines running
Unix offer something close enough to System V/BSD to allow S to be compiled
and used.
I also don't know what are the pros and cons of 386 machines vs. the 3B1/3B2
series from AT & T.  
Any information on these topics would be appreciated.
thanks
shankar subramanian  (keynes!shankar at ucsd.edu,  ...!ucsd!keynes!shankar)