Sort of hard to count properly.  DARA funded more than just CSRG.    The first DARPA money associated with large address work was in 1980 for the Vax version of Maxima, although there might have been some other projects before that.  But the CSRG project had not (yet) started when I arrived in Fall of '81, although the Stanford bake-off had and 4.1BSD was on the street - as Joy had done the FASTVAX work earlier than spring/summer.  At that point there were two 780s for CS in Evan Halls (Ernie and Kim). The 'Vaxima' port was working and I believe that the first Franz-LISP release was out.  FWIW:  The CAD group had just gotten its first in Cory which was funded by the DEC/HP/IBM and some of the Si manufacturers.  I believe that Ingress was still an 11/70, but I would defer to Eric on that - I'm not sure who/what was their funding source, although they had the original ARPAnet connection via a VDHI to LBL, so DARPA must have been involved.   The C30 IMP (for CSRG) did not arrive in Evans until at least mid 82, I think maybe '83. 

So by 1981/82 the question of how UNIX (as a system) was going to be supported for the ARPA contractors had become a ripe concern.  It must have been proposed by then and being negotiated.    IIRC Prof Fabrey was officially the PI for what would become CSRG, but some one like Eric or Mary Ann who was already there before me might have more information on the politics