brent@itm.UUCP (Brent) (11/02/84)
At NCC, Perkin-Elmer claimed they would have System V.2 by October.
Well October has come and gone, and none have been shipped. They
say they'll start shipping them Nov. 2. When we'll receive ours,
no one can say.
What is P-E's UNIX strategy? Following closely in AT&T's steps
is what they appear to be doing with V.2. To make this work, they
must either offer a better cost/performance ratio than AT&T, or offer
something AT&T doesn't. P-E has strengths in the defense industry.
This means FORTRAN. Under their OS/32, they have a super-optimized
FORTRAN compiler. There is talk of bringing this over to their UNIX.
This would continue their concentration in defense. P-E is also big
into real-time applications with OS/32. Not much under their UNIX
for this, though. They do have a utility to lock a process in main
memory, but that's about it.
What of the business-oriented community? This is where it gets
strange. If they are going for the scientific/technical community
as they appear to be doing with FORTRAN, why System V? 4.2 might
have been a better choice, to talk with the VAXen running BSD. If
they are going after the UNIX business community, as they appear to
be doing with Sys V, then why do they have an optimized FORTRAN, and
no C optimizer? How many folks running pure Bell UNIX out there
write your application programs in FORTRAN? Thank you, I see both
of you raised your hands.
More later. We'll see.
--
Brent Laminack (akgua!itm!brent)gwyn@brl-tgr.ARPA (Doug Gwyn <gwyn>) (11/05/84)
> What is P-E's UNIX strategy? Following closely in AT&T's steps > is what they appear to be doing with V.2. To make this work, they > must either offer a better cost/performance ratio than AT&T, or offer > something AT&T doesn't. For instance, a version of UNIX that runs on P-E hardware. They need to make their hardware cost/performance competitive with other 32-bit vendors offering UNIX; nothing new in that. > P-E has strengths in the defense industry. > This means FORTRAN. Under their OS/32, they have a super-optimized > FORTRAN compiler. There is talk of bringing this over to their UNIX. > This would continue their concentration in defense. Certainly, a good Fortran would help bring over their established customer base. > P-E is also big > into real-time applications with OS/32. Not much under their UNIX > for this, though. They do have a utility to lock a process in main > memory, but that's about it. A good first step is the adoption of UNIX System V, which has better IPC support for real-time applications than 4.2BSD. > What of the business-oriented community? This is where it gets > strange. If they are going for the scientific/technical community > as they appear to be doing with FORTRAN, why System V? 4.2 might > have been a better choice, to talk with the VAXen running BSD. If > they are going after the UNIX business community, as they appear to > be doing with Sys V, then why do they have an optimized FORTRAN, and > no C optimizer? 4.2BSD is not inherently more suited to scientific applications than UNIX System V. The academic community happens to be using 4.nBSD more than AT&T USG/UDL UNIX for historical reasons. Vendors also have to look at future directions. Indeed, once Berkeley and AT&T straighten out the licensing issue, there is no good reason why the two variants should not come to look more alike from the applications standpoint. The big lack in UNIX System V `a la AT&T right now is decent network support. Some software houses have developed TCP/IP and network device drivers for their UNIX System V offering; P-E is advised to do the same. > How many folks running pure Bell UNIX out there > write your application programs in FORTRAN? I venture to say that most major scientific computation is still Fortran-based, UNIX or no. Is LINPACK available in C? EISPACK? IMSL?