mark@umcp-cs.UUCP (09/30/83)
How does the "New Kernel!" compare with 2.9 unix on small machines,
which claims almost exactly the same advantages that these authors
claim?
--
spoken: mark weiser
UUCP: {seismo,allegra,brl-bmd}!umcp-cs!mark
CSNet: mark@umcp-cs
ARPA: mark.umcp-cs@UDel-Relaycarl@ucbvax.UUCP (10/02/83)
This message is in response to Mark Weiser's inquiry about 2.9BSD vs. PSUnix 1.0: > How does the "New Kernel!" compare with 2.9 unix on small machines, > which claims almost exactly the same advantages that these authors > claim? We claim many more advantages than those mentioned in John Irwin's article (260@psuvax.UUCP). Namely: - Ours is a complete distribution, not just a new kernel (although one would reasonably assume that a C compiler and loader are supplied with their version, too). - Many (literally hundreds) of bug fixes and enhancements have been made in the user level software as well as the kernel. Much of the new software that will come with 4.2BSD is also included. - There are impressive speed improvements in system calls, illegal instruction traps, and overlay switches. For example, on machines without FP11 floating point boards the USG floating point benchmark runs twice as fast on one of our systems as it does under System-III (no one has yet volunteered a System-V machine for us to test). - The 2.9 file system is twice as fast as V7 or DEC Unix/v7m file systems. Our kernel also knows how to do file system updates in the proper order, so that damage is minimized. - Our kernel supports dynamic hardware configuration at boot time. In particular, such issues as separate vs. nonseparate I/D cpus and MASSBUS vs. UNIBUS disks and tapes are handled transparently so that the same kernel is capable of running on many different machines. - Our kernel supports more devices than either V7 or DEC Unix/v7m. Both ECC and DEC standard 144 bad sector forwarding is supported for controllers capable of them. - Our kernel also supports vforks and automatic reboots. This is in no way an attempt to discourage either the Penn State group or anyone interested in their software. Indeed, there are some very clear advantages to their distribution. For example: - They are offering support. We try to be conscientious and prompt on bugs reported to us, but make no offer of support. - You wouldn't have to sign the now infamous Berkeley license agreement. - It will certainly be cheaper than the $200 we're asking. Carl Smith (ucbvax!carl ; carl@berkeley) Manager Second Berkeley Software Distribution
@pesnta.UUCP (10/15/83)
No, I'm not talking about Berkeley 4.2BSD, and I'm not talking about Bell
Labs version 8. I'm talking about PSUnix 1.0. Let me explain.
About nine months ago we received our Dec v7m (vanilla) version 7 over-
layed kernel source for our Pdp-11/34A. At that time, we had a bare-bones
11/34A with only a RK07 and no floating point hardware. Since then, we have
added several small disks and a 2K cache, but we are still basically running a
'small' system. Anyway, when we first booted ol' version 7, we found some of
its most interesting features.
#1. It was slow. Make that s.....l.....o.....w..... (spelled: zzzzzz)
I really can't see how Bell claims ten users, we found two hard to
handle.
#2. It liked to jump out the window at night. Seriously, the stupid
thing would "panic: trap" all the time!
#3. It just didn't DO a lot. Being used to working on an 11/780, it's
pretty hard to exist without ^Z stopping jobs, much less without
backspace erasing characters!
And so nine months ago we (oops, sorry, 'we' are Professor Robert Michael
Owens, John Irwin, and David Eckhardt) set out to transform version 7 into
something useable. To make a long story short, we think we've done so. So now
we have considered trying to market/give-away our UNIX. Like 4.1, it would
require a V7 license. But before we do, we need a 'Beta' test site to deter-
mine if PSUnix will actually work in the 'real world' (ie: NOT the middle of
rural Pa!). Thus this rather lengthy letter is request for such a site. The
prospective site would need the following:
1) A small Pdp-11. (ie: 24, 34, 44, 45, 60)
2) A V7 source license. (A binary might do, although I'm not sure..)
Also, A DEC v7m (overlayed) license is probably needed, as
well as a 4.1BSD for 'vi' and 'csh'
3) Some time and patience.
In return, said site would receive a working Unix that is many TIMES fas-
ter than V7, and supports the following:
1) GREATLY increased reliability.
2) 99% Berkeley 4.1bsd compatibility. (we run verbatim (word, not
floppy) copies of Berkeley's 'csh' and 'vi', as well as others)
3) An overlaying system which provides for the running of large
programs, such as netnews B2.10 and Rogue5.2.
In conclusion, we would like a small 11 to volunteer to test out our Unix
on their system. In return, they would receive the complete kernel and
sources, as well as support from us. If it works, we may give/market it in
a few months. Sorry about this long letter in many newsgroups, but I feel this
announcement is important enough to merit it.
John Irwin (814) 238-7556
The Pennsylvania State University
{allegra, ihnp4, burdvax}!psuvax!jdi
or, by USnail,
Dr. Owens (814) 863-1415
308 Whitmore Building
The Pennsylvania State University
University Park, PA 16082