mikkel@cg-d.UUCP (03/05/87)
For the last three or four years, I've heard rumors of a very
successful multi-processor UNIX system implemented at some (sub)set of
the following schools: {cmu,yale,purdue}.
The processors, I believe, were semi-tightly connected, meaning
independent caches but operating from a common physical memory address
space and a common set of peripherals. As the strategy was explained
to me, one processor ran the kernel, having total access to the
internal data structures. All system calls, from all processes on all
processors, were directed to the designated kernel processor.
Does anyone have experience with such a system? If so, where can one
find appropriate references?
Thanks. As usual, I'll collect email response and forward to those
interested in ease-dropping.
Yours,
Carl Mikkelsen
Carl Mikkelsen ..!decvax!cg-d!mikkel
Compugraphic Corporation (617) 658-5600 x 5220
200 Ballardvale St. {Not responsible for}
Wilmington, Ma. 01887 {much except myself.}ron@brl-sem.ARPA (Ron Natalie <ron>) (03/06/87)
In article <280@cg-d.UUCP>, mikkel@cg-d.UUCP (Carl Mikkelsen X5220) writes: > > For the last three or four years, I've heard rumors of a very > successful multi-processor UNIX system implemented at some (sub)set of > the following schools: {cmu,yale,purdue}. PURDUE did a multiprocessor UNIX by replacing the SBI terminator in their VAX 780 with another CPU. The master processor handles most of the kernel with time left over for user processes and user processes are run in the slave CPU. Gould uses most of the same code to run their dual processor Powernode systems (as a matter of fact, Gould sucked up a lot of the Purdue talent to do it). Doing the Master/Slaves approach is a fairly old idea. It was first published in a paper from the Navy Postgraduate School in 1975. I don't have the reference. -Ron
chinn@apcisea.UUCP (David Chinn) (03/09/87)
In article <280@cg-d.UUCP>, mikkel@cg-d.UUCP (Carl Mikkelsen X5220) writes: > > For the last three or four years, I've heard rumors of a very > successful multi-processor UNIX system implemented at some (sub)set of > the following schools: {cmu,yale,purdue}. Last year at Summmer Usenix in Atlanta, someone delivered a paper on the Mach Operating System... "Mach is a multiprocessor operating system kernel and environment... Mach provides a new foundation for UNIX development that spans networks of uniprocessors and multiprocessors..." This was the abstract from the paper "Mach: a New Kernel Foundation for Unix Development", Accetta, Baron, Bolosky, Golub, Rashid, Tevanian and Young. They are out of Carnegie Mellon University. You can look the paper up in the conference proceedings. hope this helps ...uw-beaver david m. chinnn !tikal!pilchuck apollo computer inc !apcisea!chinn bellevue, washington