davidb@inmos.co.uk (David Boreham) (08/22/90)
Well, I don't think that anyone has mentioned this yet, so here goes: Everyone seems to be making the assumption that we will want to have a *DIFFERENT* text space for *EACH* processor, and therefore make comments about not being able to afford the disks etc. However, consider the situation where every processor in a computer wanted to see the same virtual address space; or perhaps every computer within the same company; or perhaps all the computers in the world.....universe. Even keeping on the safe side of reality, with maybe 1000 processors in a computer generally available in the next few years, you start to find address bits being used up. 64-bits aught to allow everyone on the planet to use the same virtual address space :) David Boreham, INMOS Limited | mail(uk): davidb@inmos.co.uk or ukc!inmos!davidb Bristol, England | (us): uunet!inmos.com!davidb +44 454 616616 ex 547 | Internet: davidb@inmos.com
jkenton@pinocchio.encore.com (Jeff Kenton) (08/23/90)
From article <9883@ganymede.inmos.co.uk>, by davidb@inmos.co.uk (David Boreham): > > Everyone seems to be making the assumption that we will want > to have a *DIFFERENT* text space for *EACH* processor, and > . . . > > 64-bits aught to allow everyone on the planet to use the > same virtual address space :) > And the kernel code to deal with this wonderful possibility will take up most of it. - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - jeff kenton --- temporarily at jkenton@pinocchio.encore.com --- always at (617) 894-4508 --- - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - - -
aglew@lasso.crhc.uiuc.edu (Andy Glew) (08/23/90)
>> 64-bits aught to allow everyone on the planet to use the >> same virtual address space :) > >And the kernel code to deal with this wonderful possibility will >take up most of it. Only if it has to be backward compatible with every obsolete computer network ever written :-{ -- Andy Glew, a-glew@uiuc.edu [get ph nameserver from uxc.cso.uiuc.edu:net/qi]
usenet@nlm.nih.gov (usenet news poster) (08/23/90)
<9883@ganymede.inmos.co.uk> davidb@inmos.co.uk (David Boreham) writes:
% Everyone seems to be making the assumption that we will want
% to have a *DIFFERENT* text space for *EACH* processor, and
% therefore make comments about not being able to afford the
% disks etc. However, consider the situation where every processor
% in a computer wanted to see the same virtual address space...
This is not so far off. IBM will be glad to show you the space on the
RS/6000 board where the fiber cable goes, and at least one company is
marketting an add in board based on fiber optic links that will permit
shared VM between heterologous processors. The software issues for
the latter boggle my humble mind.
David Stateskhb@chiba.Eng.Sun.COM (Keith Bierman - SPD Advanced Languages) (08/23/90)
In article <1990Aug22.222045.8100@nlm.nih.gov> usenet@nlm.nih.gov (usenet news poster) writes:
..
This is not so far off. IBM will be glad to show you the space on the
RS/6000 board where the fiber cable goes, and at least one company is
marketting an add in board based on fiber optic links that will permit
shared VM between heterologous processors. The software issues for
the latter boggle my humble mind.
Not the least of the problems is that the memory model is weakly
ordered ... from even the CACHEs point of view. Only the CPU itself
has the logic which enables the out of order stores to be viewed in
the right order.
--
----------------------------------------------------------------
Keith H. Bierman kbierman@Eng.Sun.COM | khb@chiba.Eng.Sun.COM
SMI 2550 Garcia 12-33 | (415 336 2648)
Mountain View, CA 94043