CompuServe Thread

Forum unknown · Hot News & Rumors

#New Amiga's?

15 messages in this thread
#117176From: John DraperApr 3, 1988 12:04 AM
Contrary to current urban myth… Unix does not require a 68020 or an MMU to run. It on;y make it easier.
#117196From: Don Curtis/SYSOPApr 3, 1988 12:50 AM
The 68020 I understand…it should run fine with a 68010. As to the MMU, I'm only repeating what Hazy said.
#117210From: John DraperApr 3, 1988 3:16 AM
A lot of early Unix boxes ran without MMU… it ain't easy, but these days, it seems to be a generally held belief that an MMU is absolutely necessary.
#117210From: John DraperApr 3, 1988 3:16 AM
A lot of early Unix boxes ran without MMU… it ain't easy, but these days, it seems to be a generally held belief that an MMU is absolutely necessary.
#117196From: Don Curtis/SYSOPApr 3, 1988 12:50 AM
The 68020 I understand…it should run fine with a 68010. As to the MMU, I'm only repeating what Hazy said.
#117237From: Peter da SilvaApr 3, 1988 11:25 AM
UNIX needs some sort of MMU to allow an efficient implementation of the fork() system call. But it definitely doesn't need a 68020. We have a 68000 based UNIX box at work that blows the doors off the intel 80386 boxes.
#117290From: John DraperApr 3, 1988 5:28 PM
Like I said, it only makes it easier to implement Unix. Some of the truth of that depends heavily on what you consider a full implementation of Unix, and how willing you are to chance a process being clobbered by a runaway.
#117309From: Peter da SilvaApr 3, 1988 7:00 PM
It also depends on whether or not you're willing to let a processes entire address space be copied on every context switch until one That depends on whether or not you';re willing to accept having a processes entire address space be copied on every context switch between the time you do a fork() and one side either exec()s or exit()s.
#117309From: Peter da SilvaApr 3, 1988 7:00 PM
It also depends on whether or not you're willing to let a processes entire address space be copied on every context switch until one That depends on whether or not you';re willing to accept having a processes entire address space be copied on every context switch between the time you do a fork() and one side either exec()s or exit()s.
#117290From: John DraperApr 3, 1988 5:28 PM
Like I said, it only makes it easier to implement Unix. Some of the truth of that depends heavily on what you consider a full implementation of Unix, and how willing you are to chance a process being clobbered by a runaway.
#117237From: Peter da SilvaApr 3, 1988 11:25 AM
UNIX needs some sort of MMU to allow an efficient implementation of the fork() system call. But it definitely doesn't need a 68020. We have a 68000 based UNIX box at work that blows the doors off the intel 80386 boxes.
#117243From: Vic WagnerApr 3, 1988 12:18 PM
How do you implement a reliable sbrk() function without an MMU?
#117292From: John DraperApr 3, 1988 5:31 PM
No idea how Vic… just that there have been unixes (Unices?) implemented on 68000 non MMU machines.
#117292From: John DraperApr 3, 1988 5:31 PM
No idea how Vic… just that there have been unixes (Unices?) implemented on 68000 non MMU machines.
#117243From: Vic WagnerApr 3, 1988 12:18 PM
How do you implement a reliable sbrk() function without an MMU?