jsp@ius3.ius.cs.cmu.edu (John Pieper) (11/10/90)
Is anybody out there old enough to remember when we used overlays? What I want to know is what is the current state of the art in compiler-managed overlays: is this considered a solved problem, or did people just decide that since we had paging it was unimportant? I am aware of the efforts at optimizing programs for virtual memory systems, and the work at compiler-controlled caching, both of which are related problems, but I can't find anything on compiler-managed overlays. replies to jsp@cs.cmu.edu or post here. ---------
johnl@iecc.cambridge.ma.us (John R. Levine) (11/11/90)
In article <11043@pt.cs.cmu.edu> you write: >Is anybody out there old enough to remember when we used overlays? Sure. I used them under PC-DOS in 1987. My impression is that when paging is available, it usually performs much better than overlays for two reasons: one is that the need to overlay is detected by hardware without having to test at each call, the other is that getting a good structure to the overlay tree is hard enough that you're better off letting the pager determine the actual working set. Regards, John Levine, johnl@iecc.cambridge.ma.us, {spdcc|ima|world}!esegue!johnl
herrickd@iccgcc.decnet.ab.com (11/12/90)
In article <11043@pt.cs.cmu.edu>, jsp@ius3.ius.cs.cmu.edu (John Pieper) writes: > Is anybody out there old enough to remember when we used overlays? > What I want to know is what is the current state of the art in > compiler-managed overlays: is this considered a solved problem, or did people > just decide that since we had paging it was unimportant? > Boy, have you led a sheltered life. Intel / IBM / Microsoft created a memory limit of 640K for a very large universe of people who have money to spend on software. Overlay activity for that environment is real if not technically interesting. dan herrick herrickd@astro.pc.ab.com
davidsen@crdos1.crd.ge.COM (Wm E Davidsen Jr) (11/13/90)
In article <1939.273e786b@iccgcc.decnet.ab.com> herrickd@iccgcc.decnet.ab.com writes: | Boy, have you led a sheltered life. Intel / IBM / Microsoft created | a memory limit of 640K for a very large universe of people who have | money to spend on software. At the time the 8086 was designed, we were running VAXen with less than a MB in them, so the 1MB real memory limit didn't look too bad for a "toy computer" use. Early versions of MS-DOS (not PC-DOS) used all the memory there was, and I recall running at least 768k in a Tandy 2000. -- bill davidsen (davidsen@crdos1.crd.GE.COM -or- uunet!crdgw1!crdos1!davidsen) VMS is a text-only adventure game. If you win you can use unix.
dhesi%cirrusl@oliveb.ATC.olivetti.com (Rahul Dhesi) (11/13/90)
In <11043@pt.cs.cmu.edu> jsp@ius3.ius.cs.cmu.edu (John Pieper) writes: >Is anybody out there old enough to remember when we used overlays? Is anybody out there old enough to remember MS-DOS? -- Rahul Dhesi <dhesi%cirrusl@oliveb.ATC.olivetti.com> UUCP: oliveb!cirrusl!dhesi