virtual memory 4500

Brian Inglis

...

Perhaps it went to the same place where people who are unwilling to accept actual data in studies whose results don't conform to their preconceptions banished it. Of course, cavalierly dismissing a study is easiest if one hasn't bothered to read it in the first place, isn't it?

I'm not sure how that follows from the paragraph which you purport to be responding to: not only had VMS been out in the field long enough by then to have contributed data in its own right, but there were other (non-RSX) sources as well.

Funny - they seem to have been large enough for VMS to use for over 14 years, until Alpha came along with 8 KB pages in 1992.

Of course, memory pages on RSX were 8 KB from the start, but there weren't enough in a 64 KB address space to make using VM very feasible (nor arguably would there have been even at 512 B granularity). Instead, we used explicitly-controlled remapping mechanisms which demand-paged at 8 KB granularity if the desired material wasn't already memory-resident in something at least akin to a system free list, and quite effectively.

fault

Not at all: when you organize your code appropriately it can do a lot in 8 KB, and RSX had reasonably efficient system calls. Of course we had disk overlay mechanisms as well (down to 512 B granularity, but extent-based such that as many disk sectors were read in at once as made sense for the task at hand - again, as structured explicitly by the programmer).

Not as simple for trained monkeys to program as demand paging, but often noticeably more efficient in end.

- bill