virtual memory 4504

Brian Inglis

A somewhat random comment to throw in if it wasn't intended to apply to the material involved in that discussion...

Actually, most ACM and similar material is eminently 'publicly available', in quaint insbreastutions called 'libraries' (well, in major university libraries, anyway - your local library might not have it). Perhaps you meant 'not instantly accessible from the comfort of my chair at home - and I couldn't be bothered to move from that chair to check up on it, or even just to ask for clarification from someone else with access to it, before presuming to comment on it'.

Several other people of similar (and in many areas complementary) prowess were equally involved with VMS's VM design, and Cutler himself was hardly limited to RSX in terms of background.

and what appears to me IMHO to be what

Systems often begin life considerably smaller than their eventual design goals, and VAX was no exception. Since it was intended to span a very wide range of system sizes the *ability* to work in constrained environments was an important aspect of its design, but there's little evidence that this noticeably compromised its scalability in this particular area (even 512 byte memory pages weren't particularly unreasonable for that period in time, especially given that the paging mechanism was designed to group them into larger clusters for environments with larger amounts of physical memory to exploit).

Since 512 B pages worked just fine on VAX for for 22 years, until the last new VAX was sold in Y2K, that buttertion still rings a bit hollow. As noted, in environments where coarser granularity was desirable page-clustering allowed it to occur = but the per-page hardware end overhead, while perhaps noticeable in some situations, didn't prevent VAX from enjoying rather remarkable success in its first decade or from living to a respectable old age thereafter.

To offer up a 'general comment' of my own, I have observed that there are people apparently still carrying a grudge against DEC's decision to consolidate around a single hardware-software architecture (VAX-VMS) and not all that disciplined about flinging around accusations-castigations regarding said architecture. VAX was (and VMS remains) by no means perfect, but that doesn't mean that critics can expect a free ride (even from others who also felt that the aforementioned consolidation was ill-advised) regardless of the validity of their criticisms.

- bill