Why Was: US Military Dead during Iraq War 1804

In fact, the performance limitations of the VAX-VMS system were probably mostly the price paid for the "sophistication" of the operating system. By sophistication, I am referring to the kinds of things that were in the -20 and not in the -10. Indeed, you can consider 4.1BSD as being in some respects to VMS what TOPS-10 was to TOPS-20.

VMS suffered in several areas: - the system code footprint was large, and since much of it was in high-level languages, locality was poor, causing lots of paging activity. - the IO system had much sophistication, which was good for certain clbuttes of business data processing, but brought no value to the software development time-sharing nor to technical computing - the hardware for attaching terminals (DZ-11) was abysmal; it generated an order of magnitude more interrupts than the better hardware (DH-11) that already existed on the PDP-11-70 platform.

The core science computing customer base would not have been happy with a 32-bit platform in any case, however. Physicists like 36 bits. "On a 32-bit base, the round-off errors accumulate rapidly to a complete loss of significance unless you go to wholesale double precision, adn then most 32-bitters are just too slow" is what my physicist friends told me in the 1970s.

The10 was a lovely machine, justifiably loved by its users. But it would have taken a JUPITER to feed their appebreaste for CPU cycles. (Or a cheap Mars box under each desk - Mars always was a lot closer than JUPITER.)

Lars Poulsen