On Sat, Sep 19, 2015 at 12:30:13PM -0700, Al Kossow wrote: >When did the 4K user space(s?) actually swap? Did they round-robin or swap >based on activity? I would think they would stay in place until cpu-bound >jobs reached their time quantum. With only a couple of people on a 32k >machine, it may not even swap that much, depending on what the users were >running. I'd guess BASIC was pretty big.
My understanding is that it's round-robin among runnable jobs, one time slice at a time. I.e. the simplest possible way. IIRC the monitor always takes up two fields (not swappable). One more field (so, 12 KW total) is the minimum necessary to run at all -- SI, FIP, and all the users can share that field with frantic enough swapping (which causes a pretty lights show on the RF08 panel). Any more memory than that means less swapping (or none), so it's kicking out the LRU job as needed. I have a hazy memory that SI and FIP *only* run in field 2? Could be wrong. BASIC runs in your 4 KW with you. I've never seen its sources so I don't know how clever it is about overlays and/or keeping your program on disk. It's a very limited BASIC. Strings are 6 characters. Not max -- *always* 6. Line #s max out at 2046. John Wilson D Bit
