[comp.sys.ti.explorer] Extended address space

davies@m2.csc.ti.com (Byron Davies) (02/12/91)

What's been your experience with extended address space?  Does it
really work?  What are its limitations?

	-- Byron

pf@islington-terrace.csc.ti.com (Paul Fuqua) (02/13/91)

    Date: Monday, February 11, 1991  7:10pm (CST)
    From: davies at m2.csc.ti.com (Byron Davies)
    Subject: Extended address space
    
    What's been your experience with extended address space?  Does it
    really work?  What are its limitations?

I think it's great.  I manage to run for weeks at a time, doing some
intensive simulations and other heavily-consing activities like using
CLIO and the X server.  After the first level-3 GC, about 75% of my
address space is free, and it rarely drops below 35%.

Limitations:  Faulting stuff in from external space is slow.  The point
of EAS is to use swap space larger than the address space, so without
extra disk space, it's ineffective (or so I understand);  I have about
185 MB swap space.  After turning it on, the first GC hogs the machine,
though subsequent GCs are hardly noticeable.  I don't think there's a
way to turn it off short of rebooting.

I'm running 6.1 with microcode 488, in which all the problems I used to
see (I've played with this since Release 4) are fixed.  I also have a
midnight GC that does a gc-immediately and a sys:gc-external every
night, which helps even when I'm not using EAS.

Paul Fuqua                     pf@csc.ti.com, ti-csl!pf
Texas Instruments Computer Science Center, Dallas, Texas

Gentiles are people who eat mayonnaise for no reason.  -- Robin Williams