🚀 go-pugleaf

glen herrmannsfeldt <gah@ugcs.caltech.edu> writes:
> My understanding is that when running under MVS it is better to
> run the compilers without overlays.  One can, for example, load
> the Fortran H compiler into the link editor, and write it out
> again without any overlay statements.  This will remove the overlay
> structure that it would otherwise have.

lets say you have 768k byte real machine ... with 104 4k pages
available, after fixed kernel requirements (400kbytes). If this is
being heavily time-shared with lots of other users ... and the
compilers have an aggregate code size of 1mbytes to 3mbytes .... then
it would be somewhat more efficient to do it in 100-200kbyte chunks
(and in real environment not possible to do in larger chunks that
available real storage).

If you have a 1gbyte (or more) real storage with 1mbyte to 3mbytes
aggregate compiler size, then it would be more efficient to read/fetch
it as one operation ... trading off real memory utilization against
disk arm utilization.

attached are several URLs to lengthy posts about observation that over
10-15 year period, the relative system disk/dasd performance had
declined by a factor of five to ten times. when i first started making
the statement, GPD (disk division) assigned several people from their
performance & modeling group to refute the claims. They eventually
came back with conclusion that I had slightly understated the problem.
This eventually culminated in GPD doing a user group presentation on
recommendations regarding disk allocation and optimization (i.e.
compensate for the declining relative system disk performance).

whether you are doing large block fetches in real memory system or in
virtual memory system (even when there has to be emulated real I/O
with page fixing/unfixing and ccw translation) ... it still tends to
be better than simplistic memory mapped implementation that does
most of the fetches based on 4k page fault at a time.

the memory mapped file system that i had done for cms (30+ years ago):
http://www.garlic.com/~lynn/subtopic.html#mmap

had API where CMS specified virtual address range to be mapped to
filesystem area. there was some dynamic adaptive stuff that looked at
contention for real storage and decided how best to fullfill the
request. the base CMS filesystem also didn't have the concept of
contiguous allocation ... pretty much a scatter allocation somewhat
inherited from CTSS (and not all that different from various dos,
unix, etc implementations). As part of the memory-mapped enhancements,
I also added semantics to the filesystem that attempted to get
contiguous records on disk and modified the application that generated
program executables to invoke the contiguous option.

observations about relative system disk performance was declining 
(aka disks were getting faster, but rest of the system was getting
much faster than disks):
http://www.garlic.com/~lynn/93.html#31 Big I/O or Kicking the Mainframe out the Door
http://www.garlic.com/~lynn/95.html#10 Virtual Memory (A return to the past?)
http://www.garlic.com/~lynn/98.html#46 The god old days(???)
http://www.garlic.com/~lynn/99.html#4 IBM S/360
http://www.garlic.com/~lynn/99.html#103 IBM 9020 computers used by FAA (was Re: EPO stories (was: HELP IT'S HOT!!!!!))
http://www.garlic.com/~lynn/2001f.html#62 any 70's era supercomputers that ran as slow as today's supercomputers?
http://www.garlic.com/~lynn/2001l.html#40 MVS History (all parts)
http://www.garlic.com/~lynn/2001l.html#61 MVS History (all parts)
http://www.garlic.com/~lynn/2001m.html#23 Smallest Storage Capacity Hard Disk?
http://www.garlic.com/~lynn/2002.html#5 index searching
http://www.garlic.com/~lynn/2002b.html#11 Microcode? (& index searching)
http://www.garlic.com/~lynn/2002b.html#20 index searching
http://www.garlic.com/~lynn/2002e.html#8 What are some impressive page rates?
http://www.garlic.com/~lynn/2002e.html#9 What are some impressive page rates?
http://www.garlic.com/~lynn/2002i.html#16 AS/400 and MVS - clarification please
http://www.garlic.com/~lynn/2003.html#21 vax6k.openecs.org rebirth
http://www.garlic.com/~lynn/2003d.html#21 PDP10 and RISC
http://www.garlic.com/~lynn/2003f.html#50 Alpha performance, why?
http://www.garlic.com/~lynn/2003m.html#42 S/360 undocumented instructions?

-- 
Anne & Lynn Wheeler | http://www.garlic.com/~lynn/