Stan Sieler <[log in to unmask]> wrote:
|Jeff Kell asks:
|> My point (finally) is why 24Mb memory is insufficient.  We all know that it
|> isn't enough for MPE/iX, but why not?
|
|This isn't *why*, but provides a first-level reason:
|
|Here's the output of running PAGES (formerly a Nugget, now part of the
|System Manager's Toolbox from Lund Performance Solutions), showing how
|memory is currently used on our 917:
|
|   PAGES [2.22] - LPS Toolbox [A.01a]     (c) 1995 Lund Performance Solutions
|
|   For Help at the PAGES prompt enter   ?
|
|   SERIES 917LX
|   MPE/iX 5.0 (or later)
|   #CPUS: 1
|   Memory size: 88 MB (92,274,688 bytes; 22,528 logical pages)
|
|   PAGES: s
|   Memory size: 22,528 logical pages (88 MB)
|         9,310 Dirty,   15,866 Referenced,    5,817 Frozen.    1,682 Unused.
|     (   4,927 Dirty,   10,314 Referenced of the unfrozen pages)
|
|      State of in-use pages:  20,196 Present, 0 IMI, 579 ROC, 70 Absent.
|
|   Memory usage by "type" of Object Class:
|
|          Class     #LogicalPages  #MB  % total
|      ------------- -------------  ---  -------
|      SYSTEM_CODE           4,167   16    18.5%
|      SYSTEM_DATA           8,793   34    39.0%
|      UNUSED                1,682    6     7.5%
|      USER_CODE             2,332    9    10.4%
|      USER_DATA               979    3     4.3%
|      USER_STACK            1,836    7     8.1%
|      USER_FILE             2,738   10    12.2%
|
|      Totals:              22,527   87   100.0%
|
|      "User" pages are  42.5% of memory (38 MB out of 88 MB)
|
|The SYSTEM_CODE and SYSTEM_DATA categories, combined, are using 50 MB
|on my machine.  On a smaller machine, these numbers drop slightly,
|but still add up to a *LOT* of memory.
|
|Does anyone with a small member machine have PAGES, and (if so), can they
|share the output of the "S" command with us?
 
One of our systems with 64MB (not long ago only 32MB) shows:
|PAGES [2.14] - NUGGETS [K.2805.06]                  TYMLABS CORPORATION [G0700]
|(c) 1989 Software Research Northwest, Inc.                           FM (917LX)
|
|For Help at the PAGES prompt enter   ?
|
|SERIES 917LX
|MPE/iX 5.0 (or later)
|#CPUS: 1
|Memory size: 64 MB (67,108,864 bytes; 16,384 logical pages)
|
|PAGES: s
|Memory size: 16,384 logical pages (64 MB)
|      5,347 Dirty,   13,033 Referenced,    4,543 Frozen.      686 Unused.
|  (   2,266 Dirty,    8,982 Referenced of the unfrozen pages)
|
|   State of in-use pages:  14,648 Present, 0 IMI, 979 ROC, 70 Absent.
|
|Memory usage by "type" of Object Class:
|
|       Class     #LogicalPages  #MB  % total
|   ------------- -------------  ---  -------
|   SYSTEM_CODE           3,614   14    22.1%
|   SYSTEM_DATA           6,274   24    38.3%
|   UNUSED                  686    2     4.2%
|   TURBO_DATA               40    0     0.2%
|   USER_CODE             2,326    9    14.2%
|   USER_DATA               276    1     1.7%
|   USER_STACK              856    3     5.2%
|   USER_FILE             2,311    9    14.1%
|
|   Totals:              16,383   63   100.0%
|
|   "User" pages are  39.6% of memory (26 MB out of 64 MB)
 
That looks like 38 of 64MB in use by MPE to me.
 
By the way, although memory pressure is not too big a problem with 64MB on this
machine, it was almost non-existant with 96MB (which we had while waiting on
delivery of the second 32MB now in use).  However, with only 32Mb the system
was really slow with even two users active on the machine.  A single compile
would saturate the system.  This system is a development and test machine which
normally has only 2 or 3 active users online.
--
Jeff Woods <[log in to unmask]> [PGP key available]
Women and cats will do as they please. And men and dogs should relax
and get used to the idea. -- Robert A. Heinlein (1907 -1988)