Murray Lesser wrote to Will Honea on 12-07-1999

ML>      Couple of quick queries, and one not so quick:
ML> 
ML>     Some Warp 3 FixPak around 10 killed OS20MEMU.  It was revived for
ML> Warp 4 with a patch I got from Hobbes based on advice from
ML> devcon@ibm.net at the time (name of patch not recalled, but the
ML> file OS20MEMU.DLL in my OS2MEMU folder is dated 9/12/96; all the
ML> other files are dated in '94).  I am now running it under Warp 4,
ML> FixPak 5.  Are you saying that os20memu died again on a later Warp 4
ML> FixPak?  If so, another good reason to not "upgrade" if I don't have
ML> to :-).  If its only death to date was in Warp 3, I have no worry
ML> since both of my machines are running on Warp 4 FixPak 5. 

I pretty much got away from OS/2memu so I'm just reporting what I
remembered seeing - I haven't had it up in over a year.


The version is Thesues3, last copy I saw at Testcase was Th?/3000.zip
- same content as originally offersd.  The DLL is theseus5.sys

ML>     I am using Theseus 3 (theseus5.sys in config.sys).  Do you
ML> really mean you are using a later version, Theseus 5, or was that a
ML> slip of the fingers?  If you meant it, where did you get it? 

Same animal.  Works well for me.  AIR, you rename the DLL in
distribution to Theseus0.DLL and that's the only Theseusx.DLL I've ever
noticed beung loaded.

ML>     I am also still running OSRM2 Lite, which uses theseus2.sys as
ML> a driver to run a working set analysis.  The output doesn't agree
ML> with that given by Theseus 3; it appears that OSRM2 doesn't show the
ML> shared system memory.  I think I will give up the disk space taken
ML> up by OSRM2 since the vendor has apparently vanished and I now have
ML> Theseus 3.  The results shown by OSRM2, OS20MEMU, and THESEUS 3, for
ML> "owned memory" are all different, and I haven't the slightest idea
ML> of which one (if any) is "correct" :-(.  In general, I use OS20MEMU
ML> only for quick checks on the effects of coding changes (mostly for
ML> multithreaded programs) on memory usage.  Any comment?

The only place I ever heard of OSRM2 was when you mentioned it before
so I have zero idea how it works.  I trust Theseus over OS2Memu for
reporting, altho the latter was a bit quicker to interpret.  I use
Theseus all the time to monitor the programs I'm working on for memory
leaks and occasionally for hot spot analysis in some LAN based comm
threads.  Theseus is considerably more detailed in it's memory analysis
but it's also harder to learn to interpret.  Off the top of my head, I
don't recall seeing a break out for cache usage in Theseus - but
there's enough there to make missing something easy.

Will Honea 
--- Maximus/2 2.02