Date:	Wed, 5 Aug 92 11:24:20 +0100
From:	buers AT dg1 DOT chemie DOT uni-konstanz DOT de (Dieter Buerssner)
To:	rcharif AT math DOT utexas DOT edu
Cc:	djgpp AT sun DOT soe DOT clarkson DOT edu
Subject:	Paging enhancement for go32 v1.07

I tried the modified go32.exe from math.utexas.edu. With my configuration
there seems to a small speed-up. My configuration: 486DX-33,
4MB RAM, 1532 kB free XMS, 623 kB free conventional memory, ~1100 kB 
cache, DRDOS 6.0. Here some timings for compiling a medium-sized program
(200 kB source-code, plain C, maximal size paged out during compiling
about 700 kB).
    go32 from Go32Modified.zip: 3:02.52
    go32 from Go32Modified.zip recompiled with TC++ 1.0 and without
        TOPLINEINFO             4:25.35 
    dito with "TOPLINEINFO-patch" applied:
                                2:58.89
    go32 from djdev107.zip      5:40.11
    dito with "TOPLINEINFO-patch" applied:
                                3:08.77      

What I called "TOPLINEINFO-patch" is a small patch I sent some weeks
ago to the list. If anybody wants it, I can send it again.
The speed-up of Rami El Charif's patches is much better
when the memory requirements get bigger. (I didn't do any more timings)

I had two random problems with Rami El Charif's version, which I
can't explain. There happened one reboot during compilation and at
one stage gcc (or better cpp) wasn't able to find the include-files 
stdio.h etc. This was in the middle of a make, after some files
were already compiled successfully.
I checked the env. variables after this happened, and they
were still o.k. With the next make everything went fine again.
Of course, these problems could be due to some random failure of
my computer. 

Regards,
         Dieter

Dieter Buerssner  (B\"ur\ss{}ner) -- buers AT dg1 DOT chemie DOT uni-konstanz DOT de
Universitaet Konstanz -- Fakultaet Chemie -- Postfach 5560 
7750 Konstanz  -- Germany -- +49-7531-882021

- Raw text -