[Coco] compile times with copt2 and rmxlab

[Willard Goosey]

OK, I've been playing with rmxlab and copt2 a little bit.

focal0.c is the driver for the focal interpeter.  It's 22K of source
doing various things.

Compiling this file with a shell script derived from CC1:
date -t
echo 'focal0.c'
-x
echo c.prep:
C_PREP -DOS9 -DDIR_SUPPORT -DOS9AF focal0.c >ctmp.3.m
x
echo ansifront:
ansifront < ctmp.3.m > ctmp.3.an
del ctmp.3.m
echo c.comp:
C.comp ctmp.3.an -o=ctmp.3.a
del ctmp.3.an
echo rmxlab:
rmxlab ctmp.3.a > ctmp.3.ar
del ctmp.3.a
echo c.opt:
c.opt ctmp.3.ar ctmp.3.o
del ctmp.3.ar
echo copt2:
copt2 /dd/sys/patterns.dat < ctmp.3.o > ctmp.3.o2
del ctmp.3.o
echo rma:
rma ctmp.3.o2 -o=focal0.r
del ctmp.3.o2
date -t

(No I don't normally use cc1.  I use CC252 normally.  Also there were no
speed tricks used here, no preloaded modules or ram disks or even a
non-fragmented hard drive.  The only special constraint during these
compiles is that no other processes were running.)

Compile time:
no rmxlab, no copt2: 140 sec
rmxlab, no copt2   : 150 sec
no rmxlab, copt2   : 306 sec
rmxlab, copt2      : 313 sec

Rmxlab is supposed to speed up the optimizers and assember by removing
unused labels from the assembly code.  It did trim 3 seconds off the
c.opt+copt2+rma train's runtime, but it took 10 seconds to run.  

Willard
-- 
Willard Goosey  goosey at sdc.org
Socorro, New Mexico, USA
I search my heart and find Cimmeria, land of Darkness and the Night.
  -- R.E. Howard