X11R4 for i386 Unix
Piercarlo Grandi
pcg at cs.aber.ac.uk
Fri Jul 6 01:06:13 AEST 1990
In article <1990Jul03.222233.1896 at virtech.uucp> cpcahil at virtech.uucp (Conor P. Cahill) writes:
The difference is "requires to work" as opposed to "requires for usefulness".
X11R3 on any 386 system with less than 8MB will force the different processes
to be swaped out. I ran one in this configuration with only 3 xterms, an
xbiff, and an xclock. You could hear the disk fumbling around when I moved
the mouse from window to window.
This si again the result of poor swapper/page design. I cannot believe
that the combined working sets of your processes was larger than 2-3
megabytes. Hey, I cannot believe that the *total* size of your
applications was larger than that, even if xclock is rumoured to grow to
1.3 megs of address space...
On my sun3 I have 2 xterms and 1 xclock, and two remote windows, and
sizes are:
SIZE RSS
X 952 248
xclock 152 0
xinit 56 0
xterm 208 96
tcsh 104 80
uwm 120 32
xterm 216 112
tcsh 104 104
TOTALS 1912 672
i.e. well below two megs and a tad over half, and even given the great
imprecision of such measurements, we are way below eight megs. The only
explanation as usual is the catastrophic performance of the System V
(and, incidentally SunOS, Ultirx, etc...) swapper (yes, on my 4 meg sun3
I get some swapping as well, as you see form the above 0 RSS
entries...).
Note that my X server is that big also because I have a large background
bitmap.
--
Piercarlo "Peter" Grandi | ARPA: pcg%cs.aber.ac.uk at nsfnet-relay.ac.uk
Dept of CS, UCW Aberystwyth | UUCP: ...!mcsun!ukc!aber-cs!pcg
Penglais, Aberystwyth SY23 3BZ, UK | INET: pcg at cs.aber.ac.uk
More information about the Comp.unix.i386
mailing list