[comp.sys.sgi] X Window System performance on Iris

jgarb@CSD360B.ERIM.ORG (Joe Garbarino) (05/09/89)

Mark Callow of SGI writes:
> On present SGI hardware XOR performance is at best adequate.  We therefore
> recommend you run our window manager which does rubberbanding in the overlay
> planes (it's written in PostScript, but that's another story) rather
> than something like uwm which uses xor all over the place.

I replied:
@ Until this XOR/window manager problem is solved, a major piece of SGI's X
@ Window System implementation is severely inadequate, and I will not be
@ satisfied with its performance.

Mark again:
> I said the performance was "adequate".  How you translate that to "severely
> inadequate" without even having seen it is beyond me.  uwm is usable.

"at best adequate" says to me that while it may be adequate in some
respects, it is mostly inadequate.  If you can follow the drawing of the
outlines of the windows as they are being moved/resized with
awm/twm/uwm, as you can in the current implementation, this is indeed
severely inadequate.  Movement/resizing of the windows is one of the
more important functions of the window manager; making it a very slow
function actually makes that window manager unuseable.

					Joe Garbarino
					ERIM
					P.O. Box 8618
					Ann Arbor, Mi.  48107
					(313)994-1200 x2508
					jgarb@csd360b.erim.org

baskett@SGI.COM (05/12/89)

"Movement/resizing of the windows is one of the more important
functions of the window manager;"

True.

"making it a very slow function actually makes that window manager
unuseable."

Perhaps for you.  For most people I know, the window manager needs to
process input events quickly and efficiently, not use too much memory,
not use too many CPU cycles, and do menus and such tools quickly.  Most
people I know don't move or resize windows very many times per minute/
hour/day so the speed of those operations is not especially critical as
long as it is fast enough to not be irritating.  What is fast enough to
not be irritating is a personal judgement, too.  But Mark is right in
that you should try it before you make that judgement.

Forest Baskett