OUTMEM error problem

[Loki Jorgenson Rm421]

	Hey ho.....

	I have been suffering with symptoms which I can't trace to an
algorithmic error.  I am wondering if it isn't a system bug of some
strange sort.

	On a 4D/25 running IRIXv3.2.1, I have been running a simple program
which reads in a 2D array of values, +1/-1, opens a window and draws the
rectangular array by coloring a pixel if +1 and leaving it black if -1.
Strangely enough, I have been encountering a problem wherein the program
crashes, reporting a graphics error #2 OUTMEM (and a couple of #75
NOWINOPEN after that) and leaving a core.  Dbx shows the reason for the
crash to be a segmentation fault.

	This may seem obvious to many but the hitch is this: This only 
happens on certain size arrays, typically my smaller ones!  For example,
72x72 sized systems.  For larger systems, like 192x192, I have no problems.

	Dbx has narrowed the offending subrountine down to my call "ortho2()"
and eventually to the internal graphics call "i_ortho2()".  Repeated
examinations of the core, the code and the results have yielded very little
consistent info upon which to define the source of the problem.  I have
used edge to trace the program and check variable values but nothing has
appeared to be amiss.

	Note that there are a couple of malloc()'s in my program but I
have no reason to suspect that there is actually a shortage of
memory; the machine has 16M and the cores are actually quite
small.... about 30k.

	Has anyone had any similar experiences?

Thanks,

Loki Jorgenson			node:  loki at physics.mcgill.ca
Physics, McGill University	fax:   (514) 398-3733
Montreal Quebec CANADA		phone: (514) 398-6531

		<<  If I only had a brain  >>