Answers Database
Workstation XDE: Whatever happened to get_memsize.csh
Record #211
Problem Title:
Workstation XDE: Whatever happened to get_memsize.csh
Problem Description:
Solution 1:
The reason we used get_memsize.csh was because on certain computers
(normally Apollo) it was common for the machine to have only 8 Mbytes
of RAM. The workstation XACT takes about 15 Mbytes to draw a 3090
at its default scale, so that made the X server start swapping,
effectively killing the machine. Even machines with 16 or 24 Mbytes
could be swamped.
What was really wanted was the amount of RAM on the machine; but that's
almost impossible to determine on most Unix systems.
The only platform where get_memsize.csh would return a reliably useful number
was Apollo.
We modified XACT to default to 14 Mbytes unless that value is overridden
by the user using XACT_MEMSIZE, and stopped distributing get_memsize.csh.
A good rule of thumb is that 14 Mbytes is fine; if your system has more
than 32 Mbytes, then the XACT_MEMSIZE environment variable may be set
to as much as 1/2 of the total. X servers generally are pretty vulnerable
to running out of memory, so a conservative approach is wise.
End of Record #211 - Last Modified: 11/06/96 03:23 |
