[18] in 1993-clients
first impressions
daemon@ATHENA.MIT.EDU (John Carr)
Mon Mar 9 19:08:28 1992
To: 1993-clients@Athena.MIT.EDU
Date: Mon, 09 Mar 92 19:08:13 EST
From: John Carr <jfc@Athena.MIT.EDU>
I logged into both test machines tonight.
There were three problems I found on the RS/60000, all of which are
probably related to the early stage of the AIX 3.2 port:
1. When I typed an incorrect username xlogin exited too quickly
for me to read an error message (there might have been an X
error in the console window).
2. aklog printed too many warning messages
3. IBM mh was installed, so my path found /bin/<mh program> before
/usr/athena/bin/<mh program> and I couldn't read mail without
using an explicit path. The fix is not to install IBM mh.
A footnote to (3): like Sun OS, AIX 3.2 makes /bin a symbolic link
pointing to /usr/bin so $athena_path will need to be changed to reference
only one of these directories.
I did not try performance tests on the RS/6000 because it isn't running
AFS yet and the large packages I would have tried to compile are in AFS.
Both versions of X had minor bugs (stray pixels in italic fonts on the
IBM, blank regions in the DEC xterm while scrolling).
The DECstation had a screensaver running when I found it. It took a long
time to reboot it (when we typed the same key sequence the machine ignored
me but rebooted for Lucien), and the kernel took a long time to come up.
Compared to my (very low) expectations the DEC workstation was very good.
It is the first DEC workstation I have used with a usable keyboard or
reasonable disk I/O (the lk402 and rz25 are responsible for this). Ultrix
doesn't have integrated VM and filesystem code like AIX, but the fast disk
makes up a much of the difference. With the fast disk, sufficient memory,
and new keyboard it is possible for me to get work done on a DECstation.
Jeff suggested that the DECstation might make a good server. Based on my
I/O tests I agree. More tests are required to verify this, including
trying to put several disks onto one machine.
