earlw@pesnta.UUCP (10/15/83)
they may in fact do that. Any data they may receive
doesn't filter down to my level and the users of the network would
never see the reports anyway. By using the network to gain my
information, I have accepted the obligation to distribute any data
to the net users.
Since the responses are short, I will list each one but
information about the originator of the article has been removed.
All responses are listed in this report - good and bad. The format
was cleaned up a bit to make it easier to read. The content is the
same. I wish to thank the people that responded to the survey
request. A copy of this report will be given to P-E management.
User #1 (Contains general comments and foul language)
===================================================================
Worst system (OS/32, not the computer) I ever used. No good for
realtime, no good for software development. The machine could have
supported a fine realtime system, if PE didn't have shit-for-brains
software people. Software overhead on SVC's used to be miserable!
The machine architecture was OK, but I disliked the software
simulate interrupt instruction, which was unable to hold and
remember interrupts at lower levels than they were executed (ie.
the instruction forced an immediate interrupt regardless of the
current priority) and the queue manipulation instructions only
worked for addresses below 2^16 (too bad!). Also really needed an
MVC! Maybe they've fixed these things in the past 5 years. Looks
like what started as an attempt to make IBM assembly source
compatible that got lost.
User #2
===================================================================
Model: 3210
Options: Floating Point
Memory: 1 Megabyte
Disks: MSM300 on Selch F1
Ampex 96 on Selch F0
Tapes: 800 bpi on Selch F0
Terminals: 13
(also two Teletype 40s, one NEC 5510 and one Hayes
Smartmodem 1200)
Operating System: UNIX
Users: 10
Type of work: Transaction processing - database update
Reliability: (Crashes, etc.)
Always hardware so far. Mostly things get scrambled on
power fails. Sure would be nice if Edition 7 had the
fancy power-fail auto-restart that 4.1 on VAXen has.
P-E Customer Service response time, analysts/CE's ability to do the
job, ...
Usually we know more about UNIX than the local support
people, but they call O'port and get the answers.
Does the soft/hardware live up to your expectations? (handles the
user load, stays up, ...)
All in all, I've been pleased. The machine has proven
very reliable, the hardware support is very good. I'm
filling up disk faster than I counted on, but who
doesn't? I should've bought a faster tape drive. The
only thing I'm not real wild about is the 1251 CRTs.
They don't have keyboard rollover, and the special
attributes are brain-damaged (i.e. can't scroll). For
the money, I would buy VT-100 look-alikes if I had it
to do over again.
What additions/deletions/enhancements would you want P-E to make?
a) Provide a driver and spooler for the NEC spinwriter
b) Power-fail auto restart
c) Get System 3
d) get news (how come you've got B 2.10.1.1 when I had
to bootleg an old A+ ??)
e) Add the -O option (optimizer) on the C compiler
f) tell me how to get files onto the removable 13 Meg
platter on my Ampex drive. I haven't been able to get
anyone to tell me how to do it successfully yet.
User #3
===================================================================
Model: 3240
Options: Floating Point
Memory: 3 Megabyte (soon to be upgraded to 6Mb)
Disks: 2 300Mb on Selch F0
1 300Mb on Selch F1
Tapes: Kennedy 800/1600 125ips on Selch F2
Terminals: 9 (including console)
Other Peripherals: Versatec V80 serves as our printer and graphics
hardcopy device. CSPI Map300 array processor.
Operating System: Unix Edition VII release 2.2.1.1 (from twg)
Load Average: 1.0 (heavily underused at present during design phase
of projects for which the system was purchased [see
below].)
Work Type: Development of Spatial Database Management System (i.e.
currently most of the load has been compiles and runs
of database code etc.) [Also, a lot of games have
immigrated to the P.E. because of its superior speed
(compared with PDP11/34 and PDP11/70) and greater
address space (warp 6.0 required modification to our
200+Kb kernel).]
Hard/Software/Service/Sales/Company Comments
Reliability: Most crashes/panics have been easily traceable to
either hardware or software. Most of the software
problems have been as a result of our modifications to
the kernel, some of which didn't quite work at first.
We have had 1 5V power supply (out of 4?) give up, one
head crash, and one mysterious power-supply problem
that turned out to be a loose connection in the
voltage-sensing circuits.
PE response time: Response time has been great (a couple hours)
since PE headquarters for the province (maybe
Canada) is just 1/2 Hr away. The service
personnel have been, for the most part,
competent. Some-times, the people come in pairs,
one of the people being a relatively
unexperienced trainee. On the whole, the machine
has been very reliable. Response from PE has
been good when it comes to hardware; when it
comes to software complaints and suggestions,
they almost seem to fall on deaf ears. We have
made two very simple suggestions as to how to
improve something in the kernel (told them what
file to fix and where) and the only response was
the acknowledgement copy of our report was
returned.
Additions/Changes/Enhancements:
The speed selection for the 2- and 8-line communications
multiplexors sucks. While there are 16 possible speeds for
each line, they come in 4 groups of 4 speeds. In order to
change which speed group a particular line uses, you have
to change 1 or 2 hardware switches on the board which means
taking the system down. All speeds should be selectable
via software.@pesnta.UUCP (10/15/83)
they may in fact do that. Any data they may receive
doesn't filter down to my level and the users of the network would
never see the reports anyway. By using the network to gain my
information, I have accepted the obligation to distribute any data
to the net users.
Since the responses are short, I will list each one but
information about the originator of the article has been removed.
All responses are listed in this report - good and bad. The format
was cleaned up a bit to make it easier to read. The content is the
same. I wish to thank the people that responded to the survey
request. A copy of this report will be given to P-E management.
User #1 (Contains general comments and foul language)
===================================================================
Worst system (OS/32, not the computer) I ever used. No good for
realtime, no good for software development. The machine could have
supported a fine realtime system, if PE didn't have shit-for-brains
software people. Software overhead on SVC's used to be miserable!
The machine architecture was OK, but I disliked the software
simulate interrupt instruction, which was unable to hold and
remember interrupts at lower levels than they were executed (ie.
the instruction forced an immediate interrupt regardless of the
current priority) and the queue manipulation instructions only
worked for addresses below 2^16 (too bad!). Also really needed an
MVC! Maybe they've fixed these things in the past 5 years. Looks
like what started as an attempt to make IBM assembly source
compatible that got lost.
User #2
===================================================================
Model: 3210
Options: Floating Point
Memory: 1 Megabyte
Disks: MSM300 on Selch F1
Ampex 96 on Selch F0
Tapes: 800 bpi on Selch F0
Terminals: 13
(also two Teletype 40s, one NEC 5510 and one Hayes
Smartmodem 1200)
Operating System: UNIX
Users: 10
Type of work: Transaction processing - database update
Reliability: (Crashes, etc.)
Always hardware so far. Mostly things get scrambled on
power fails. Sure would be nice if Edition 7 had the
fancy power-fail auto-restart that 4.1 on VAXen has.
P-E Customer Service response time, analysts/CE's ability to do the
job, ...
Usually we know more about UNIX than the local support
people, but they call O'port and get the answers.
Does the soft/hardware live up to your expectations? (handles the
user load, stays up, ...)
All in all, I've been pleased. The machine has proven
very reliable, the hardware support is very good. I'm
filling up disk faster than I counted on, but who
doesn't? I should've bought a faster tape drive. The
only thing I'm not real wild about is the 1251 CRTs.
They don't have keyboard rollover, and the special
attributes are brain-damaged (i.e. can't scroll). For
the money, I would buy VT-100 look-alikes if I had it
to do over again.
What additions/deletions/enhancements would you want P-E to make?
a) Provide a driver and spooler for the NEC spinwriter
b) Power-fail auto restart
c) Get System 3
d) get news (how come you've got B 2.10.1.1 when I had
to bootleg an old A+ ??)
e) Add the -O option (optimizer) on the C compiler
f) tell me how to get files onto the removable 13 Meg
platter on my Ampex drive. I haven't been able to get
anyone to tell me how to do it successfully yet.
User #3
===================================================================
Model: 3240
Options: Floating Point
Memory: 3 Megabyte (soon to be upgraded to 6Mb)
Disks: 2 300Mb on Selch F0
1 300Mb on Selch F1
Tapes: Kennedy 800/1600 125ips on Selch F2
Terminals: 9 (including console)
Other Peripherals: Versatec V80 serves as our printer and graphics
hardcopy device. CSPI Map300 array processor.
Operating System: Unix Edition VII release 2.2.1.1 (from twg)
Load Average: 1.0 (heavily underused at present during design phase
of projects for which the system was purchased [see
below].)
Work Type: Development of Spatial Database Management System (i.e.
currently most of the load has been compiles and runs
of database code etc.) [Also, a lot of games have
immigrated to the P.E. because of its superior speed
(compared with PDP11/34 and PDP11/70) and greater
address space (warp 6.0 required modification to our
200+Kb kernel).]
Hard/Software/Service/Sales/Company Comments
Reliability: Most crashes/panics have been easily traceable to
either hardware or software. Most of the software
problems have been as a result of our modifications to
the kernel, some of which didn't quite work at first.
We have had 1 5V power supply (out of 4?) give up, one
head crash, and one mysterious power-supply problem
that turned out to be a loose connection in the
voltage-sensing circuits.
PE response time: Response time has been great (a couple hours)
since PE headquarters for the province (maybe
Canada) is just 1/2 Hr away. The service
personnel have been, for the most part,
competent. Some-times, the people come in pairs,
one of the people being a relatively
unexperienced trainee. On the whole, the machine
has been very reliable. Response from PE has
been good when it comes to hardware; when it
comes to software complaints and suggestions,
they almost seem to fall on deaf ears. We have
made two very simple suggestions as to how to
improve something in the kernel (told them what
file to fix and where) and the only response was
the acknowledgement copy of our report was
returned.
Additions/Changes/Enhancements:
The speed selection for the 2- and 8-line communications
multiplexors sucks. While there are 16 possible speeds for
each line, they come in 4 groups of 4 speeds. In order to
change which speed group a particular line uses, you have
to change 1 or 2 hardware switches on the board which means
taking the system down. All speeds should be selectable
via software.