|
| |
Gathering Point for Physics Service Courses Status Report - 4th Quarter
 | Accomplishments |
| Next Quarter Plans |
| Contacts |
| Equipment Utilization |
| Feedback and Problems
Accomplishments There are four courses currently making substantial use
of the computing facility.
 | Materials Simulations Jim Sethna's course, developed explicitly to
promote and utilize the new Intel machines, is being taught again this spring. We again
have over twenty participants, including one faculty member, ten of the brightest young
physicists, and ten other graduate students from engineering. This spring, the focus is on
materials simulations. We meet for six hours per week in two afternoon lab sessions,
implementing state-of-the art simulations techniques spanning at least four topics.
|
| Peter Stein is teaching a non-majors undergraduate course using Python. |
| Physics 214, the large third-semester engineering course on waves, particles, and
fields, is using the facility for two weeks solidly in introducing their class to
interactive simulations, using the simulations galileo, pythag, huygens, and schrdgr. |
| Karl Berkelman is using the facility regularly in his honors Waves course, Physics 218. |
Next Quarter Plans
We intend to put up an X-server to run some of the important plotting programs not
available under Windows.
Contacts
James P. Sethna, Professor, Department of Physics
Ralph B. Robinson, Programmer Analyst, Physics Department
Equipment Utilization
All machines are active and running over Fast Ethernet using Windows NT. The final
software configuration has been installed, and is all we could wish for.
Feedback
Developer's Studio C++ is an excellent software development environment. It is the sole
rationale we have found for shifting from Unix/Linux to the Windows platform. We had an
excellent experience with it in our course and are continuing to enjoy working with it in
research contexts.
Last modified on: 07/07/98
|
|
Gathering Point for Physics Service Courses Status Report - 2nd Quarter
| Accomplishments |
| Next Quarter Plans |
| Contacts |
| Equipment Utilization |
| Feedback and Problems
Accomplishments
In the first part of this second quarter, Jim Sethna continued our
new course course, Physics 683 designed to show off the fast hardware
and flexible software development environment made possible by the Intel
grant. Between ten and twenty of the brightest and computationally
most sophisticated graduate students and post-docs in the department met
for six hours per week in two afternoon lab sessions, implementing
state-of-the art simulations techniques spanning eight topics.
The summer period is a quiet time for this teaching facility. The
equipment is being used by a variety of graduate students doing various
projects. One faculty member and four of his students used the room to
engage in a virtual conference on dislocation dynamics being held at
Livermore National Labs, using RealVideo technology: see http://multiscale.llnl.gov/multiscale-folder/course/ecourse.html.
Serious simulations on multiple processers are exploring dendritic
growth. Finally, the software component for a pedagogical article on
hysteresis and avalanche behavior - an outgrowth of Physics 683 (above)
- is being developed and tested.
Next Quarter Plans
In the next quarter, we intend to begin the integration of this
facility with it's intended audience: the freshman and sophomore
engineering and biology students taking introductory physics.
The Web interfaces for all physics courses will be centralized into the
new facility. The FTP site for downloads of course software will be
shifted. A 100 megabit connection from the facility to the
backbone will be installed, if Cornell Information Technologies provides
it. Distribution of a finalized workstation configuration will
proceed next month before classes begin. One of the machines will
be incorporated into the classroom demonstrations. Several
demonstrations are planned in the third semester Waves course for the
four labs galileo, pythag, huygens, and schrdgr.
One or perhaps two of the cooperative learning sessions in the
first-semester engineering mechanics course will be taught in the
facility, using galileo and perhaps jupiter. This is a large course, and
each learning session will fill the facility constantly for a couple of
weeks. Explorations of downloads of multimedia software into the
dorm rooms will begin.
Contacts
James P. Sethna, Professor, Department of Physics
Ralph B. Robinson, Programmer Analyst, Physics Department
Equipment Utilization
Fast Ethernet has been installed. All machines are active and running
Windows NT. A final software configuration is in development, and should
be installed soon. We had some difficulty cloning machines with Drive
Image Professional, and are shifting to Ghost, which has been reasonably
successful in LASSP.
Feedback
Developer's Studio C++ is an excellent software development
environment. It is the sole rationale we have found for shifting from
Unix/Linux to the Windows platform. We had an excellent experience with
it in our course and are continuing to enjoy working with it in research
contexts.
Setting up large numbers of identical NT workstations is much too
painful. We're pretty sophisticated, but we've worked long and hard to
find a way to clone one software installation over several machines.
The software we chose to clone the systems with, Drive Image
Professional, was unable to strip the machine-dependent information from
the various registry and other files. We're planning to try Ghost.
LAST QUARTER's FEEDBACK is largely still applicable:
The SCSI systems on the Dell workstation machines are configured
improperly! Unfortunately, Dell insists on providing SCSI CD-ROM drives
in systems with SCSI hard disks. This causes problems if one desires to
connect SCSI devices to the external SCSI connector on the systems. We
eventually decided not to use a Jazz drive to clone the system, but to
do it over the network.
The Dell software distribution does not properly share files over the
network: we needed to patch the registry to fix the Server service. They
also distribute their software on a FAT partition, which wastes a huge
amount of space and demands extra steps to re-establish file security
even after converting to NTFS.
We have yet to find a sensible XY plotting package for the machines. In
particular, we want a package which is commonly available (free or
mainstream, preferably both), and which can be driven from within a C++
program from the command line, and from a scripting language such as
Python. We explored Excel: in the current release of Python we
could not make the COM interface to work at all, and we had no clue how
to use it from the command line or from within a C++ program.
(Presumably OLE allows us to do this.) Gnu plotting also
didn't work well. We continue to explore: we have hopes for an NT port
of xmgr, and are looking into Genplot and Super Mongo.
We have had problems with strange erratic flickering in the Dell 1200HS
19" monitors (purchased as cost-sharing for the Intel grant). The
flickering looks quite similar to, but milder than, that produced when
degaussing the monitor. The problem is exhibited in two separate rooms,
but is not local to an individual monitor: several nearby monitors will
flicker in synchrony. We are exploring whether the problem is with stray
magnetic fields or is transmitted through the power supply. We do not
know at this time whether Dell's monitor is particularly susceptible, or
whether other brands are better shielded.
Last modified on: 07/07/98
|
Gathering Point for Physics Service Courses Status Report - 1st Quarter
Shifting the department to a new platform is a substantial task. It was
crucial to the research efforts, to the computing administration efforts,
and to the teaching effort that we devote substantial time and effort to
making immediate use of these machines, in demanding tasks, to exercise the
hardware and software and introduce ourselves to the new environment.
To do this, Jim Sethna
in Physics developed a new course, Physics
683 designed to show off the fast hardware and flexible software
development environment made possible by the Intel grant. Between ten and
twenty of the brightest and computationally most sophisticated graduate
students and post-docs in the department met for six hours per week in two
afternoon lab sessions, implementing state-of-the art simulations techniques
spanning six topics:
As the earliest of the facilities in Physics to get its equipment, the
Physics Service Course room has seen much activity with broad impact for
physics adoption of the Intel architecture in general. We expect that this
activity will spill over into the physics service courses: the core of
talented graduate students will bring to their teaching the skills developed
in this semester, and of course we've worked out many bugs in the system.
The next quarter is summer break. We expect the graduate student use of
this facility to drop precipitously as the LASSP and LNS equipment arrives
and as the course ends. Planning and software development for the fall
semester, the installation of fast Ethernet and the distribution of a
finalized workstation configuration will dominate the next quarter usage.
James P. Sethna,
Professor, Department of Physics
Ralph B. Robinson, Programmer
Analyst, Physics Department
Two days after the client machines were delivered, a group of physics
graduate students and post-docs assisted in unpacking the systems,
installing the operating system and the software, and setting up the
network: within 24 hours all seventeen machines were active and usable
running Windows NT Workstation. (One machine is on loan to the Intel project
centered in LASSP, being used by a new faculty
member). The server did not arrive until a month later, so we started by
setting up the students with accounts on an individual node, and by
distributing software via shared directories over the network (after
patching the registry to fix the Server
service in the Dell software distribution.)
Getting the server configured properly and getting a workable
configuration for the workstations took a few weeks after the arrival of the
server. We then ran into two serious obstacles to cloning the software
configuration from one workstation to the other 18 (see Feedback).
| The SCSI systems on the Dell workstation machines are configured
improperly! Unfortunately, Dell insists on providing SCSI CD-ROM drives
in systems with SCSI hard disks. This causes problems
if one desires to connect SCSI devices to the external SCSI connector on
the systems. We eventually decided not to use a Jazz drive to clone the
system, but to do it over the network. |
| The Dell software distribution does not properly share files over the
network: we needed to patch the registry to fix the Server
service. They also distribute their software on a FAT partition,
which wastes a huge amount of space and demands extra steps to re-establish
file security even after converting to NTFS. |
| The software we chose to clone the systems with, Drive Image
Professional, was unable to strip the machine-dependent information from
the various registry and other files. There is a patch on their Web page
which looks like it might be useful. |
| We have yet to find a sensible XY plotting package for the machines.
In particular, we want a package which is commonly available (free or
mainstream, preferably both), and which can be driven from within a C++
program from the command line, and from a scripting language such as
Python. We explored Excel: in the current release of Python we could not
make the COM interface to work at all, and we had no clue how to use it
from the command line or from within a C++ program. (Presumably OLE
allows us to do this.) Gnu plotting also didn't work well. We continue
to explore: we have hopes for an NT port of xmgr, and are looking into
Genplot and Super Mongo. |
| We have had problems with strange erratic flickering in the Dell
1200HS 19" monitors (purchased as cost-sharing for the Intel
grant). The flickering looks quite similar to, but milder than, that
produced when degaussing the monitor. The problem is exhibited in two
separate rooms, but is not local to an individual monitor: several
nearby monitors will flicker in synchrony. We are exploring whether the
problem is with stray magnetic fields or is transmitted through the
power supply. We do not know at this time whether Dell's monitor is
particularly susceptible, or whether other brands are better shielded. |
|
|
