gtaylor@lasspvax.UUCP (Greg Taylor) (05/02/85)
CORNELL UNIVERSITY CENTER FOR THEORY AND
SIMULATION IN SCIENCE AND ENGINEERING (THEORY CENTER)
_W_H_O _W_E _A_R_E _A_N_D _W_H_A_T _W_E _D_O
The Theory Center is looking for experienced and creative
people interested in helping us build a "1990's environment"
for scientific computing. We have initiatives planned or in
progress in parallel systems, graphics,
networking/distributed systems, and
compilers/languages/environments for scientific worksta-
tions.
As the recipient of one of the NSF Supercomputer Facility
awards, and the only Facility (of four) committed to paral-
lel processing we can offer very high visibility. The na-
tional press has on several occasions called Cornell the
"one to watch".
_W_H_A_T _W_E _W_A_N_T / _W_H_O _W_E _W_A_N_T
Our immediate need is for:
1) Several experienced people, preferably with solid back-
grounds in Computer Science, who understand UNIX inter-
nals and networking and are interested in becoming in-
volved in the challenges of parallel processing and
distributed systems.
2) We want at least one person able to write compilers and
contribute to an optimization project which will in-
clude an AI component to let the user help with the op-
timization.
3) We need one person to undertake the coordination of all
the UNIX machines on our network (about 12 at present).
This involves teaching the support staff at these sites
about UNIX, about graphics, about scientific computing
and cajoling them into conforming to some loose stan-
dards. This person is also vital as the provider of
feedback to the rest of the group on what's happening
"out in the field".
4) We are looking for two Research (Computational) Associ-
ates who will be responsible for working directly with
scientists in helping them get their applications run-
ning, partcularly parallel applications. This person
must have significant competance in science, but must
also have experience in languages other than FORTRAN,
and UNIX experience.
These are not at present management positions, but we are
looking for people who have experience as project leaders,
or at least that potential, since we expect to form groups
working in several of the above areas over the next few
years.
_A _B_I_T _M_O_R_E _B_A_C_K_G_R_O_U_N_D
Our current system is an IBM 3084/QX with four attached FPS
264 Scientific Processors. There is a group which has as
its responsiblity the care of that system, but there is
another group whose mission is rather different and it's the
latter group for which we are recruiting. The immediate
need is to build a UNIX interface which will allow the
scientists (who mostly are users of UNIX machines on a net-
work called Theorynet) to access the facilities of the
IBM/FPS complex remotely without needing to become involved
in the IBM operating system. We expect to base this
software on the SUN network file system RPC and XDT proto-
cols and define our own applications layer.
We will be acquiring a Gould 9050 to serve as a UNIX home
for remote users, who will come in mainly over ARPANET.
This will also be a development machine for the staff,
although we expect to have various workstations (SUN, HP,
etc.) as well.
We have one parallel system at present, which consists of
eight FPS scientifics processors and a shared mass memory on
a high-speed bus. This system is attached to a UNIX host
and the software is UNIX-based.
The longer-term plan (say 18 months) is to acquire a paral-
lel system with >64 processors and at least 400 megaflops
which would be under the purview of the UNIX group. We have
a system in mind, but cannot talk about it. We don't plan
to build a system, but to buy it. However, we expect to
develop much of the software ourselves, and this effort has
already begun in a very modest way. There is plenty of
challenge here. There will also be a great deal of interest
in what we do, and how we do it, so if we are successful our
staff will have fame in abundance.
_W_H_O _T_O _C_O_N_T_A_C_T
Please send a resume and a cover letter indicating which
area you are interested in, to
Alison Brown
Cornell Theory Center
265 Olin Hall
Ithaca, N.Y. 14853
ARPA: alison@cornell
UUCP: ....{cmcl2,decvax,ihnp4,seismo}!cornell!alisonhestenes@sdcsla.UUCP (Eric Hestenes) (05/11/85)
> > CORNELL UNIVERSITY CENTER FOR THEORY AND > SIMULATION IN SCIENCE AND ENGINEERING (THEORY CENTER) > > Alison Brown You may think you will be the only ones doing parallel processing work, but in reality parallel processing work is certain to emerge ( and is planned at least in S.D. ) on all the 4 supercomputers. How could anyone pass up the chance to speak in *vectors* rather than in serial cycles? Eric ------------------------------------- |this *opinion* is exclusively my own | ------------------------------------- hestenes@nprdc ucbvax!sdcsvax!sdcsla!hestenes --------- Expect it .... ....... when you least expect it. --------