eugene@ames.UUCP (Eugene Miya) (08/13/84)
][ Using the Massively Parallel Processor for CFD Applications John Bruno Computer Science Dept. University of California Santa Barbara, CA and Research Institute for Advanced Computer Science This report presents the results of an investigation into the feasibility of using the MPP for direct and large eddy simulation on the Navier-Stokes equations. A major part of this study has been devoted to the "implementation" of two of the standard numerical algorithms for CFD. These implementations were not run because the MPP delivered to NASA Goddard SFC does not have sufficient capacity. Instead, we designed detailed implementation plans and estimate time and space requirements of our algorithms on a suitably configured MPP. In addition we consider other issues related to practical implementation of these algorithms on MPP-like architectures: namely, adaptive grid generation, zonal boundary conditions, table lookup problem, and the software interface. Our performance estimates show that the architectural components of the MPP, the staging memory and the array unit, appear to be well suited to the numerical algorithms of CFD. This combined with the prospect of building a faster and larger MPP-like machines holds the promise of achieving sustained gigaflop rates that are required for numerical simulations like CFD. Wednesday, August 22, 1984, 9:30-10:30 AM Building 200, Room 213 (Committee Room) Foreign nationals must request permission to attend, preferably 1 week in advance. Please mail requests to: emiya@ames-vmsb.ARPA. Include country of origin, whether you have a green card, and be prepared to show passport and other forms of ID on the day of seminar.
berry@zinfandel.UUCP (Berry Kercheval) (08/14/84)
Pardon my ignorance but what is "CFD". Also, if there is no requirement for US Citizens to prove thier citizenship to attend, how secure is this seminar? The only people asked to supply documentation are the only ones who might be kept out. What is to stop someone from just saying "I am a US citizen, let me in."? -- Berry Kercheval Zehntel Inc. (ihnp4!zehntel!zinfandel!berry) (415)932-6900
eugene@ames.UUCP (Eugene Miya) (01/30/85)
Joint RIACS and ECT Branch Seminar
SPEAKER: Vason Srini
Lawrence Berkeley Laboratory
TOPIC: A Message-Based Processor for Data Flow
ABSTRACT: The architecture of a message-based processor for a dataflow
system is described. The primary objective in developing the architecture
is to determine the overhead in executing dataflow graphs and the various
tradeoffs related to the overhead. Nine classes of instructions have been
identified for the message-based processor to support the execution of
program graphs. These instructions support message communication, string
operations, floating point operations, fixed point operations, and
manipulate structures such as tables and streams of data. To reduce the
execution overhead in the dataflow system, several features are included in
the processor. There is a data memory (table memory) in the processor to do
string operations on a vector of data.
DATE: 5 Feb 1985 TIME: 10 AM BLDG: 233 ROOM: 172
POINT OF CONTACT: E. Miya PHONE NUMBER: (415)-694-6453
VISITORS ARE WELCOME: Register and obtain vehicle pass at Ames Visitor
Reception Building (N-253) or the Security Station near Gate 18. See map
below. Do not use the Navy Main Gate.
Non-citizens (except Permanent Residents) must have prior approval from the
Director's Office one week in advance. Submit requests to the point of
contact indicated above. Non-citizens must register at the Visitor
Reception Building. Permanent Residents are required to show Alien
Registration Card at the time of registration.
Excuse the late posting, I have been at a meeting.