🚀 go-pugleaf

There was a mistake about the date of the following seminar:

In article <104786@bu.edu> dme@cs.bu.edu (David Evans) writes:

>  Mon 14 Dec  2:00 pm
   ^^^^^^^^^^
>MIT Building NE43, Rm 518, 545 Technology Square, Cambridge (On Main Street
>near intersection of Albany Street, @ Polaroid Building) Refreshments at 1.45.
>Speaker: Kathy Yelick, University of California, Berkeley
>"Programming Irregular Applications on Distributed Memory Multiprocessors"
>Host: MIT Practice of Computation Seminar and John Guttag
>For Info: <alr@anne.lcs.mit.edu> Anne Rubin

The actual date is Tuesday December 15. Below is the abstract of for the talk:


P O C S    S E M I N A R


DATE:  TUESDAY, DECEMBER 15, 1992

TIME:  REFRESHMENTS AT 1:45 
TALK AT 2:00

PLACE:  NE43-518



Programming Irregular Applications on
Distributed Memory Multiprocessors

Kathy Yelick
University of California, Berkeley


The difficulty of parallelizing a given program is
strongly correlated to the degree of irregularity in
the program's structures.  Most of the applications 
that have been parallelized successfully on large scale 
multiprocessors are those with a high degree of regularity.
The goal of our research is to provide languages and libraries
that will simplify parallel programming for irregular applications,
and consequently increase the class of applications that can be
effectively parallelized.  We have approached this problem by
implementing irregular applications on a CM5 multiprocessor, while
at the same time developing layers of system level software to
provide a common base for the applications.  

In this talk we describe two irregular applications: the Grobner
basis computation and an event-based circuit simulator.  Both
perform well on the CM5, beating the speedups obtained previously
on both shared and distributed memory machines.  We identify common
features of these applications and others, and describe language and
library support currently under development.  The implementations
are organized around a set of nondeterministic "transition-axioms"
that access local and distributed data structures.  The system support
includes a distributed object library called Multipol and scheduling
for multiple levels of threading among the transitions.  We conclude
with a discussion of how this work fits into a larger collaborative
effort to build an integrated software environment for distributed
memory multiprocessors.)


HOST:  John Guttag



-- 
 ---
	Gene Itkis		(gene@cs.bu.edu)

Newsgroups: ne.seminars
Subject: Another CORRECTION for Kathy Yelick's talk (was Re: Boston Combined Schedule)
References: <104786@bu.edu> <104822@bu.edu>
Organization: Computer Science Department, Boston University, Boston, MA, USA
From: gene@cs.bu.edu (Gene Itkis) 

There is another change, this time of room:

= P O C S    S E M I N A R
= 
= Programming Irregular Applications on
= Distributed Memory Multiprocessors
= 
= Kathy Yelick
= University of California, Berkeley
= 
= DATE:  TUESDAY, DECEMBER 15, 1992
= 
= TIME:  REFRESHMENTS AT 1:45 
= TALK AT 2:00
= 
= PLACE:  NE43-518
	       ^^^

This has being changed to the 6th floor playroom (instead of room 518).

Sorry for the confusion.
-- 
 ---
	Gene Itkis		(gene@cs.bu.edu)