|
Home
Program
People
Maps
Charm Homepage
Live Webcast
|
Objectives
This workshop is aimed at taking stock of the funded
collaborative interdisciplinary research projects which the Parallel
Programming Laboratory (the developers of Charm++ and AMPI) are engaged in.
In addition to the computer scientists at PPL, the collaborators
represented at this workshop include those from:
- University of Illinois
at Urbana Champaign (CS, ME, TAM, MS, Physics, Beckman)
- University of
Washington, Seattle.
- Princeton University
- NYU
- IBM
- PSC (Pittsburgh
Supercomputing Center)
The interdisciplinary projects represented are:
- NIH: Parallel
Scalable Molecular Dynamics: NAMD (With Klaus Schulten, Bob Skeel)
- DOE: Rocket
Simulation (CSAR, director: Michael Heath)
- NSF: Material
Simulation (CPSD: with Robert Haber, Jonathan Dantzig, Duane Johnson)
- NSF: Computational
Cosmology Simulations (With Tom Quinn)
- NASA: Scalable
Visualizations for Computational Cosmology (With Tom Quinn)
- NSF: Quantum
Mechanical and classical modeling (QM/MM) (Roberto Car, Mike Klein,
Josep Toerrellas, Glenn Martyna, Mark Tuckerman, Nick Nystrom)
- NSF: Programming
Environments for Blue Gene Supercomputer and Applications (with S. Adve,
D. Padua, P. Geubelle)
Registration
The 2002 workshop has passed. Please see the videos and slides. We look forward to seeing
you next year.
Hotel
We have reserved a block of rooms at the Chancellor Hotel for the
Charm++ workshop. Participants need to call the Chancellor Hotel by 5th
October, 02 to reserve a room. (Phone: 1-800-257-6667.) Refer to the rooms
for "CHARM Workshop".
Schedule:
(may change slightly)
(All talks and tutorials are
open to the public.)
Oct 9
The morning sessions will all run
simultaneously:
- 9:15am-10:00am
Breakfast in 2240 DCL (for registered participants)
- 10am-noon: Charm++ tutorial
in 2240 DCL (Sindhura Bandhakavi)
[PowerPoint]
This hands-on tutorial will give a basic introduction to Charm++ for people
who have never written a Charm++ program. It will cover the basic
virtualization concept, proxies, and basic uses of
arrays.
- 10am-noon:
AMPI tutorial in 2261 DCL (Chao Huang)
[PowerPoint]
[PDF]
This hands-on tutorial will give a basic introduction to AMPI, our
adaptive version of MPI on top of Charm++. Some familiarity with MPI is
assumed. It will cover the major differences between MPI and AMPI, and
discuss threadsafety issues like global variables.
The afternoon sessions also run simultaneously:
- 2pm-4pm: FEM
framework tutorial in 2240 DCL (Gunavardan
Kakulapati)
Basics [PowerPoint]
Advanced [PowerPoint]
This tutorial will give a basic introduction to the Charm++
Finite-Element Framework (FEM Framework). It will cover the concepts and
major functions of the FEM framework, and show how to convert a serial
FEM program into a parallel FEM program running in the framework. Its
coverage will be much the same as that of the FEM Workshop.
- Discussions on
Quantum Simulation in 2508 DCL.
Oct 10
The formal workshop begins in
2240 DCL. The morning session will consist of talks followed by a discussion
session:
After a break for lunch, the afternoon session includes:
- 1:30-2:00pm:
Faucets: Efficient Utilization of Multiple Clusters
(Jay
DeSouza)
[PowerPoint]
- 2:00-2:30pm:
CSAR and Multi-component simulations (Orion
Lawlor)
[PowerPoint]
- 2:30-3:00pm:
Unstructured Grid applications in material process simulation (Jon Dantzig, Mechanical Engineering)
- 3:30-4:30pm:
Parallel Quantum Dynamics (Roberto Car, Glenn Martyna, Mark Tuckerman, Ramkumar Vadali
[PowerPoint])
- Discussion and
conclusion
A formal banquet will follow the talks at the Silvercreek
Restaurant in the Greenhouse room, starting at 6:30 pm. (for registered
participants)
Dinner choices will be Fusilli Primavera
(vegetarian) and Apple Amaretto Chicken--we will need your choice by
Monday, October 4.
Dinner will include soup or salad, beverage, vegetable, rice, and
fresh baked breads. For dessert, we will have a small dessert buffet where
guests can have either chocolate layer cake or mountain berry flan tart... or
a little of both. There will be a cash bar.
October 11
The morning sessions will run
simultaneously:
- 9:15am-10:00am
Breakfast in 2240 DCL (for registered participants)
- 10am-noon:
Clusters tutorial in 2240 DCL (Jay DeSouza
and Sameer Kumar)
[PowerPoint]
installation files
This tutorial will provide a basic introduction to the clustering
technology provided by Charm++, and demonstrate how that technology can
improve cluster utilization.
Outline
- Introduction
- High-Level
- Description of
Faucets
- Description of
Adaptive Jobs
- Description of the
Adaptive Queuing System
- Demonstration of
the Complete System
- Performance Results
- Lower Level
- How to Write an
Adaptive Program
- Installation of
the Adaptive Queuing System
- Connecting your
cluster to an existing Faucets server
- Installing your
own Faucets Server
- Conclusion
- Discussion session
on the N-Chilada project with U. Washington computational cosmologists
in 2508 DCL.
|