ChaNGa (Charm N-body GrAvity)
Cosmological simulation framework "ChaNGa" (originally "N-Chilada") is a collaborative project with Prof. Thomas Quinn (University of Washington: N-Body Shop) supported by the NSF.
ChaNGa (CHArm++ N-body GrAvity solver) was built as a Charm++ application that employs a tree algorithm to represent the simulation space. This tree is constructed globally over all particles in the simulation and segmented into elements named TreePieces. The various TreePieces are distributed by the Charm++ runtime system to the available processors for parallel computation of the gravitational forces. Each TreePiece is implemented in ChaNGa as a chare and can migrate between processors (e.g. during load balancing) as the simulation evolves. Because various TreePieces may reside on the same processor, ChaNGa employs a software caching mechanism that accelerates repeated accesses to the same remote particle data. Each processor contains a cache, implemented as a Charm++ group, to store remotely fetched TreePieces. Another important feature in ChaNGa is the use of a multi-step integration scheme, which on the smallest (most frequent) timestep updates only those particles with the highest speeds.
MAJOR GOALS: The main goal of this work is to produce the most scalable code available for Cosmological simulations; to achieve that, load balancing is a fundamental capability, both at singlestepping and at multistepping. Also, achieving good sequential performance is important.
---
Historically, this project was called N-Chilada, a framework for enabling N-Body based parallel simulations, and especially those in Cosmology and Astronomy. N-Chilada was a collaborative project with Prof. Thomas Quinn (University of Washington) and Prof. Marianne Winslett (University of Illinois). This research page has been archived and can be found here.
People
Papers / Talks
-
23-022023
MS Thesis -
22-062022
PaperParaTreeT: A Fast, General Framework for Spatial Tree Traversal
- Joseph Hutter
- Justin Szaday
- Jaemin Choi
- Spencer Wallace
- Simeng Liu
- Laxmikant Vasudeo Kale
- Thomas Quinn
-
18-022018
PaperMulti-level Load Balancing with an Integrated Runtime Approach
- Seonmyeong Bak
- Harshitha Menon
- Sam White
- Matthias Diener
- Laxmikant Vasudeo Kale
-
16-192016
PaperHandling Transient and Persistent Imbalance Together in Distributed and Shared Memory
- Harshitha Menon
- Seonmyeong Bak
- Phil Miller
- Sam White
- Nitin Bhat
- Laxmikant Vasudeo Kale
-
15-042015
Phd Thesis -
14-302014
PaperAdaptive Techniques for Clustered N-Body Cosmological Simulations
- Harshitha Menon
- Lukasz Wesolowski
- Gengbin Zheng
- Pritish Jetley
- Laxmikant Vasudeo Kale
- Thomas Quinn
- Fabio Governato
-
14-182014
PaperTRAM: Optimizing Fine-grained Communication with Topological Routing and Aggregation of Messages
- Lukasz Wesolowski
- Ramprasad Venkataraman
- Abhishek Gupta
- Jae-Seung Yeom
- Keith Bisset
- Yanhua Sun
- Pritish Jetley
- Thomas Quinn
- Laxmikant Vasudeo Kale
-
13-412013
Paper- Thomas Quinn
- Pritish Jetley
- Laxmikant Vasudeo Kale
- Filippo Gioachin
-
13-162013
PaperParallel Science and Engineering Applications: The Charm++ Approach
- Laxmikant Vasudeo Kale
- Abhinav Bhatele
-
10-162010
PaperScaling Hierarchical N-Body Simulations on GPU Clusters
- Pritish Jetley
- Lukasz Wesolowski
- Filippo Gioachin
- Laxmikant Vasudeo Kale
- Thomas Quinn
-
08-032008
PaperMassively Parallel Cosmological Simulations with ChaNGa
- Pritish Jetley
- Filippo Gioachin
- Celso Mendes
- Laxmikant Vasudeo Kale
- Thomas Quinn
-
07-092007
PaperTowards Petascale Cosmological Simulations with ChaNGa
- Filippo Gioachin
- Pritish Jetley
- Celso Mendes
- Laxmikant Vasudeo Kale
- Thomas Quinn
-
06-212006
PosterCosmological Simulations on Supercomputers
- Filippo Gioachin
- Sayantan Chakravorty
- Celso Mendes
- Laxmikant Vasudeo Kale
- Thomas Quinn
-
06-092006
MS Thesis -
06-012006
PaperScalable Cosmological Simulations on Parallel Machines
- Filippo Gioachin
- Amit Sharma
- Sayantan Chakravorty
- Celso Mendes
- Laxmikant Vasudeo Kale
- Thomas Quinn
-
04-192004
PosterSalsa: a Parallel, Interactive, Particle-Based Analysis Tool
- Thomas Quinn
- Laxmikant Vasudeo Kale
- Filippo Gioachin
- Orion Lawlor
- Graeme Lufkin
- Gregory Stinson
-
98-071998
PaperStatic Networks: A Powerful and Elegant Extension to Concurrent Object-Oriented Languages
- Joshua Yelon
- Laxmikant Vasudeo Kale
-
93-131993
PaperA Load Balancing Strategy For Prioritized Execution of Tasks
- Amitabh Sinha
- Laxmikant Vasudeo Kale
-
92-101992
PaperDynamic Adaptive Scheduling in an Implementation of a Data Parallel Language
- Ed Kornkven
- Laxmikant Vasudeo Kale