GPU Gems for Astro- and Particle Physics (GGAPP)
Proposal to build a major computing facility based on special hardware
Link to technical data / proposal writing Principal Investigator:
Co-Investigators:
-
Ralf Klessen,
Institut für Theoretische Astrophysik (ITA) ,
Zentrum für Astronomie der Univ. Heidelberg (ZAH)
-
Udo Kebschull,
Kirchhoff-Institut für Physik (KIP), Universität Heidelberg
-
Reinhard Männer, Informatik V,
Zentrales Institut für Technische Informatik (ZITI), Univ. Heidelberg,
Standort Mannheim
Industrial Partner:
Scientific Advisory Panel:
-
Douglas N.C. Lin ,
Professor, Univ. of California Observatories, UC Santa Cruz, USA
and Founding Director of
The Kavli Institute for
Astronomy and Astrophysics at
Peking University , Beijing, China.
-
Mordechai-Marc MacLow ,
Curator and Curator-in-Charge,
Department of Astrophysics, Chair Division of Physical Sciences,
American Museum of Natural History, New York, USA ,
-
David Merritt ,
Professor,
Astrophysics Group, Rochester Institute of Technology, New York State, USA
Outline of the project (duration 5 years):
-
Build a supercomputer based on graphics processing units (GPU, e.g. NVIDIA GeForce 9600 GT), used for high-performance
numerical computation, on the scale of more than 100 Tflop/s. Location: ZITI, Standort Mannheim.
- From Newton to Einstein:
Model Binary and Multiple Black Holes in Galactic Nuclei from Galactic Mergers down
to Relatvistic Coalescence, predict gravitational wave generation, realistic
galaxy models, spin dynamics of black holes interacting with stellar clusters,
accretion disks, and other black holes, recoil effects. Development of direct
N-body and smoothed particle hydrodynamics simulation software for parallel GPU
clusters, utilization of grid technology. (ARI-ZAH, Team of R. Spurzem)
- Star and Star Cluster Formation in Dense Molecular Clouds;
Software Development for Tree-SPH smoothed particle hydrodynamics simulation on
parallel GPU clusters. (ITA-ZAH, Team of
R. Klessen).
- Fast Particle Tracking for (CERN, Switzerland) and CBM (FAIR/GSI, Germany).
The terabytes of experimental data must be carefully and in depth analyzed.
We expect a few orders of magnitude speed-up when running the track finder application
on the proposed cluster hardware. The proposed cluster can therefore play a significant
role not only in development
of the reconstruction algorithms, but also in the analysis of real data.
Implementation, Test and Application of automated cluster management.
(KIP, Team of U. Kebschull).
-
Algorithm Development for particle based simulations, hardware and network configuration.
Application Programmer Interfaces to efficiently use the GPU hardware have to be developed
and supported; high-level language tools need to adapted to ease the programmability of the
new devices. Application users have to supported in optimizing their algorithm. The network
infrastructure and memory and data flow architecture of the entire system and of GPUs has to
be optimized. (ZITI, Team of R. Männer)
The following links are for internal use of the teams only:
Proposal Files (Sources and pdf) Direct Link to File List (Last Update Feb. 28, 13:30)
Proposal Files (Complete tar.gz ball) ggapp.tar.gz (Feb. 25, 03:00 a.m.)
Contact: R. Spurzem
NVIDIA - Visual Computing Techniques - Douglas N.C. Lin , Professor, Univ. of California Observatories, UC Santa Cruz, USA and Founding Director of The Kavli Institute for Astronomy and Astrophysics at Peking University , Beijing, China.
- Mordechai-Marc MacLow , Curator and Curator-in-Charge, Department of Astrophysics, Chair Division of Physical Sciences, American Museum of Natural History, New York, USA ,
- David Merritt , Professor, Astrophysics Group, Rochester Institute of Technology, New York State, USA
-
Build a supercomputer based on graphics processing units (GPU, e.g. NVIDIA GeForce 9600 GT), used for high-performance
numerical computation, on the scale of more than 100 Tflop/s. Location: ZITI, Standort Mannheim.
- From Newton to Einstein: Model Binary and Multiple Black Holes in Galactic Nuclei from Galactic Mergers down to Relatvistic Coalescence, predict gravitational wave generation, realistic galaxy models, spin dynamics of black holes interacting with stellar clusters, accretion disks, and other black holes, recoil effects. Development of direct N-body and smoothed particle hydrodynamics simulation software for parallel GPU clusters, utilization of grid technology. (ARI-ZAH, Team of R. Spurzem)
- Star and Star Cluster Formation in Dense Molecular Clouds; Software Development for Tree-SPH smoothed particle hydrodynamics simulation on parallel GPU clusters. (ITA-ZAH, Team of R. Klessen).
- Fast Particle Tracking for (CERN, Switzerland) and CBM (FAIR/GSI, Germany). The terabytes of experimental data must be carefully and in depth analyzed. We expect a few orders of magnitude speed-up when running the track finder application on the proposed cluster hardware. The proposed cluster can therefore play a significant role not only in development of the reconstruction algorithms, but also in the analysis of real data. Implementation, Test and Application of automated cluster management. (KIP, Team of U. Kebschull).
- Algorithm Development for particle based simulations, hardware and network configuration. Application Programmer Interfaces to efficiently use the GPU hardware have to be developed and supported; high-level language tools need to adapted to ease the programmability of the new devices. Application users have to supported in optimizing their algorithm. The network infrastructure and memory and data flow architecture of the entire system and of GPUs has to be optimized. (ZITI, Team of R. Männer)
The following links are for internal use of the teams only:
Proposal Files (Sources and pdf) Direct Link to File List (Last Update Feb. 28, 13:30)
Proposal Files (Complete tar.gz ball) ggapp.tar.gz (Feb. 25, 03:00 a.m.)
Contact: R. Spurzem