The GRACE Project Intro

Astrophysical Computer Simulations using Programmable Hardware

GRACE-2 Project Continuation 2010/2011 (Application Approved by Volkswagenstiftung)

The GRACE Project and Volkswagenstiftung in Wikipedia

The Silk Road Project
       - GRACE follow-up in Beijing, China ...

• December 2009: Workshop Accelerator-based Computing and Manycore (Manycore and Accelerator-based Computing for Physics and Astronomy Applications), jointly organised by Hemant Shukla and John Shalf (LBL/NERSC) , Tom Abel (KIPAC/Stanford) with Rainer Spurzem (GRACE Teams), at Kavli Institute for Particle Astrophysics and Cosmology (KIPAC) . In this workshop future transcontinental cooperation on software and hardware issues for novel many-core systems will be prepared, in which the GRACE-2 proposed project will play an important role.
• September 2009: : Rainer Spurzem appointed special visiting professor of Chinese Academy of Sciences, at National Astronomical Observatories of China , to build a computational astrophysics group and supervise the design and operation of a special accelerated supercomputer with 170 Tesla GPU's at NAOC (joint appointment with Kavli Institute of Astronomy and Astrophysics of at Peking University, China . The project is called the Silk Road Project . In this project also FPGA based accelerator cards MPRACE will be used and further explored as a follow-up of the GRACE project and in collaboration with the other GRACE teams.
• June 2009: GRACE project presented at invited talk of Rainer Spurzem during the International Supercomputing Conference in Hamburg in June 2009, titled Accelerating Astrophysical Particle Simulations with Programmable Hardware (FPGA & GPU) .
• March 2009: Peter Berczik obtained record benchmark of sustained 25 Teraflop/s for a single five million body direct N-body simulation on 100 Tesla graphical processing units (GPU) on the GPU clusters of the Institute of Process Engineering IPE of Chinese Academy of Sciences in Beijing in March 2009. The IPE is operating of order 1000 GPU cards with a cumulated peak speed of 1 Petaflop/s. It's hardware is the prototype for similar clusters to be opened at other CAS institutions (see top news). See preliminary plot (this work is in collaboration with Keigo Nitadori and Tsuyoshi Hamada , and also partly shows results obtained from the kolob frontier kolob frontier GPU cluster see below.
• November 2008: Kolob - First Supercomputer based on Graphical Processing Units at Heidelberg University (see University of Heidelberg Press Notice ); The project is funded by the University of Heidelberg Excellence Scheme , in collaboration with Ralf Klessen , Robi Banerjee (ITA) and Reiner Männer , Andreas Kugel , Guillermo Marcus (ZITI) . Graphic cards (=GPU: graphical processing units) offer a huge power for parallel computing tasks, which accelerates traditional processors, as they are used in standard PCs.
• April 2008: Gravitational Waves visualized from Black Hole Merger in GRACE Cluster titan (April 2008, avi-File 95 MB)
  

GRACE is an interdisciplinary project of theoretical astrophysics and computer engineering to use reconfigurable hardware (FPGA = field programmable gate array) for astrophysical particle simulations. Teams of Astronomisches Rechen-Institut, Zentrum für Astronomie Univ. Heidelberg (ARI-ZAH) , Lehrstuhl V, Central Institute for Computer Engineering (ZITI), Univ. Heidelberg, Location Mannheim (ZITI) and Lehrstuhl Computational Astrophysics (CAST) at the Universitätssternwarte München (USM) collaborate to use Programmable Hardware (FPGA) for gravitational and non-gravitational force computations (e.g. for smoothed particle hydrodynamics SPH) in particle based large scale astrophysical computer simulations. The dynamics of galactic nuclei with black holes, of merging galaxies, of star formation is modelled using this new computer architecture. Our main international collaborators are David Merritt and Hans-Peter Bischof with the GRAPEcluster at RIT as well as Naohito Nakasato, nakasato@riken.jp and Tsuyoshi Hamada, thamada@riken.jp , and the PROGRAPE/PGR project , RIKEN Japan

The GRACE cluster ''titan'' at the ARI ZAH (Heidelberg, Germany)
was assembled and delivered by sysgen GmbH
using MPRACE Boards of UMA with Xilinx Virtex-2 FPGA hardware (see the Mannheim FPGA Webpage );
micro-GRAPE6 hardware, produced by Hamamatsu Metrix Co. Japan