Today AMD announced that the company’s FirePro GPU accelerators are delivering more than 3 petaFLOPS to the GSI GmbH for physics research. Located in Darmstadt, Germany, GSI is a research facility for heavy ion research including cancer therapy.
AMD is proud to collaborate with ASUS, the Frankfurt Institute for Advanced Studies, (FIAS) and GSI to support such important physics and computer science research,” said David Cummings, senior director and general manager, professional graphics, AMD. “This installation reaffirms AMD’s leading role in HPC with the implementation of the AMD FirePro S9150 server GPUs in this three petaFLOPS supercomputer cluster. AMD and ASUS are enabling OpenCL applications for critical science research usage for this cluster. We’re committed to building our HPC leadership position in the industry as a foremost provider of computing applications, tools and technologies.”
Each of the 160 ASUS ESC4000 G2S nodes1 of the new L-CSC cluster include four AMD FirePro S9150 server GPUs, for a total GPU peak of 3.25 petaFLOPS single precision and 1.62 petaFLOPS double precision performance. The cluster is enabling Lattice Quantum Chromo Dynamics (Lattice QCD) computational research using one of the fastest OpenCL implementations for research applications in the world. Lattice QCD calculations are applicable to a range of studies in high energy and heavy ion physics.
We had excellent cooperation with ASUS and AMD to make this project happen in such a short timeframe,” said Professor Doctor Volker Lindenstruth, professor at Goethe University of Frankfurt and chairman of Frankfurt Institute for Advanced Studies. “The ASUS ESC4000 G2S servers and the AMD FirePro S9150 GPUs are an extremely powerful basis for the L-CSC cluster and they provide the compute capabilities we need for our research. The large 16GB of memory of the AMD FirePro S9150 server GPUs allows us to run most LQCD computations on one GPU without inter-GPU and inter-node communication resulting in very efficient LQCD application operation. We have chosen a multi-GPU approach with four GPU boards per server to enable an extremely cost-efficient and energy-efficient configuration.”
High energy physics experiments, such as at GSI, need huge compute farms for simulations. The L-CSC cluster is aimed at QCD computations, which require extremely high memory bandwidth. The implementation of the Lattice QCD algorithm was developed at FIAS using an OpenCL™ based implementation for portability and flexibility.
AMD is exhibiting with AMD FirePro S9150 server GPUs and AMD server technologies at SC14 booth #839 from November 17 to 20.
Sign up for our insideHPC Newsletter.