Petapath

Contact Michal Harasimiuk +44 117 911 7544 [email protected] NVIDIA Bio Workbench: more science more quickly

NVIDIA has always recog- By harnessing the perform- The codes can be divided nised the value of building an ance of NVIDIA Tesla Com- into two groups, and include: open ecosystem that puting Solutions, it can turn a allows users to make use of GPU standard PC into a “computa- ACE MD accelerated applications with the tional laboratory” capable of AMBER minimum of effort. running complex bioscience, BigDFT (ABINIT) The NVIDIA® Tesla™ Bio particularly Molecular Dynamics GROMACS Workbench is a new and criti- and codes. HOOMD Bio Workbench cally important part of this eco- These include well established LAMMPS Workshop: system, specifically addressing codes such as AMBER and other NAMD, TeraChem, VMD users in the Life Sciences and newer codes written from the Bio Informatics The University of . The ground up to support GPUs. All CUDA-BLASTP Nottingham Bio Workbench enables bio- offer substantial improvements in CUDA-EC physicists and computational throughput, enabling users to CUDA-MEME 23rd Nov 2010 to push the boundaries obtain more results more quickly. CUDASW++ of biochemical research in Bio Workbench codes can also GPU-HMMER Room A103 a practical and cost effective way. be run on GPU equipped clus- HEX Protein Centre for Astronomy ters. Jacket (MATLAB Plugin) MUMmerGPU and Particle , Cripps Building Benchmark figures compiled by Ross C. Walker from the San Diego Supercomputer Center show performance Petapath and improvements, over an x86 reference machine (2.8GHz Intel E5462 dual quad core), ranging from a minimum NVIDIA would like to of 3x faster on a single NVIDIA Tesla C2050 to over 17x depending on the simulation. invite researchers, stu- 8 core C1060 C2050 dents and industrial Benchmark Speedup users to a series of (ns/day) (ns/day) (ns/day) seminars and work- TRPCage (304 atoms) 116 254 368.2 3.17 x shops dedicated to the AMBER Bio Workbench, with Implicit Solvent Myoglobin (2,492 atoms) 4.4 27.9 49.9 11.34 x a particular focus on (GB) Nucleosome (25,095 atoms) 0.06 0.52 1.04 AMBER 11. 17.33 x AMBER DHFR NVE (23,558 atoms) 5.94 11.93 20.7 3.48 x To register visit: Explicit Solvent DHFR NPT (23,558 atoms) 5.33 9.8 18.11 3.39 x petapath.com/nvidia (PME) FactorIX NVE (90,906 atoms) 1.67 3.51 5.19 3.1 x A lot can be achieved in one day

Petapath Plan of the day

10:00 a.m. Registration opens, morning coffee

11 a.m. - 1 p.m. BioWorkbench seminar

1 p.m. - 2 p.m. Lunch and discussions

2 p.m - 5 p.m. CUDA workshop

5 p.m. Day end

Contact There will be coffee breaks with light snacks throughout the day Michal Harasimiuk +44 117 911 7544 [email protected] Our seminars are aimed Taking a selection of Please help us make the at users who already have ex- these codes we would like to seminars relevant to your perience with molecular dy- demonstrate and analyse ad- compute problems by sending namics and bio informatics. vantage of Fermi compute us the input files and bench- While we will start, taking power and which hardware marks ahead of the sessions. If AMBER as an example, by configurations will work best you would like to see how your introducing the latest devel- for the different class of prob- simulations perform on the opments in AMBER 11 with lems. We will then give an latest NVIDIA hardware and particular emphasis on new overview of the different op- share this experience with methods and GPU support, tions available on the GPU other attendees, then please our goal is to give researches and how to improve perform- contact us at least two weeks an opportunity to trial their ance by selecting specific op- before the date of the seminar. preferred codes and data on tions. the GPU enabled cluster. The Bio Station Petapath has worked with Supermicro to architect a deskside supercomputer for Bio Workbench applications. Supports up-to four We created the configuration double width GPUs that provides the right balance of bandwidth and compute, 92.31% efficient power supply and the memory footprint sufficient for large systems.

Bio Station can be config- ured with up-to 12 CPU cores, • Intel® Xeon® processor 5600/5500 series, with QPI up-to 6.4 GT/s four Tesla M2050 Computing • Up-to 192GB DDR3 1333/ 1066/ 800MHz ECC Registered DIMM Modules or professional • 4U tower or rack mount Quadro boards.

Registration is open:

www.petapath.com/nvidia