Status of the US LQCD Infrastructure Project



Similar documents
SciDAC Software Infrastructure for Lattice Gauge Theory

Workflow Requirements (Dec. 12, 2006)

Symmetric Multiprocessing

A Study on the Scalability of Hybrid LS-DYNA on Multicore Architectures

Multi-Threading Performance on Commodity Multi-Core Processors

1 Bull, 2011 Bull Extreme Computing

Simulation Platform Overview


Parallel Programming Survey

Multi-core Curriculum Development at Georgia Tech: Experience and Future Steps

Agenda. HPC Software Stack. HPC Post-Processing Visualization. Case Study National Scientific Center. European HPC Benchmark Center Montpellier PSSC

Microsoft Compute Clusters in High Performance Technical Computing. Björn Tromsdorf, HPC Product Manager, Microsoft Corporation

How To Compare Amazon Ec2 To A Supercomputer For Scientific Applications

Designing and Building Applications for Extreme Scale Systems CS598 William Gropp

High Performance Computing. Course Notes HPC Fundamentals

OpenMP Programming on ScaleMP

benchmarking Amazon EC2 for high-performance scientific computing

GPU System Architecture. Alan Gray EPCC The University of Edinburgh

Building an energy dashboard. Energy measurement and visualization in current HPC systems

Supercomputing Status und Trends (Conference Report) Peter Wegner

Turbomachinery CFD on many-core platforms experiences and strategies

BSC - Barcelona Supercomputer Center

Parallel Compression and Decompression of DNA Sequence Reads in FASTQ Format

Part I Courses Syllabus

Finite Elements Infinite Possibilities. Virtual Simulation and High-Performance Computing

Data Centric Systems (DCS)

Lattice QCD Performance. on Multi core Linux Servers

Petascale Visualization: Approaches and Initial Results

numascale White Paper The Numascale Solution: Extreme BIG DATA Computing Hardware Accellerated Data Intensive Computing By: Einar Rustad ABSTRACT

White Paper The Numascale Solution: Extreme BIG DATA Computing

LBM BASED FLOW SIMULATION USING GPU COMPUTING PROCESSOR

Performance of the JMA NWP models on the PC cluster TSUBAME.

Cluster Scalability of ANSYS FLUENT 12 for a Large Aerodynamics Case on the Darwin Supercomputer

Cellular Computing on a Linux Cluster

Clusters: Mainstream Technology for CAE

Introduction to High Performance Cluster Computing. Cluster Training for UCL Part 1

2015 The MathWorks, Inc. 1

Next Generation GPU Architecture Code-named Fermi

Text Mining Approach for Big Data Analysis Using Clustering and Classification Methodologies

Speeding up MATLAB and Simulink Applications

Building a Top500-class Supercomputing Cluster at LNS-BUAP

Cloud-WIEN2k. A Scientific Cloud Computing Platform for Condensed Matter Physics

White Paper The Numascale Solution: Affordable BIG DATA Computing

Data Deduplication in a Hybrid Architecture for Improving Write Performance

Comparing the performance of the Landmark Nexus reservoir simulator on HP servers

Overview of HPC systems and software available within

Large Vector-Field Visualization, Theory and Practice: Large Data and Parallel Visualization Hank Childs + D. Pugmire, D. Camp, C. Garth, G.

Cluster Computing in a College of Criminal Justice

Multicore Parallel Computing with OpenMP

So#ware Tools and Techniques for HPC, Clouds, and Server- Class SoCs Ron Brightwell

Activities and Results

Pedraforca: ARM + GPU prototype

HPC performance applications on Virtual Clusters

Self service for software development tools

Technical Computing Suite Job Management Software

The Lattice Project: A Multi-Model Grid Computing System. Center for Bioinformatics and Computational Biology University of Maryland

Making Multicore Work and Measuring its Benefits. Markus Levy, president EEMBC and Multicore Association

Computer Organization

Overview of HPC Resources at Vanderbilt

Introduction to Cloud Computing

Performance Analysis of a Hybrid MPI/OpenMP Application on Multi-core Clusters

Cluster, Grid, Cloud Concepts

Interoperability between Sun Grid Engine and the Windows Compute Cluster

HPC and Big Data. EPCC The University of Edinburgh. Adrian Jackson Technical Architect

Using the Windows Cluster

Lecture 1 Introduction to Parallel Programming

Parallel Computing. Introduction

Application of Predictive Analytics for Better Alignment of Business and IT

Cluster Implementation and Management; Scheduling

LS-DYNA Best-Practices: Networking, MPI and Parallel File System Effect on LS-DYNA Performance

Lecture 11: Multi-Core and GPU. Multithreading. Integration of multiple processor cores on a single chip.

Numerical Calculation of Laminar Flame Propagation with Parallelism Assignment ZERO, CS 267, UC Berkeley, Spring 2015

Performance Characteristics of a Cost-Effective Medium-Sized Beowulf Cluster Supercomputer

High Productivity Computing With Windows

Data Analytics at NERSC. Joaquin Correa NERSC Data and Analytics Services

Simple Introduction to Clusters

Programming models for heterogeneous computing. Manuel Ujaldón Nvidia CUDA Fellow and A/Prof. Computer Architecture Department University of Malaga

Load Balancing on a Non-dedicated Heterogeneous Network of Workstations

MPI / ClusterTools Update and Plans

Position Classification Flysheet for Computer Science Series, GS Table of Contents

Is a Data Scientist the New Quant? Stuart Kozola MathWorks

Unleashing the Performance Potential of GPUs for Atmospheric Dynamic Solvers

Transcription:

Status of the US LQCD Infrastructure Project Bob Sugar Allhands Meeting, April 6-7, 2006 p.1/15

Overview SciDAC-2 Proposal The LQCD Computing Project Allhands Meeting, April 6-7, 2006 p.2/15

SciDAC-2 Proposal Two areas of emphasis Software Hardware Research and Development Fourteen participating institutions Three national laboratories Eleven universities Allhands Meeting, April 6-7, 2006 p.3/15

Major Components of SciDAC-2 Software Machine specific software Infrastructure for physics applications Uniform computing environment Allhands Meeting, April 6-7, 2006 p.4/15

SciDAC Software Level 4 QCD Physics Toolbox Shared Algorithms, Building Blocks, Visualization, Performance Tools Workflow and Data Analysis tools Level 3 QOP (Optimized in asm) Dirac operators, inverters, force terms, etc. Uniform User Environment Runtime, accounting, grid tools Level 2 QDP (QCD Data Parallel) Lattice wide operations, Data shifts. Hides message passing/layout QIO (QCD IO) Binary/XML files & ILDG tools Level 1 QLA (QCD Linear Algebra) C, C++ and asm QMP (QCD Message Passing) MPI, native QCDOC, GigE, etc QMC (QCD multi core) Threaded interface and SMP lib Allhands Meeting, April 6-7, 2006 p.5/15

Increased Participation of Computer Scientists Ted Bapty (Vanderbilt): Monitoring and controlling large computers Massimo DiPierro (DePaul): Visualization Dan Reed and Ying Zheng (North Carolina): Performance analysis Xien-He Sun (IIT): Workflow for large physics analysis projects Allhands Meeting, April 6-7, 2006 p.6/15

Hardware Research and Development Investigation of Cluster Components Investigation of a specialized computer for lattice QCD Allhands Meeting, April 6-7, 2006 p.7/15

DOE Strategic Timelines High Energy Physics Program (2009): Use computer simulations to calculate strong interactions between particles so precisely that theoretical uncertainties no longer limit our understanding of their interactions. Nuclear Physics Program (2017): Provide precise lattice calculations to compare with established nucleon properties. Allhands Meeting, April 6-7, 2006 p.8/15

Lattice QCD Computing Project Objectives Acquire the most capable and cost-effective hardware for lattice QCD on a yearly basis Operate the SciDAC Prototype clusters, QCDOC and hardware acquired in the Project Budget FY 2006 FY 2007 FY 2008 FY 2009 Hardware 1,850 1,694 1,630 798 Operations 650 806 870 902 Total 2,500 2,500 2,500 1,700 Budget figures are in thousands of dollars. Allhands Meeting, April 6-7, 2006 p.9/15

Plan for FY 2006 Baseline: October, 2005 SciDAC Prototype Clusters 1.7 Tflop/s QCDOC 4.2 Tflop/s FY 2006 JLab Cluster 128 + 128 dual core Pentium 830-D Nodes Infiniband interconnect 0.6 Tflop/s $0.86 per Mflop/s FY 2006 FNAL Cluster 450 dual-core, dual-processor nodes Infiniband interconnect 1.7 + 0.3 = 2.0 Tflop/s Allhands Meeting, April 6-7, 2006 p.10/15

Plan for FY 2007 Baseline: October 2006 Clusters 4.3 Tflop/s QCDOC 4.2 Tflop/s Large cluster at JLab Multi-core, multi-processor nodes Infiniband interconnect Milestone is 3.1 Tflop/s Allhands Meeting, April 6-7, 2006 p.11/15

Project Management Federal Project Manager: John Kogut Contract Project Manager: Don Holmgren Associate Contract Project Manger: Bakul Banerjee Site Mangers BNL: Tom Schlagel FNAL: Amitoj Singh JLab: Chip Watson Allhands Meeting, April 6-7, 2006 p.12/15

Project with a Capitol P Well defined work responsibilities and budgets Regular reporting on progress in all aspects of the work Milestones which must be met Hardware acquisition Hardware operation Science Yearly submission of Exhibit 300 to the Office of Management and Budget Yearly project review Allhands Meeting, April 6-7, 2006 p.13/15

Our Responsibilities Perform calculations that will have major impacts on research in high energy and nuclear physics Raise the visibility of our field in the broader physics community Provide data on our scientific progress in a timely fashion Enable Don Holmgren to report progress in meeting science milestones Enable John Kogut to keep interested parties within the DOE up to date on our progress Provide input on user needs to the project management Allhands Meeting, April 6-7, 2006 p.14/15

HAPPY RETIREMENT JEFF Allhands Meeting, April 6-7, 2006 p.15/15