Getting Access to ALCF Resources and Services

Getting Access to ALCF
Resources and Services
David E. Martin
Manager, User Services and Outreach
dem@alcf.anl.gov
SC10 – November 17, 2010

Argonne Leadership Computing Facility
 ALCF was established in 2006 at Argonne to provide the
computational science community with a leading-edge
computing capability dedicated to breakthrough science and
engineering
 One of two DOE national Leadership Computing Facilities (the
other is the National Center for Computational Sciences at
Oak Ridge National Laboratory)
 Supports the primary mission of DOE’s Office of Science
Advanced Scientific Computing Research (ASCR) program to
discover, develop, and deploy the computational and
networking tools that enable researchers in the scientific
disciplines to analyze, model, simulate, and predict complex
phenomena important to DOE.
2

3
ALCF Compute and Storage Resources
 Intrepid - ALCF Blue Gene/P System:
– 40,960 nodes / 163,840 PPC cores
– 80 Terabytes of memory
– Peak flop rate: 557 Teraflops
– Linpack flop rate: 450.3
– #9 on the Top500 list
 Eureka - ALCF Visualization System:
– 100 nodes / 800 2.0 GHz Xeon cores
– 3.2 Terabytes of memory
– 200 NVIDIA FX5600 GPUs
– Peak flop rate: 100 Teraflops
 Storage:
– 6+ Petabytes of disk storage with an I/O rate of 80 GB/s (GPFS and
PVFS)
– 5+ Petabytes of archival storage, 10,000 volume tape archive
(HPSS)

4 @ 10 Gig
ALCF Compute Resources
4 4
Surveyor (Dev)
1 rack/4k cores
13.9TF
Intrepid
40 racks/160k cores
557 TF
Networks
(via ESnet, Internet2
UltraScienceNet)
/gpfs/home 105TB
Rate: 8+ GB/s
Switch
128TB
Rate: 2+ GB/s
Tape Library 5PB
6500 LT04 @ 800GB each
24 drives @ 120 MB/s each
I/O
I/O
SwitchComplex
/intrepid-fs0 (GPFS) 3PB
/intrepid-fs1 (PVFS) 2PB
Rate: 60+ GB/s
(4) DDN 9550 - 16 file servers
640 @ 10 Gig
16 @ 10 Gig
Eureka (Viz)
100 nodes/800 cores
200 NVIDIA GPUs
100 TF
Gadzooks (Viz)
4 nodes/32 cores
100 @ 10 Gig

Variety of Allocation Programs
6

7
DOE INCITE Program
Innovative and Novel Computational Impact on Theory and Experiment
 Solicits large computationally intensive research projects
– To enable high-impact scientific advances
– Call for proposal opened once per year (call closed 6/30/2010)
– INCITE Program web site: www.er.doe.gov/ascr/incite
 Open to all scientific researchers and organizations
– Scientific Discipline Peer Review
– Computational Readiness Review
 Provides large computer time & data storage allocations
– To a small number of projects for 1-3 years
– Academic, Federal Lab and Industry, with DOE or other support
 Primary vehicle for selecting principal science projects for the Leadership
Computing Facilities
– 60% of time at Leadership Facilities
 In 2010, 35 INCITE projects allocated more than 600M CPU hours at the ALCF

DOE ALCC Program
ASCR Leadership Computing Challenge
 Allocations for projects of special interest to DOE with an emphasis on high risk,
high payoff simulations in areas of interest to the department’s energy mission
– 30% of the core hours at Leadership Facilities
 Awards granted in June, 2010
– 10 awards at ALCF in 2010 for 300+ million core hours
 ALCC applications submitted from August 1, 2010 through February 15, 2011 will
be considered for allocation in 2011
– http://www.science.doe.gov/ascr/facilities/alcc.html
8

Director’s Discretionary Allocations
 Time is available for projects without INCITE or ALCC
allocations!
 ALCF Discretionary allocations provide time for:
– Porting, scaling, and tuning applications
– Benchmarking codes and preparing INCITE proposals
– Preliminary science runs prior to an INCITE award
– Early Science Program
 To apply go to the ALCF allocations page
– www.alcf.anl.gov/support/gettingstarted
9

10
• Startup assistance
• User administration assistance
• Job management services
• Technical support (Standard and
Emergency)
ALCF
Services
• ALCF science liaison
• Assistance with proposals, planning,
reporting
• Collaboration within science domains
• Performance engineering
• Application tuning
• Data analytics
• Data management services
• Workshops & seminars
• Customized training programs
• On-line content & user guides
• Educational and industry outreach
programs
ALCF Service Offerings

Programming Models and Development Environment
 Languages:
– Full language support with IBM XL and GNU compilers
– Languages: Fortran, C, C++, Python
 MPI:
– Based on MPICH2 1.0.x base code:
• MPI-IO supported
• One-sided communication supported
• No process management (MPI_Spawn(), MPI_Connect(), etc)
– Utilizes the 3 different BG/P networks for different MPI functions
 Threads:
– OpenMP 2.5
– NPTL Pthreads
 Linux development environment:
– Compute Node Kernel provides look and feel of a Linux environment
• POSIX routines (with some restrictions: no fork() or system())
• BG/P adds pthread support, additional socket support
– Supports statically and dynamically linked libraries (static is default)
– Cross compile since login nodes and compute nodes have different processor & OS

Performance Engineering Team
The mission of the Performance Engineering group is to help ALCF users achieve the
best performance out of their applications. To this end, ALCF performance
engineers work closely with the users in porting, tuning and parallelizing their
applications on ALCF computers. They also assist in resolving performance-
inhibiting I/O issues.
ALCF Performance Engineers have extensive experience in:
 Porting, performance tuning and parallelizing of scientific applications and other
software
 Computer architectures
 Computational algorithms
 I/O
12

Supported Libraries and Programs
Program Location Description
TotalView /soft/apps/totalview-8.5.0-0 Multithreaded, multiprocess source code debugger for high
performance computing.
Coreprocessor /soft/apps/coreprocessor.pl A tool to debug and provide postmortem analysis of dead applications.
TAU-2.17 /soft/apps/tau A portable profiling and tracing toolkit for performance analysis of
parallel programs written in Fortran, C++, and C
HPCT /soft/apps/hpct_bgp MPI profiling and tracing library, which collects profiling and tracing
data for MPI programs.
13
Program Location Description
armci /bgsys/drivers/ppcfloor/comm The Aggregate Remote Memory Copy (ARMCI) library
HDF5 /soft/apps/hdf5-1.6.6 The Hierarchical Data Format (HDF) is a model for managing and
storing data.
NetCDF /soft/apps/netcdf-3.6.2 A set of software libraries and machine-independent data formats
that supports the creation, access, and sharing of array-oriented
scientific data.
Parallel NetCDF /soft/apps/parallel-netcdf-
1.0.2
A library providing high-performance I/O while still maintaining file-
format compatibility with Unidata's NetCDF.
mercurial-0.9.5 /soft/apps/mercurial-0.9.5 A distributed version-control system
Scons /soft/apps/scons-0.97 A cross-platform substitute for the classic Make utility
tcl-8.4.14 /soft/apps/tcl-8.4.14 A dynamic programming language,

Catalyst Team
The Catalyst team provides key projects with an assigned expert, the “Catalyst,” to
maximize and accelerate research. Catalysts are computational scientists that are
experts in their fields: computational chemistry, physics, fluid dynamics,
astrophysics, etc., and actively participate in the research projects.
In addition to diagnosing problems using their knowledge of the scientific
calculations being done, Catalysts also provide:
14
 Assistance with algorithm development and
scaling efforts
A single point of contact for key research
initiatives
Tailored services for unique project requirements
Full project lifecycle assistance
Value-added services and support in conjunction
with ALCF resources

Data Analytics and Visualization Team
The Data Analytics and Visualization team has expertise in tools and methods for high-
performance post processing of large datasets, interactive data exploration, batch
visualization, and production visualization.
Team members help users with their visualization and analysis needs using ALCF high-
performance resources and a suite of tools maintained for these purposes.
 Production tools for high performance visualization (ParaView, VisIt)
 Analysis tools (R, MATLAB)
 Presentation graphics (PowerPoint, Keynote, Final Cut Pro)
 The ALCF Data Analytics and Visualization team has strong connections to Argonne’s
Mathematics and Computer Science research and development in the area of
visualization and analysis.
15

Operations Team
16
The ALCF Operations team consists of the Systems Group and the Advanced
Integration Group.
The Systems Group is responsible for:
Hardware maintenance
Software maintenance
Resolution of user tickets related to system issues
Responding to requests for new software or versions of software
Developing systems tools, particularly ones related to the unique system
architectures and scale of ALCF resources
The Advanced Integration Group is responsible
for:
Ensuring the entire system software stack
works together
Assisting with I/O performance issues
Bug fixes and feature requests for systems
software

User Services and Outreach
The USO team provides frontline services and support to existing and potential ALCF
users. The team also provides marketing and outreach to users, DOE and the broader
community.
USO provides the following services:
 Account/Project Administration
– Account management, foreign national clearance, secure access
 Training and Education
– Documentations, workshops
 Help Desk
– Problem resolution, User Communications
 User Outreach
– Tours, Classes, Conferences
17

Getting Access to ALCF Resources and Services

More Related Content

Viewers also liked (8)

Similar to Getting Access to ALCF Resources and Services (20)

Recently uploaded (20)

Getting Access to ALCF Resources and Services