B07.01 HPCMP-Unclassified-Overview-Distro-A Slide Deck 2.pdf

PDF 3 MB Posted

Attached to
Post Industry Day Notice HITS-UIII Federal contract opportunity
Solicitation number
PANERD25P0000_001596
Issued by
Department of the Army Corps of Engineers Engineer Research and Development Center

About this file

This is a presentation document outlining the High Performance Computing Modernization Program (HPCMP), a Department of Defense initiative.

The HPCMP, initiated in 1993 under Congressional mandate, is a department-wide program administered by the U.S. Army Engineer Research and Development Center in Vicksburg, Mississippi, on behalf of Army ASLT with joint service execution across the Air Force, Army, Navy, and department agencies including DTRA, MDA, and DARPA. The program's mission is to accelerate the development and transition of warfighting capabilities through advanced computing, networking, software, security, and computational expertise, supporting approximately 33,000 Department scientists and engineers. Major functions include five DoD Supercomputing Resource Centers (DSRCs) located at Wright-Patterson Air Force Base (Ohio), Aberdeen Proving Ground (Maryland), Vicksburg (Mississippi), Stennis Space Center (Mississippi), and Kihei, Maui (Hawaii), providing large-scale systems with capabilities ranging from 4.2 to 17.65 peak PFLOPS. The program operates the Defense Research and Engineering Network (DREN), a high-bandwidth, low-latency wide-area network with a 100 Gbps dedicated backbone connecting 238 DREN sites and 112 SDREN sites with aggregate site bandwidth of 1,741 Gbps.

The program delivers Computational Research and Engineering Acquisition Tools and Environments (CREATE) software for design, modeling, analysis, and validation across multiple domains (CREATE-AV for aircraft, CREATE-GV for ground vehicles, CREATE-SH for ships, CREATE-RF for radio frequency, and CREATE-FT for finite element analysis). Productivity Enhancement and Training (PET) resources provide access to computational subject matter experts through training, mission projects, outreach initiatives, and special projects designed to improve software scaling, portability, and functionality. Access to HPCMP resources is available to Department researchers and engineers, Department contractors, university staff working on Department research grants, and non-U.S. citizens under specified conditions. The program is critical to Department of Defense operations, enabling improvements to over 50 Major Defense Acquisition Programs and supporting development of next-generation multi-domain operational capabilities while complementing iterative developmental and operational testing.

View the file

Other files for this federal contract opportunity

On GovTribe

Work with this file on GovTribe

  • Download the original file
  • Contacts named in this file
  • Similar government files
  • Ask GovTribe AI about this file

Text version

High Performance Computing Modernization Program

Kelly Dalton

Director, HPCMP

Unclassified

The Agenda

• Our Mission and Resources

• CREATE

• PET and Available Training

• Workforce Development

• Additional Resources and Strategic Vector

• Q&A

• The High Performance Computing Modernization Program (HPCMP) initiated in 1993 in response to Congressional mandate

• Mandate: Advance the supercomputing capability for the DoD, provide a national research network, provide software and computational science expertise and support to better enable DoD RDT&E activities

• Department-wide program administered by U.S. Army Engineering Research & Development Center (Vicksburg, MS) on behalf of Army ASLT

• Joint Service execution

• Substantial “in kind” funding from services

HPC Modernization Program Supports Air Force, Army, Navy, and Department Agencies

High Performance Computing Modernization Program

Accelerating technology development

Enabling new warfighting capabilities

Supporting the Department’s 33,000 scientists and engineers

What is the Mission?

To accelerate the development and transition of warfighting capabilities through the application of advanced computing, networking, software, security, and computational expertise.

ADVANCED

SUPERCOMPUTERS

AGILE

NETWORKS

WORLD-CLASS

SOFTWARE

CYBERSECURITY

TRAINING & COMPUTATIONAL EXPERTISE

WORKFORCE DEVELOPMENT

USERS

SUPERIOR

WARFIGHTING OUTCOMES

U.S. Air Force Research Laboratory

DSRC

U.S. Army Research

Laboratory

DSRC

U.S. Army Engineer Research and Development Center DSRC

Maui High Performance

Computing Center DSRC

U.S. Navy

DSRC

Defense Research & Engineering Network

(DREN)

Computer Network Defense, Security R&D, and Security Integration

Core Software

Computational Environments

Education and Training

HPC User Support

Hardware + Network + Software Science and Technology

Major Functions

Army Navy

Air Force

DTRA

MDA

DARPA

Services & Agencies

Federal Agencies + PEO’s/PM’s + Academia + Industry Partners

Test and Evaluation

Acquisition Engineering

DoD Supercomputing Resource Centers (DSRCs) Networking and Security Software Applications

A program and national asset providing high-performance computing capabilities and expertise to solve our most critical mission challenges

DoD Supercomputing Resource Centers (DSRCs)

Large Scale Systems, Breadth of Capabilities HPE Cray EX4000 system 1,092 standard compute nodes 4 large-memory nodes 12 1-GPU visualization nodes Total of 212,736 compute cores 382 TB of usable memory; Rated at 8.2 peak PFLOPS

HPE Cray EX4000 system 1,280 standard compute nodes 16 large-memory nodes 32 AI/ML nodes 24 visualization nodes Total of 256,512 compute cores

Liqid system 494 standard compute nodes 24 large-memory nodes 36 AI/ML nodes 24 visualization nodes.

HPE Cray EX4000 1,632 standard compute nodes 4 large-memory nodes 8 GPU nodes Total of 313,344 compute cores 585 TB of useable memory; Rated at 17.65 peak PFLOPS

Penguin Computing TrueHPC system 1,384 standard compute nodes, 16 large-memory nodes 16 visualization accelerated nodes, 32 AI/ML nodes 32 High Core Performance nodes Total of 1,480 compute nodes or 186,368 compute cores 386 TB of memory; Rated at 8.5 peak PFLOPS

HPE Cray EX system 2,304 standard compute nodes; 26 large-memory nodes 16 visualization accelerated nodes 32 Single-GPU MLA accelerated nodes 32 Dual-GPU MLA accelerated nodes 2,410 compute nodes 308,480 compute cores 640 TB of memory; Rated at 13.5 peak PFLOPS.

Penguin Computing TrueHPC 1,480 standard compute nodes, 8 large-memory nodes 24 Visualization nodes, 32 MLA nodes 64 High Clock nodes Total of 199,680 compute cores 447 TB of memory and is rated at 9 peak PFLOPS

HPE Cray EX system 640 standard compute nodes 4 large-memory nodes 16 AI/ML nodes 24 visualization nodes

HPE Cray EX system 1,024 standard compute nodes 4 large-memory nodes 24 1-GPU visualization nodes, 40 2-GPU Machine-Learning nodes Total of 1,092 compute nodes or 139,776 compute cores 564 TB of memory; Rated at 6.86 peak PFLOPS

Liqid system 770 standard compute nodes 7 large-memory nodes 16 1-GPU visualization nodes, 68 4-GPU MLA nodes 28 6-GPU MLA nodes (a total of 81,502 compute cores). It has 376 TB of useable memory and is rated at 4.2 PFLOP

0 Mbps

1-9 Gbps

10-24 Gbps

25-49 Gbps

50-99 Gbps

100-199 Gbps

200-299 Gbps

300 + Gbps

Defense Research and Engineering Network (DREN)

• Department’s premier wide area network – for S&T, T&E and Acquisition Engineering – with separation from Warfighting networks

• High-bandwidth, low-latency, low-jitter, full-service network fully supporting IPv6

• Connects high performance computing (HPC) centers and users

• Secret-level network overlay on DREN backbone (SDREN)

• DREN 4 provides 1 Gbps to 100 Gbps service to DoD sites; across a 100 Gbps dedicated backbone

1 Gbps 10 Gbps

Site Speeds Backbone

100 Gbps

DSRC

Network # of Sites

DREN 238

SDREN 112

As of Jun 2024

Aggregate Site Bandwidth Ordered: 1,741 Gbps

Robust, Scalable, Reliable

…built to drive exceptional performance

Powerful Software Tools Built for a Purpose

Stakeholder Requirements

Design Model and Analyze Build and Validate Model

Maintenance and Sustainment

Refine and Optimize

Computational Research and Engineering Acquisition Tools and Environments

Computational Research & Engineering Acquisition

Tools and Environments

(CREATE)

CREATE Simulation: NavyFOAM

CREATE Simulation: Explosive Modeling

Above Surface Below Surface

Productivity Enhancement and Training (PET)

Four Elements of Support Provide Provide access to computational

Subject Matter Experts

Enable Enable HPCMP users to become more productive by effectively employing HPCMP resources

Improve Improve the scaling, portability, accuracy and functionality of application software on a variety of hardware platforms

Build Build awareness of and promote transfer of scientific and technical HPC capabilities to reduce cost and schedule of DoD priorities

PET Resources Training Provides access to HPC training resources and events in applicable tools/technologies using the latest collaboration, distance learning, and training technologies

Mission Projects (MPs)

Designed to assist users with solving complex issues in mission-critical areas

Outreach Cultivates and maintains collaborative relationships with enterprise-wide stakeholders to improve HPCMP productivity enhancement

Special Projects (SPs)

Enables collaboration between to solve HPCMP and mission-related problems

How to Access

HPCMP

Resources

Department researchers and engineers

Department contractors

University staff working on a Department research grant

Non-US Citizens may have access to HPCMP resources

HPCMP is Critical to DoW

• Enabling improvements to > 50 Major Defense Acquisition Programs and existing weapons platforms

• Supporting the development of next-generation platforms and capabilities for multi-domain operations

• Complementing iterative developmental and operational testing

File details come from the government source that posted it. Updated .