skip to main content
OSTI.GOV title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: The Co Design Architecture for Exascale Systems, a Novel Approach for Scalable Designs

Abstract

High performance computing (HPC) has begun scaling beyond the Petaflop range towards the Exaflop (1000 Petaflops) mark. One of the major concerns throughout the development toward such performance capability is scalability both at the system level and the application layer. In this paper we present a novel approach for a new design concept the Co Design approach with enables a tighter development of both the application communication libraries and the underlying hardware interconnect solution in order to overcome scalability issues and to enable a more efficient design approach towards Exascale computing. We have suggested a new application programing interface and have demonstrated a 50x improvement of performance and scalability increases.

Authors:
 [1];  [1];  [2];  [2];  [1];  [1];  [1];  [1];  [1];  [2]
  1. Mellanox Technologies, Inc.
  2. ORNL
Publication Date:
Research Org.:
Oak Ridge National Lab. (ORNL), Oak Ridge, TN (United States)
Sponsoring Org.:
USDOE Office of Science (SC)
OSTI Identifier:
1055032
DOE Contract Number:
DE-AC05-00OR22725
Resource Type:
Conference
Resource Relation:
Conference: International Supercomputing Conference - ISC, Hamburg, Germany, 20120617, 20120617
Country of Publication:
United States
Language:
English

Citation Formats

Kagan, Michael, Shainer, Gilad, Poole, Stephen W, Shamis, Pavel, Wilde, Todd, Pak, Lui, Liu, Tong, Dubman, Mike, Shahar, Yiftah, and Graham, Richard L. The Co Design Architecture for Exascale Systems, a Novel Approach for Scalable Designs. United States: N. p., 2012. Web.
Kagan, Michael, Shainer, Gilad, Poole, Stephen W, Shamis, Pavel, Wilde, Todd, Pak, Lui, Liu, Tong, Dubman, Mike, Shahar, Yiftah, & Graham, Richard L. The Co Design Architecture for Exascale Systems, a Novel Approach for Scalable Designs. United States.
Kagan, Michael, Shainer, Gilad, Poole, Stephen W, Shamis, Pavel, Wilde, Todd, Pak, Lui, Liu, Tong, Dubman, Mike, Shahar, Yiftah, and Graham, Richard L. 2012. "The Co Design Architecture for Exascale Systems, a Novel Approach for Scalable Designs". United States. doi:.
@article{osti_1055032,
title = {The Co Design Architecture for Exascale Systems, a Novel Approach for Scalable Designs},
author = {Kagan, Michael and Shainer, Gilad and Poole, Stephen W and Shamis, Pavel and Wilde, Todd and Pak, Lui and Liu, Tong and Dubman, Mike and Shahar, Yiftah and Graham, Richard L},
abstractNote = {High performance computing (HPC) has begun scaling beyond the Petaflop range towards the Exaflop (1000 Petaflops) mark. One of the major concerns throughout the development toward such performance capability is scalability both at the system level and the application layer. In this paper we present a novel approach for a new design concept the Co Design approach with enables a tighter development of both the application communication libraries and the underlying hardware interconnect solution in order to overcome scalability issues and to enable a more efficient design approach towards Exascale computing. We have suggested a new application programing interface and have demonstrated a 50x improvement of performance and scalability increases.},
doi = {},
journal = {},
number = ,
volume = ,
place = {United States},
year = 2012,
month = 1
}

Conference:
Other availability
Please see Document Availability for additional information on obtaining the full-text document. Library patrons may search WorldCat to identify libraries that hold this conference proceeding.

Save / Share:
  • Energy network optimization (ENO) models identify new strategies for designing, installing, and controlling stationary combined heat and power (CHP) fuel cell systems (FCSs) with the goals of (1) minimizing electricity and heating costs for building owners and (2) reducing emissions of the primary greenhouse gas (GHG) - carbon dioxide (CO{sub 2}). A goal of this work is to employ relatively inexpensive simulation studies to discover more financially and environmentally effective approaches for installing CHP FCSs. ENO models quantify the impact of different choices made by power generation operators, FCS manufacturers, building owners, and governments with respect to two primary goalsmore » - energy cost savings for building owners and CO{sub 2} emission reductions. These types of models are crucial for identifying cost and CO{sub 2} optima for particular installations. Optimal strategies change with varying economic and environmental conditions, FCS performance, the characteristics of building demand for electricity and heat, and many other factors. ENO models evaluate both 'business-as-usual' and novel FCS operating strategies. For the scenarios examined here, relative to a base case of no FCSs installed, model results indicate that novel strategies could reduce building energy costs by 25% and CO{sub 2} emissions by 80%. Part I of II articles discusses model assumptions and methodology. Part II of II articles illustrates model results for a university campus town and generalizes these results for diverse communities.« less
  • Scalable parallel processing is a driving force in high performance computing research. The notion of scalable parallel algorithms continues to elude computational science as a whole. Some problems are trivially parallelized while others have hidden parallelism and pipelining. Von Neumann languages, such as FORTRAN and C, are a limiting factor influencing the performance of parallel software. This paper proposes a software architecture to support the notion of scalable parallel software architecture.
  • Emerging needs in transportation network modeling and simulation are raising new challenges with respect to scal-ability of network size and vehicular traffic intensity, speed of simulation for simulation-based optimization, and fidel-ity of vehicular behavior for accurate capture of event phe-nomena. Parallel execution is warranted to sustain the re-quired detail, size and speed. However, few parallel simulators exist for such applications, partly due to the challenges underlying their development. Moreover, many simulators are based on time-stepped models, which can be computationally inefficient for the purposes of modeling evacuation traffic. Here an approach is presented to de-signing a simulator with memory andmore » speed efficiency as the goals from the outset, and, specifically, scalability via parallel execution. The design makes use of discrete event modeling techniques as well as parallel simulation meth-ods. Our simulator, called SCATTER, is being developed, incorporating such design considerations. Preliminary per-formance results are presented on benchmark road net-works, showing scalability to one million vehicles simu-lated on one processor.« less
  • A multi-channel laser-based chemical sensor platform is presented, in which a modular architecture allows the exchange of complete sensor channels without disruption to overall operation. Each sensor channel contains custom optical and electronics packages, which can be selected to access laser wavelengths, interaction path lengths and modulation techniques optimal for a given application or mission. Although intended primarily to accommodate mid-infrared (MIR) external cavity quantum cascade lasers (ECQCLs)and astigmatic Herriott cells, channels using visible or near infrared (NIR) lasers or other gas cell architectures can also be used, making this a truly versatile platform. Analog and digital resources have beenmore » carefully chosen to facilitate small footprint, rapid spectral scanning, ow-noise signal recovery, failsafe autonomous operation, and in-situ chemometric data analysis, storage and transmission. Results from the demonstration of a two-channel version of this platform are also presented.« less