Engineering topics
Thigpen, William W.
Publications and source records attributed to Thigpen, William W..
Expanding a Supercomputer Facility Using Modular Data Center Technology
With the expansion of high-end computing resources needed to support NASA's increasing demands for physics-based simulations, the facility housing Pleiades-the agency's largest supercomputer-recently reached its power and cooling capacity. In response, the NASA Advanced Supercomputing Division at Ames Research Center undertook a prototype project that resulted in a new facility based on modular data center technology. The facility, a ~1000 square-foot module on a concrete pad with room for 16-18 compute racks, was completed in fall 2016 and an SGI computer system, named Electra, was deployed there in early 2017. Cooling is performed via an evaporative system built into the module, and preliminary experience shows a Power Usage Effectiveness (PUE) of ~1.03. Electra achieved over a petaflop on the LINPACK benchmark, sufficient to rank number 96 on the November 2016 TOP500 list. The system consists of 1,152 InfiniBand-connected Intel Xeon Broadwell-based nodes. Its users access their files on a facility wide file system shared by all compute assets via Mellanox MetroX InfiniBand extenders, which connect the Electra fabric to Lustre routers InfiniBand fabric over fiber-optic links about 300 meters long. The prototype has exceeded expectations and is serving as a blueprint for future expansions.*†
NASA Advanced Supercomputing Facility Expansion
The NASA Advanced Supercomputing (NAS) Division enables advances in high-end computing technologies and in modeling and simulation methods to tackle some of the toughest science and engineering challenges facing NASA today. The name "NAS" has long been associated with leadership and innovation throughout the high-end computing (HEC) community. We play a significant role in shaping HEC standards and paradigms, and provide leadership in the areas of large-scale InfiniBand fabrics, Lustre open-source filesystems, and hyperwall technologies. We provide an integrated high-end computing environment to accelerate NASA missions and make revolutionary advances in science. Pleiades, a petaflop-scale supercomputer, is used by scientists throughout the U.S. to support NASA missions, and is ranked among the most powerful systems in the world. One of our key focus areas is in modeling and simulation to support NASA's real-world engineering applications and make fundamental advances in modeling and simulation methods.
Spaceflight Operations Services Grid (SOSG)
In an effort to adapt existing space flight operations services to new emerging Grid technologies we are developing a Grid-based prototype space flight operations Grid. This prototype is based on the operational services being provided to the International Space Station's Payload operations located at the Marshall Space Flight Center, Alabama. The prototype services will be Grid or Web enabled and provided to four user communities through portal technology. Users will have the opportunity to assess the value and feasibility of Grid technologies to their specific areas or disciplines. In this presentation descriptions of the prototype development, User-based services, Grid-based services and status of the project will be presented. Expected benefits, findings and observations (if any) to date will also be discussed. The focus of the presentation will be on the project in general, status to date and future plans. The End-use services to be included in the prototype are voice, video, telemetry, commanding, collaboration tools and visualization among others. Security is addressed throughout the project and is being designed into the Grid technologies and standards development. The project is divided into three phases. Phase One establishes the baseline User-based services required for space flight operations listed above. Phase Two involves applying Gridlweb technologies to the User-based services and development of portals for access by users. Phase Three will allow NASA and end users to evaluate the services and determine the future of the technology as applied to space flight operational services. Although, Phase One, which includes the development of the quasi-operational User-based services of the prototype, development will be completed by March 2004, the application of Grid technologies to these services will have just begun. We will provide status of the Grid technologies to the individual User-based services. This effort will result in an extensible environment that incorporates existing and new spaceflight services into a standards-based framework providing current and future NASA programs with cost savings and new and evolvable methods to conduct science. This project will demonstrate how the use of new programming paradigms such as web and grid services can provide three significant benefits to the cost-effective delivery of spaceflight services. They will enable applications to operate more efficiently by being able to utilize pooled resources. They will also permit the reuse of common services to rapidly construct new and more powerful applications. Finally they will permit easy and secure access to services via a combination of grid and portal technology by a distributed user community consisting of NASA operations centers, scientists, the educational community and even the general population as outreach. The approach will be to deploy existing mission support applications such as the Telescience Resource Kit (TReK) and new applications under development, such as the Grid Video Distribution System (GViDS), together with existing grid applications and services such as high-performance computing and visualization services provided by NASA s Information Power Grid (IPG) in the MSFC s Payload Operations Integration Center (POIC) HOSC Annex. Once the initial applications have been moved to the grid, a process will begin to apply the new programming paradigms to integrate them where possible. For example, with GViDS, instead of viewing the Distribution service as an application that must run on a single node, the new approach is to build it such that it can be dispatched across a pool of resources in response to dynamic loads. To make this a reality, reusable services will be critical, such as a brokering service to locate appropriate resource within the pool. This brokering service can then be used by other applications such as the TReK. To expand further, if the GViDS application is constructed using a services-based mel, then other applications such as the Video Auditorium can then use GViDS as a service to easily incorporate these video streams into a collaborative conference. Finally, as these applications are re-factored into this new services-based paradigm, the construction of portals to integrate them will be a simple process. As a result, portals can be tailored to meet the requirements of specific user communities.
Spaceflight Operations Services Grid (SOSG) Prototype Implementation and Feasibility Study
Science Operations Services Grid is focusing on building a prototype grid-based environment that incorporates existing and new spaceflight services to enable current and future NASA programs with cost savings and new and evolvable methods to conduct science in a distributed environment. The Science Operations Services Grid (SOSG) will provide a distributed environment for widely disparate organizations to conduct their systems and processes in a more efficient and cost effective manner. These organizations include those that: 1) engage in space-based science and operations, 2) develop space-based systems and processes, and 3) conduct scientific research, bringing together disparate scientific disciplines like geology and oceanography to create new information. In addition educational outreach will be significantly enhanced by providing to schools the same tools used by NASA with the ability of the schools to actively participate on many levels in the science generated by NASA from space and on the ground. The services range from voice, video and telemetry processing and display to data mining, high level processing and visualization tools all accessible from a single portal. In this environment, users would not require high end systems or processes at their home locations to use these services. Also, the user would need to know minimal details about the applications in order to utilize the services. In addition, security at all levels is an underlying goal of the project. The Science Operations Services Grid will focus on four tools that are currently used by the ISS Payload community along with nine more that are new to the community. Under the prototype four Grid virtual organizations PO) will be developed to represent four types of users. They are a Payload (experimenters) VO, a Flight Controllers VO, an Engineering and Science Collaborators VO and an Education and Public Outreach VO. The User-based services will be implemented to replicate the operational voice, video, telemetry and commanding systems. Once the User-based services are in place, they will be analyzed to establish feasibility for Grid enabling. If feasible then each User-based service will be Grid enabled. The remaining non-Grid services if not already Web enabled will be so enabled. In the end, four portals will be developed one for each VO. Each portal will contain the appropriate User-based services required for that VO to operate.
OpenMP Experiences and Comparisons
The document discusses message passage interfaces (MPI), open message passage issues and parallelization. A comparison of the two vector platforms, C90 and SVlex is presented. Sections of the document are written in computer code.
Code Optimization and Parallelization on the Origins: Looking from Users' Perspective
Parallel machines are becoming the main compute engines for high performance computing. Despite their increasing popularity, it is still a challenge for most users to learn the basic techniques to optimize/parallelize their codes on such platforms. In this paper, we present some experiences on learning these techniques for the Origin systems at the NASA Advanced Supercomputing Division. Emphasis of this paper will be on a few essential issues (with examples) that general users should master when they work with the Origins as well as other parallel systems.
Using Modules with MPICH-G2 (and "Loose Ends")
A new approach to running complex, distributed MPI jobs using the MPICH-G2 library is described. This approach allows the user to switch between different versions of compilers, system libraries, MPI libraries, etc. via the "module" command. The key idea is a departure from the prescribed "(jobtype=mpi)" approach to running distributed MPI jobs. The new method requires the user to provide a script that will be run as the "executable" with the "(jobtype=single)" RSL attribute. The major advantage of the proposed method is to enable users to decide in their own script what modules, environment, etc. they would like to have in running their job.
Pricing the Computing Resources: Reading Between the Lines and Beyond
Distributed computing systems have the potential to increase the usefulness of existing facilities for computation without adding anything physical, but that is realized only when necessary administrative features are in place. In a distributed environment, the best match is sought between a computing job to be run and a computer to run the job (global scheduling), which is a function that has not been required by conventional systems. Viewing the computers as 'suppliers' and the users as 'consumers' of computing services, markets for computing services/resources have been examined as one of the most promising mechanisms for global scheduling. We first establish why economics can contribute to scheduling. We further define the criterion for a scheme to qualify as an application of economics. Many studies to date have claimed to have applied economics to scheduling. If their scheduling mechanisms do not utilize economics, contrary to their claims, their favorable results do not contribute to the assertion that markets provide the best framework for global scheduling. We examine the well-known scheduling schemes, which concern pricing and markets, using our criterion of what application of economics is. Our conclusion is that none of the schemes examined makes full use of economics.
Towards Scalable 1024 Processor Shared Memory Systems
Over the past 3 years, NASA Ames has been involved in a cooperative effort with SGI to develop the largest single system image systems available. Currently a 1024 Origin3OOO is under development, with first boot expected later in the summer of 2001. This paper discusses some early results with a 512p Origin3OOO system and some arcane IRIX system calls that can dramatically improve scaling performance.
Comparison of 250 MHz R10K Origin 2000 and 400 MHz Origin 2000 Using NAS Parallel Benchmarks
This report describes results of benchmark tests on Steger, a 250 MHz Origin 2000 system with R10K processors, currently installed at the NASA Ames National Advanced Supercomputing (NAS) facility. For comparison purposes, the tests were also run on Lomax, a 400 MHz Origin 2000 with R12K processors. The BT, LU, and SP application benchmarks in the NAS Parallel Benchmark Suite and the kernel benchmark FT were chosen to measure system performance. Having been written to measure performance on Computational Fluid Dynamics applications, these benchmarks are assumed appropriate to represent the NAS workload. Since the NAS runs both message passing (MPI) and shared-memory, compiler directive type codes, both MPI and OpenMP versions of the benchmarks were used. The MPI versions used were the latest official release of the NAS Parallel Benchmarks, version 2.3. The OpenMP versions used were PBN3b2, a beta version that is in the process of being released. NPB 2.3 and PBN3b2 are technically different benchmarks, and NPB results are not directly comparable to PBN results.