Search NASA⌕ Search

SEARCH · Search NASA

Results for “MPI Sessions”

Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.

Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.

MPI Sessions: Second Demonstration and Evaluation of MPI Sessions Prototype

MPI Sessions presents a new paradigm for applications to use MPI functionality within an application. This paradigm offers the promise of more flexible ways for applications to express the capabilities they require from MPI at a fine grain level. This report presents a second evaluation of a prototype implementation of the MPI Sessions proposal, including additional development work on the prototype and investigations using Sessions in the DASK data analytics framework.

97 MATHEMATICS AND COMPUTING↗

MPI Session: External Network Transport Implementation (V.1.0)

The MPI Sessions extensions to the MPI standard have been accepted by the MPI Forum and will be included in the upcoming MPI 4 version of the standard. MPI Sessions has the potential to address several limitations of MPI’s current specification: MPI cannot be initialized within an MPI process from different application components without a priori knowledge or coordination; MPI cannot be initialized more than once; and, MPI cannot be reinitialized after MPI finalization. MPI Sessions also offers the possibility for more flexible ways for individual components of an application to express the capabilities they require from MPI at a finer granularity than is presently possible. A prototype of MPI Sessions, based on the Open MPI implementation of the MPI standard, was developed to facilitate acceptance of the Sessions proposal by the Forum. The initial implementation had some limitations, one of the more significant ones being that it was limited in its ability to fully exploit modern network APIs such as OFI libfabric and OpenUCX and underlying network hardware. This report presents enhancements to the prototype implementation of MPI Sessions that removes this restriction for the networks to be used in the next generation of DOE exa–scale systems. Open MPI was used as the implementation vehicle, but results here are also relevant to other middleware stacks.

97 MATHEMATICS AND COMPUTING↗

VerifyIO: Verifying Adherence to Parallel I/O Consistency Semantics

VerifyIO is a tool designed for verifying I/O consistency semantics in High-Performance Computing (HPC) applications. It addresses the challenges of ensuring correctness and portability across different I/O consistency models, such as POSIX, Commit, Session, and MPI-IO. By analyzing execution traces, detecting conflicts, and verifying synchronization adherence, VerifyIO provides actionable insights for both application developers and I/O library designers.

Wang, Chen [Lawrence Livermore National Laboratory↗

pFUnit 3.0 Tutorial Advanced

This tutorial will introduce Fortran developers to unit-testing and test-driven development (TDD) using pFUnit. As with other unit-testing frameworks, pFUnit, simplifies the process of writing, collecting, and executing tests while providing clear diagnostic messages for failing tests. pFUnit specifically targets the development of scientific-technical software written in Fortran and includes customized features such as: assertions for multi-dimensional arrays, distributed (MPI) and thread-based (OpenMP) parallellism, and flexible parameterized tests.These sessions will include numerous examples and hands-on exercises that gradually build in complexity. Attendees are expected to have working knowledge of F90, but familiarity with object-oriented syntax in F2003 and MPI will be of benefit for the more advanced examples. By the end of the tutorial the audience should feel comfortable in applying pFUnit within their own development environment.

Test Driven using pFUnit↗

Taking the MPI standard and the open MPI library to exascale

The Open MPI for Exascale (OMPI-X) project was one of two in the Exascale Computing Project (ECP) focused on advancing the MPI ecosystem. The OMPI-X team worked with other MPI Forum members to champion several important features for inclusion in the MPI 4.0, 4.1, and upcoming 5.0 MPI standard versions, in support of the needs of exascale applications and systems. The team also worked with the larger Open MPI community to bring implementations of these new features and other enhancements into Open MPI, one of the leading open-source implementations of the MPI interface. Here, this paper describes the motivation for the work of the OMPI-X project in the context of exascale computing needs, the nature of the resulting new capabilities in the MPI standard, and how they were implemented in the Open MPI library. Features include improved support for “MPI + X” programming models through partitioned communications and support for user-level threading, sessions, fault tolerance through the user-level fault mitigation (ULFM) and Reinit models, and other features. We also discuss enhancements to Open MPI providing improved performance and scalability for existing features, such as collective operations, one-sided operations, support for the Slingshot-11 interconnect of the initial exascale systems, and how the OMPI-X team worked to improve quality assurance for the Open MPI library, particularly on platforms of interest to the Department of Energy community.

97 MATHEMATICS AND COMPUTING↗

Workshop on Extraterrestrial Materials from Cold and Hot Deserts

The workshop was held July 6-8, 1999 before the Meteoritical Society meeting in Johannesburg, South Africa. The venue was Kwa Maritane Resort in the Pilanesburg Game Reserve. Conveners were Ludolf Schultz (Chair, MPI fur Chemie), Ian Franchi (Open University), Arch Reid (University of Houston), and Mike Zolensky (NASA JSC). Extended abstracts will be published as an LPI Technical Report. In the first session, Marvin discussed three African iron meteorites: Cape of Good Hope, Gibeon, and Hoba. Grady presented a statistical analysis of meteorites from hot and cold deserts. Wasson discussed types of Antarctic iron meteorites. Several presentations characterized populations of meteorites from individual desert areas: Libyan Desert (Weber et al.), Nullarbor Region (Bevan et al.), and Mojave Desert (Kring et al. and Verish et al). Pairing among EET87503-group howardites was discussed by Buchanan et al. Based on 14C terrestrial ages of Allan Hills ordinary chondrites, Bland et al. suggested that ice flow may be the principal sink for Antarctic meteorites. The effects of preterrestrial and terrestrial alteration were considered in the second session. Nakamura et al. and Lipschutz discussed asteroidal metamorphism of carbonaceous chondrites. Zolensky presented evidence for preterrestrial halide and sulfide in meteorites. Crozaz and Wadhwa described terrestrial alteration of Dar al Gani 476. Welten and Nishiizumi discussed terrestrial weathering of chondrites from Frontier Mountain, Antarctica. Most of the third session dealt with terrestrial meteorite ages. Based on 14C-10Be ages, Jull et al. discussed the exponential decay in numbers of meteorites with increased age. Nishiizumi et al. concluded that some Allan Hills meteorites have much older terrestrial ages than any meteorites from Lewis Cliffs. Welten et al. discussed terrestrial ages determined by 41Ca/36CI of metal separates from hot desert meteorites. Based on a comparison with large IDPs, Flynn et al. suggested that polar micrometeorites, lost a water-soluble sulfate phase by terrestrial alteration. Nyquist suggested that Type I cosmic spheres from deep sea sediments and polar ice were derived from carbonaceous chondrite-like asteroidal sources. The final session considered noble gases, cosmic ray effects, and thermoluminescence. Calculations presented by Reedy indicate that cosmic ray-produced nuclides are more likely to be preserved in small objects than in larger objects. Murty and Mohapatra. reported that trapped gas in Dar al Gani 476 includes a Martian atmospheric component. Wieler et al. and Scherer et al. reported noble gas abundances in different types of desert meteorites. Patzer and Schultz discussed the influence of terrestrial weathering on cosmic ray exposure ages of enstatite chondrites. Franchi et al. suggested that it is difficult to discriminate whether differences in gas release profiles of lunar meteorites from hot deserts and returned lunar samples are the result of terrestrial weathering or shock metamorphism. Benoit and Sears surveyed natural and induced thermoluminescence of Antarctic ordinary chondrites. Merchel et al. analyzed Saharan meteorites with short or complex exposure histories.

Buchanan, Paul C.↗

Foreign Travel Trip Report for LLNL travel with DOE FES funding, May 19th-30th, 2012

I attended the 20th biannual International Conference on Plasma Surface Interaction (PSI) in Fusion Devices in Aachen, Germany, hosted this year by the Forschungszentrum Julich (FZJ) research center. The PSI conference is one of the main international forums for the presentation and discussion of results on plasma surface interactions and edge plasma physics relevant to magnetic confinement fusion devices. I disseminated the recent results of FESP/LLNL tokamak research by presenting three posters on: (i) understanding reconnection and controlling edge localized modes (ELMs) using the BOUT++ code, (ii) simulation of resistive ballooning mode turbulence, and (iii) innovative design of Snowflake divertors. I learned of many new and recent results from international tokamak facilities and had the opportunity for discussion of these topics with other scientists at the poster sessions, conference lunches/receptions, etc. Some of the major highlights of the PSI conference topics were: (1) Review of the progress in using metallic tungsten and beryllium (ITER-like) walls at international tokamak facilities: JET (Culham, UK), TEXTOR (FZJ, Germany) and Alcator CMOD (MIT, USA). Results included: effect of small and large-area melting on plasma impurity content and recovery, expected reduction in retention of hydrogenic species, increased heat load during disruptions and need for mitigation with massive gas injection. (2) A review of ELM control in general (T. Evans, GA) and recent results of ELM control using n=2 external magnetic perturbations on ASDEX-Upgrade (MPI-Garching, Germany). (3) General agreement among the international tokamak database that, along the outer midplane of a low collisionality tokamak, the SOL power width in current experiments varies inversely with respect to plasma current (Ip), roughly as 1/Ip, with little dependence on other plasma parameters. This would imply roughly a factor of 1/4 of the width that was assumed for the design of the ITER tokamak. The first studies of the implications for ITER (A. Kukushkin, ITER) have shown a great reduction in operational parameter space that, at present, can only be lifted by increasing target plate heat flux limits. During my visit to the CRPP at the EPFL, I delivered an invited talk in order to disseminate new results of the recent publication [1] on using non-axisymmetric perturbations of the SOL to control the edge plasma. I was given a tour of both the TCV tokamak and the TORPEX simple magnetized plasma device/divertor simulator. TORPEX is an excellent laboratory for exploring the physics of simple magnetized plasmas that are relevant to the scrape-off layer of a tokamak. Properly designed experiments on TORPEX can potentially be used to test the theory of controlling the edge plasma using non-axisymmetric potentials and currents in the SOL developed by LLNL described in [1].

70 PLASMA PHYSICS AND FUSION TECHNOLOGY↗

Roofline Analysis in the Intel® Advisor to Deliver Optimized Performance for applications on Intel® Xeon Phi™ Processor

In this session we show, in two case studies, how the roofline feature of Intel Advisor has been utilized to optimize the performance of kernels of the XGC1 and PICSAR codes in preparation for Intel Knights Landing architecture. The impact of the implemented optimizations and the benefits of using the automatic roofline feature of Intel Advisor to study performance of large applications will be presented. This demonstrates an effective optimization strategy that has enabled these science applications to achieve up to 4.6 times speed-up and prepare for future exascale architectures. # Goal/Relevance of Session The roofline model [1,2] is a powerful tool for analyzing the performance of applications with respect to the theoretical peak achievable on a given computer architecture. It allows one to graphically represent the performance of an application in terms of operational intensity, i.e. the ratio of flops performed and bytes moved from memory in order to guide optimization efforts. Given the scale and complexity of modern science applications, it can often be a tedious task for the user to perform the analysis on the level of functions or loops to identify where performance gains can be made. With new Intel tools, it is now possible to automate this task, as well as base the estimates of peak performance on measurements rather than vendor specifications. The goal of this session is to demonstrate how the roofline feature of Intel Advisor can be used to balance memory vs. computation related optimization efforts and effectively identify performance bottlenecks. A series of typical optimization techniques: cache blocking, structure refactoring, data alignment, and vectorization illustrated by the kernel cases will be addressed. # Description of the codes ## XGC1 The XGC1 code [3] is a magnetic fusion Particle-In-Cell code that uses an unstructured mesh for its Poisson solver that allows it to accurately resolve the edge plasma of a magnetic fusion device. After recent optimizations to its collision kernel [4], most of the computing time is spent in the electron push (pushe) kernel, where these optimization efforts have been focused. The kernel code scaled well with MPI+OpenMP but had almost no automatic compiler vectorization, in part due to indirect memory addresses and in part due to low trip counts of low-level loops that would be candidates for vectorization. Particle blocking and sorting have been implemented to increase trip counts of low-level loops and improve memory locality, and OpenMP directives have been added to vectorize compute-intensive loops that were identified by Advisor. The optimizations have improved the performance of the pushe kernel 2x on Haswell processors and 1.7x on KNL. The KNL node-for-node performance has been brought to within 30% of a NERSC Cori phase I Haswell node and we expect to bridge this gap by reducing the memory footprint of compute intensive routines to improve cache reuse. ## PICSAR is a Fortran/Python high-performance Particle-In-Cell library targeting at MIC architectures first designed to be coupled with the PIC code WARP for the simulation of laser-matter interaction and particle accelerators. PICSAR also contains a FORTRAN stand-alone kernel for performance studies and benchmarks. A MPI domain decomposition is used between NUMA domains and a tile decomposition (cache-blocking) handled by OpenMP has been added for shared-memory parallelism and better cache management. The so-called current deposition and field gathering steps that compose the PIC time loop constitute major hotspots that have been rewritten to enable more efficient vectorization. Particle communications between tiles and MPI domain has been merged and parallelized. All considered, these improvements provide speedups of 3.1 for order 1 and 4.6 for order 3 interpolation shape factors on KNL configured in SNC4 quadrant flat mode. Performance is similar between a node of cori phase 1 and KNL at order 1 and better on KNL by a factor 1.6 at order 3 with the considered test case (homogeneous thermal plasma).

97 MATHEMATICS AND COMPUTING↗