CEED-MS34: Improve performance and capabilities of CEED-enabled ECP applications on Summit/Sierra. Exascale Computing Project Milestone Report

Kolev, Tzanio; Fischer, Paul; Abdelfattah, Ahmad; Ananthan, Shreyas; Barra, Valeria; Beams, Natalie; Bleile, Ryan; Brown, J. L.; Carson, Robert; Camier, Jean-Sylvain; Churchfield, Matthew; Dobrev, Veselin; Dongarra, Jack; Dudouit, Yohann; Karakus, Ali; Kerkemeier, Stefan; Lan, Yu-Hsiang; Medina, David; Merzari, Elia; Min, Misun; Parker, Scott; Ratnayaka, Thilina; Smith, Craig F.; Sprague, Michael; Stitt, Thomas; Thompson, Jeremy; Tomboulides, Ananias; Tomov, Stanimire; Tomov, Vladimir; Vargas, Anielle de; Warburton, Tim; Weiss, Kenneth

doi:10.2172/1614965

Cited by 4 publications

(13 citation statements)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The use of the batch BLAS operations increases the chances of performance portability, since BLAS is often highly optimized by vendors and other open source numerical software. This was recently illustrated with the MAGMA port and CEED backend for AMD GPUs (Brown et al, 2020a, 2020b; Kolev et al, 2020).…”

Section: High-order Software Ecosystemmentioning

confidence: 95%

“…For the unsteady ExaSMR simulations, the GPU-based NekRS code has made significant advances in development. For the 17×17 rod-bundle in Figure 17(a), we have demonstrated improved NekRS simulation capabilities to extend to largest problem size to date, with 175 million spectral elements (n=61 billion grid points) using 3520 nodes (21120 V100s) on Summit as well as strong and weak scaling performance at large scale in (Kolev et al, 2020; Tomov et al, 2019) (see Figure 14).…”

Section: Application Integrationmentioning

confidence: 95%

“…The outcomes of the CEED technologies have been integrated into the open source codes, Nek5000/CEM/RS, MFEM, libCEED and libParanumal. Their impact in various application problems have been demonstrated in the CEED milestone reports (Brown et al, 2017; Kolev et al, 2020; Min et al, 2017, 2019a; Shephard et al, 2019; Tomov et al, 2018, 2019) and other project reports (Ameen et al, 2020; Merzari et al, 2017; Min et al, 2019b). Some of the results are shown in Figures 17 and 18.…”

Section: Application Integrationmentioning

confidence: 96%

“…In addition, CEED has modes of operation where the elementwise operator evaluation can be recast as standard batch on medium-to-large-sized matrices (Abdelfattah et al, 2016b); the MAGMA backend for libCEED exploits this to improve performance for non-tensor finite elements (Kolev et al, 2020). The use of the batch BLAS operations increases the chances of performance portability, since BLAS is often highly optimized by vendors and other open source numerical software.…”

Section: High-order Software Ecosystemmentioning

confidence: 99%

“…Our target is to capture highly turbulent flow structures in the solution at minimal cost by using relatively few elements (≈300 per sphere) of high order (p=7). Algorithmic strategies including efficient edge collapse, tessellation, smoothing, and projection, are presented in (Kolev et al, 2020) along with quality measurements, flow simulations, validation, and performance results for pebble-bed geometries ranging from hundreds to thousands of pebbles. Figure 17(a) shows a case of 3344 pebbles in an annular domain using 1.1M spectral elements.…”

Section: Application Integrationmentioning

confidence: 99%

See 4 more Smart Citations

Efficient exascale discretizations: High-order finite element methods

Kolev

Fischer

Min

et al. 2021

The International Journal of High Performance Computing Applica

Self Cite

View full text Add to dashboard Cite

Efficient exploitation of exascale architectures requires rethinking of the numerical algorithms used in many large-scale applications. These architectures favor algorithms that expose ultra fine-grain parallelism and maximize the ratio of floating point operations to energy intensive data movement. One of the few viable approaches to achieve high efficiency in the area of PDE discretizations on unstructured grids is to use matrix-free/partially assembled high-order finite element methods, since these methods can increase the accuracy and/or lower the computational time due to reduced data motion. In this paper we provide an overview of the research and development activities in the Center for Efficient Exascale Discretizations (CEED), a co-design center in the Exascale Computing Project that is focused on the development of next-generation discretization software and algorithms to enable a wide range of finite element applications to run efficiently on future hardware. CEED is a research partnership involving more than 30 computational scientists from two US national labs and five universities, including members of the Nek5000, MFEM, MAGMA and PETSc projects. We discuss the CEED co-design activities based on targeted benchmarks, miniapps and discretization libraries and our work on performance optimizations for large-scale GPU architectures. We also provide a broad overview of research and development activities in areas such as unstructured adaptive mesh refinement algorithms, matrix-free linear solvers, high-order data visualization, and list examples of collaborations with several ECP and external applications.

show abstract

Section: High-order Software Ecosystemmentioning

confidence: 95%

Section: Application Integrationmentioning

confidence: 95%

Section: Application Integrationmentioning

confidence: 96%

Section: High-order Software Ecosystemmentioning

confidence: 99%

Section: Application Integrationmentioning

confidence: 99%

See 3 more Smart Citations

Efficient exascale discretizations: High-order finite element methods

Kolev

Fischer

Min

et al. 2021

The International Journal of High Performance Computing Applica

Self Cite

View full text Add to dashboard Cite

show abstract

HPC Hardware Design Reliability Benchmarking With HDFIT

Omland

Netti

Peng

et al. 2023

IEEE Trans. Parallel Distrib. Syst.

View full text Add to dashboard Cite

Large-Scale direct numerical simulations of turbulence using GPUs and modern Fortran

Karp

Massaro

Jansson

et al. 2023

The International Journal of High Performance Computing Applica

View full text Add to dashboard Cite

We present our approach to making direct numerical simulations of turbulence with applications in sustainable shipping. We use modern Fortran and the spectral element method to leverage and scale on supercomputers powered by the Nvidia A100 and the recent AMD Instinct MI250X GPUs, while still providing support for user software developed in Fortran. We demonstrate the efficiency of our approach by performing the world’s first direct numerical simulation of the flow around a Flettner rotor at Re = 30,000 and its interaction with a turbulent boundary layer. We present a performance comparison between the AMD Instinct MI250X and Nvidia A100 GPUs for scalable computational fluid dynamics. Our results show that one MI250X offers performance on par with two A100 GPUs and has a similar power efficiency based on readings from on-chip energy sensors.

show abstract

CEED-MS34: Improve performance and capabilities of CEED-enabled ECP applications on Summit/Sierra. Exascale Computing Project Milestone Report

Cited by 4 publications

References 0 publications

Efficient exascale discretizations: High-order finite element methods

Efficient exascale discretizations: High-order finite element methods

HPC Hardware Design Reliability Benchmarking With HDFIT

Large-Scale direct numerical simulations of turbulence using GPUs and modern Fortran

Contact Info

Product

Resources

About