1887

Abstract

Summary

The complexity involved in using APIs and resource managers from mulple cloud providers, supercompung centers, and partners highlights the need for having a single, unified way to launch and manage computaon workloads on mulple target machines – on-premises or in the cloud. We were able to efficiently run a large scale 3D RTM benchmark with thousands of shots on an on-demand cloud-based GPU-cluster providing 5 PFlops (peak SP). For this run, we used a domain decomposion and process placement schema that achieves a sustained communicaon performance close to 160Gb/s on the border exchange used by the wave propagaon finite difference method.

Loading

Article metrics loading...

/content/papers/10.3997/2214-4609.202011916
2020-07-15
2021-05-11
Loading full text...

Full text loading...

References

  1. Bukhamsin, A & Schonewille, A.
    , Containerizing Parallel MPI-based HPC Applications, 2017. 10.3997/2214‑4609.201702320.
    https://doi.org/10.3997/2214-4609.201702320 [Google Scholar]
  2. Souza Filho, P. & Sardinha, A. & Ávila, C. & Azambuja, A. & Sierra, F. & De Paula, D. & Vecino, M. & Silva, L. & Ji, N.
    , Seismic Processing with Hybrid HPC. Fourth EAGE Workshop on High Performance Computing for Upstream, 2019,
    [Google Scholar]
  3. Breuer, Alexander & Cui, Yifeng & Heinecke, Alexander
    . (2019). Petaflop Seismic Simulations in the Public Cloud. 10.1007/978‑3‑030‑20656‑7_9.
    https://doi.org/10.1007/978-3-030-20656-7_9 [Google Scholar]
  4. PauliusMicikevicius
    . 3D finite difference computation on GPUs using CUDA. 2009. In Proceedings of 2nd Workshop on General Purpose Processing on Graphics Processing Units (GPGPU-2). ACM, New York, NY, USA, 79–84. http://dx.doi.org/10.1145/1513895.1513905
    [Google Scholar]
  5. Kari N.Erickson & Faith V.Van Wig & Luke A.Kachelmeier
    , Comparison of High Performance Network Options: EDR InfiniBand vs. 100Gb RDMA Capable Ethernet. SuperComputing 2016 Poster. http://sc16.supercomputing.org/sc-archive/tech_poster/poster_files/post149s2-file3.pdf
    [Google Scholar]
  6. Amazon FSx for Lustre Reference
    Amazon FSx for Lustre Reference. https://docs.aws.amazon.com/fsx/latest/LustreGuide/performance.html
  7. The Floodgates Are Open – Increased Network Bandwidth for EC2 Instances. https://aws.amazon.com/pt/blogs/aws/the-floodgates-are-open-increased-network-bandwidth-for-ec2-instances/
    [Google Scholar]
  8. Widest path problem. https://en.wikipedia.org/wiki/Widest_path_problem
    [Google Scholar]
http://instance.metastore.ingenta.com/content/papers/10.3997/2214-4609.202011916
Loading
/content/papers/10.3997/2214-4609.202011916
Loading

Data & Media loading...

This is a required field
Please enter a valid email address
Approval was a Success
Invalid data
An Error Occurred
Approval was partially successful, following selected items could not be processed due to error