Tracing of GPU-Aware MPI Applications: First Benchmarks for the Angara Interconnect

Ismagilov T.; Mukosey A.; F. Smirnov; V. Galigerov; Grishichkin Y.; V. Stegailov; A. Timofeev

doi:10.1007/978-3-031-85697-6_17

Publications

?

Tracing of GPU-Aware MPI Applications: First Benchmarks for the Angara Interconnect

P. 256–270.

Ismagilov T., Mukosey A., Smirnov F., Galigerov V., Grishichkin Y., Stegailov V., Timofeev A.

The efficiency of data transfer is one of the most important issues of supercomputer development in the post-Moore era. The rise of heterogeneous computing systems introduces such complicated patterns of data transfers as, for instance, the GPU-aware MPI technology. The practical deployment of this technology in applications requires the development of the dedicated system software as well as the analysis tools for tracing the runtime behavior of the corresponding applied algorithms. In this work we present the UCX API for GPU-aware MPI implementation over Angara interconnect and analyze the execution patterns of the rocHPL benchmark using the Score-P infrastructure. This analysis allows us to make a comparison of the GPU-aware MPI implementation for the Angara interconnect with the InfiniBand implementation.

Language: English

DOI

Text on another site

Keywords: HPL HIP RDMA Low-latency communication Score-P

In book

Parallel Processing and Applied Mathematics. 15th International Conference, PPAM 2024, Ostrava, Czech Republic, September 8–11, 2024, Revised Selected Papers, Part I

Vol. 15579: Lecture Notes in Computer Science. , Springer, 2025.

Towards performance analysis of GPU-aware MPI over Angara interconnect

Ismagilov T., Mukosey A., Smirnov F. et al., International Journal of High Performance Computing Applications 2026 Vol. 40 No. 2 P. 240–253

One of the most important aspects of supercomputer development in the post-Moore era is the interconnect technologies that allow one to unite a multitude of processing elements into a well-synchronized computing system. Novel types of supercomputer interconnect require careful benchmarking and compliance with the requirements of modern hardware trends. GPU-based heterogeneous computing is one of ...

Added: May 7, 2026

Performance of Supercomputers Based on Angara Interconnect and Novel AMD CPUs/GPUs

Shamsutdinov A., Khalilov M., Ismagilov T. et al., , in: 22nd International Conference, MMST 2022, Nizhny Novgorod, Russia, November 14–17, 2022, Revised Selected Papers.: Springer, 2022. P. 401–416.

A low-latency high bandwidth interconnect that makes a unified system from a collection of nodes is a heart of any modern su- percomputer. At the moment, Infiniband is the main commercially avail- able type of interconnect without any other real competition world-wide. Proprietary interconnects are known to stand behind effcient supercomputer systems. Since 2016, the ...

Added: May 16, 2023

Multi-GPU GEMM Algorithm Performance Analysis for Nvidia and AMD GPUs Connected by NVLink and PCIe

Choi Y. R., Stegailov V., , in: 22nd International Conference, MMST 2022, Nizhny Novgorod, Russia, November 14–17, 2022, Revised Selected Papers.: Springer, 2022. Ch. 23 P. 281–292.

Modern types of multi-GPU servers combine up to 8 A100 GPUs connected by NVLink 3.0 links through NVSwitch. This connectivity provides unprecedented capabilities for multi-GPU algorithms. In this work, we analyze the performance of matrix-matrix multiplication algorithm developed by us previously. Tuning principles and limits for maximum performance are discussed. Algorithm performance for much more ...

Added: December 26, 2022