The impact of the AC922 Architecture on Performance of Deep Neural Network Training - Publikacja - MOST Wiedzy

Twoje konto

zaloguj

Wyszukiwarka

The impact of the AC922 Architecture on Performance of Deep Neural Network Training

Abstrakt

Practical deep learning applications require more and more computing power. New computing architectures emerge, specifically designed for the artificial intelligence applications, including the IBM Power System AC922. In this paper we confront an AC922 (8335-GTG) server equipped with 4 NVIDIA Volta V100 GPUs with selected deep neural network training applications, including four convolutional and one recurrent model. We report performance results depending on batch sizes and GPU selection and compare them with the results from another contemporary workstation based on the same set of GPUs – NVIDIA® DGX Station ™ . The results show that the AC922 performs better in all tested configurations, achieving improvements up to 10.3%. Profiling indicates that the improvement is due to the efficient I/O pipeline. The performance differences depend on the specific model, rather than on the model class (RNN/CNN). Both systems offer good scalability up to 4 GPUs. In certain cases there is a significant difference in performance depending on exactly which GPUs are used for computations.

Cytowania

  • 3

    CrossRef

  • 0

    Web of Science

  • 4

    Scopus

Autorzy (3)

Cytuj jako

Pełna treść

pełna treść publikacji nie jest dostępna w portalu

Słowa kluczowe

Informacje szczegółowe

Kategoria:
Aktywność konferencyjna
Typ:
publikacja w wydawnictwie zbiorowym recenzowanym (także w materiałach konferencyjnych)
Tytuł wydania:
2019 International Conference on High Performance Computing & Simulation (HPCS) strony 666 - 673
Język:
angielski
Rok wydania:
2020
Opis bibliograficzny:
Rościszewski P., Iwański M., Czarnul P.: The impact of the AC922 Architecture on Performance of Deep Neural Network Training// 2019 International Conference on High Performance Computing & Simulation (HPCS)/ : , 2020, s.666-673
DOI:
Cyfrowy identyfikator dokumentu elektronicznego (otwiera się w nowej karcie) 10.1109/hpcs48598.2019.9188164
Weryfikacja:
Politechnika Gdańska

wyświetlono 99 razy

Publikacje, które mogą cię zainteresować

Performance evaluation of Unified Memory with prefetching and oversubscription for selected parallel CUDA applications on NVIDIA Pascal and Volta GPUs

2019

Performance/energy aware optimization of parallel applications on GPUs under power capping

2020

Exploring Neural Networks for Musical Instrument Identification in Polyphonic Audio

2024

Benchmarking Deep Neural Network Training Using Multi- and Many-Core Processors

2020
Meta Tagi