• Chinese Optics Letters
  • Vol. 18, Issue 2, 020902 (2020)
Hiromi Sannomiya1, Naoki Takada2、*, Tomoya Sakaguchi1, Hirotaka Nakayama3, Minoru Oikawa2, Yuichiro Mori2, Takashi Kakue4, Tomoyoshi Shimobaba4, and Tomoyoshi Ito4
Author Affiliations
  • 1Graduate School of Integrated Arts and Sciences, Kochi University, Kochi, Kochi 780-8520, Japan
  • 2Research and Education Faculty, Kochi University, Kochi, Kochi 780-8520, Japan
  • 3National Astronomical Observatory of Japan, Mitaka, Tokyo 181-8588, Japan
  • 4Graduate School of Engineering, Chiba University, Chiba, Chiba 263-8522, Japan
  • show less
    DOI: 10.3788/COL202018.020902 Cite this Article Set citation alerts
    Hiromi Sannomiya, Naoki Takada, Tomoya Sakaguchi, Hirotaka Nakayama, Minoru Oikawa, Yuichiro Mori, Takashi Kakue, Tomoyoshi Shimobaba, Tomoyoshi Ito. Real-time electroholography using a single spatial light modulator and a cluster of graphics-processing units connected by a gigabit Ethernet network[J]. Chinese Optics Letters, 2020, 18(2): 020902 Copy Citation Text show less
    Proposed system with a single SLM and a multi-GPU cluster connected by gigabit Ethernet network.
    Fig. 1. Proposed system with a single SLM and a multi-GPU cluster connected by gigabit Ethernet network.
    Pipeline processing for real-time electroholography using the proposed system.
    Fig. 2. Pipeline processing for real-time electroholography using the proposed system.
    Outline of the packing process used to reduce the volume of data transferred.
    Fig. 3. Outline of the packing process used to reduce the volume of data transferred.
    Outline of the unpacking process used to reproduce the binary CGH.
    Fig. 4. Outline of the unpacking process used to reproduce the binary CGH.
    Comparison of a multi-GPU cluster connected with InfiniBand QDR and a multi-GPU cluster connected with a gigabit Ethernet.
    Fig. 5. Comparison of a multi-GPU cluster connected with InfiniBand QDR and a multi-GPU cluster connected with a gigabit Ethernet.
    Comparison of the multi-GPU cluster using the NVIDIA GeForce GTX TITAN X GPU and the multi-GPU cluster using the GeForce GTX 1080 Ti GPU.
    Fig. 6. Comparison of the multi-GPU cluster using the NVIDIA GeForce GTX TITAN X GPU and the multi-GPU cluster using the GeForce GTX 1080 Ti GPU.
    Snapshot of a reconstructed 3D video (Video 1).
    Fig. 7. Snapshot of a reconstructed 3D video (Video 1).
    ItemSpecification
    CPUIntel Core i7 4770 (Clock speed: 3.4 GHz)
    Main memoryDDR3-1600 4 GB
    OSLinux (CentOS 7.3 x86_64)
    SoftwareNVIDIA CUDA 8.0 SDK, OpenGL, MPICH 3.2
    GPUNVIDIA GeForce GTX TITAN X
    Table 1. Specifications of the Multi-GPU Cluster Using NVIDIA GeForce GTX TITAN X
    ItemSpecification
    CPUIntel Core i7 7800X (Clock speed: 3.5 GHz)
    Main memoryDDR4-2666 16 GB
    OSLinux (CentOS 7.6 x86_64)
    SoftwareNVIDIA CUDA 10.1 SDK, OpenGL, MPICH 3.2
    GPUNVIDIA GeForce GTX 1080 Ti
    Table 2. Specifications of the Multi-GPU Cluster Using NVIDIA GeForce GTX 1080 Ti
    Hiromi Sannomiya, Naoki Takada, Tomoya Sakaguchi, Hirotaka Nakayama, Minoru Oikawa, Yuichiro Mori, Takashi Kakue, Tomoyoshi Shimobaba, Tomoyoshi Ito. Real-time electroholography using a single spatial light modulator and a cluster of graphics-processing units connected by a gigabit Ethernet network[J]. Chinese Optics Letters, 2020, 18(2): 020902
    Download Citation