TN and NN implementation on FPGA

Europe/Zurich
CERN

CERN

Description

This scientific working group focuses on the implementation of tensor network states and neural network algorithms on specialized hardware platforms, including FPGAs (Field Programmable Gate Arrays) and GPUs (Graphics Processing Units). The group aims to explore innovative approaches to accelerate computations in quantum many-body systems, quantum field theory simulations, and optimization problems.

Key objectives include:

  • Developing hardware-optimized algorithms for tensor contractions and neural network inference/training.
  • Leveraging the parallel processing capabilities of GPUs and the reconfigurability of FPGAs to enhance performance for large-scale simulations.
  • Addressing challenges such as memory bandwidth limitations, precision trade-offs, and energy efficiency in high-performance computing contexts.

By integrating expertise from quantum information science, machine learning, and hardware acceleration, the group seeks to bridge theoretical developments with practical implementations, enabling breakthroughs in scalable simulations and real-time quantum data processing.

    • 10:00 10:15
      Kickoff 4/2-037 - TH meeting room

      4/2-037 - TH meeting room

      CERN

      18
      Show room on map

      Revisit timetable, bring everyone interested to the same page

    • 10:15 11:45
      Quantum TEA (Seminar) 4/2-037 - TH meeting room

      4/2-037 - TH meeting room

      CERN

      18
      Show room on map

      Presentation of work related to the hands-on sessions

    • 11:45 12:30
      HPC/group tasks & people 4/2-037 - TH meeting room

      4/2-037 - TH meeting room

      CERN

      18
      Show room on map

      Install cluster, coordinate between everyone who wants to get involved in the hands-on sessions

    • 12:30 14:00
      Lunch
    • 14:00 15:00
      HPC/group tasks & people 4/2-037 - TH meeting room

      4/2-037 - TH meeting room

      CERN

      18
      Show room on map

      Install cluster, coordinate between everyone who wants to get involved in the hands-on sessions

    • 15:00 18:00
      Hands-on 4/2-037 - TH meeting room

      4/2-037 - TH meeting room

      CERN

      18
      Show room on map

      Hands-on GPU TN: Run tensor network machine learning on GPU for the given data set
      Hands-on FPGA TN: Run tensor network machine learning inference on FPGA for the given data set
      Hands-on GPU NN: Run NN from GitHub for the given data set
      Hands-on FPGA NN: Run NN inference from GitHub model for the given data set

    • 10:00 10:15
      Sync 4/S-056

      4/S-056

      CERN

      20
      Show room on map

      Update of what happened the day before, discuss potential blocking steps

    • 10:15 11:45
      FPGA in-depth (Seminar) 4/S-056

      4/S-056

      CERN

      20
      Show room on map

      Presentation of work related to the hands-on sessions

    • 11:45 12:30
      Hands-on 4/S-056

      4/S-056

      CERN

      20
      Show room on map

      Hands-on GPU TN: Run tensor network machine learning on GPU for the given data set
      Hands-on FPGA TN: Run tensor network machine learning inference on FPGA for the given data set
      Hands-on GPU NN: Run NN from GitHub for the given data set
      Hands-on FPGA NN: Run NN inference from GitHub model for the given data set

    • 12:30 14:00
      Lunch
    • 14:00 18:00
      Hands-on 4/S-056

      4/S-056

      CERN

      20
      Show room on map

      Hands-on GPU TN: Run tensor network machine learning on GPU for the given data set
      Hands-on FPGA TN: Run tensor network machine learning inference on FPGA for the given data set
      Hands-on GPU NN: Run NN from GitHub for the given data set
      Hands-on FPGA NN: Run NN inference from GitHub model for the given data set

    • 10:00 10:15
      Sync 4/2-037 - TH meeting room

      4/2-037 - TH meeting room

      CERN

      18
      Show room on map

      Update of what happened the day before, discuss potential blocking steps

    • 10:15 11:45
      Chai TEA (Seminar)

      Presentation of work related to the hands-on sessions

    • 11:45 12:30
      Hands-on 4/2-037 - TH meeting room

      4/2-037 - TH meeting room

      CERN

      18
      Show room on map

      Hands-on GPU TN: Run tensor network machine learning on GPU for the given data set
      Hands-on FPGA TN: Run tensor network machine learning inference on FPGA for the given data set
      Hands-on GPU NN: Run NN from GitHub for the given data set
      Hands-on FPGA NN: Run NN inference from GitHub model for the given data set

    • 12:30 14:00
      Lunch
    • 14:00 18:00
      Hands-on: Hands-om 4/2-037 - TH meeting room

      4/2-037 - TH meeting room

      CERN

      18
      Show room on map

      Hands-on GPU TN: Run tensor network machine learning on GPU for the given data set
      Hands-on FPGA TN: Run tensor network machine learning inference on FPGA for the given data set
      Hands-on GPU NN: Run NN from GitHub for the given data set
      Hands-on FPGA NN: Run NN inference from GitHub model for the given data set

    • 10:00 10:15
      Sync 4/2-037 - TH meeting room

      4/2-037 - TH meeting room

      CERN

      18
      Show room on map

      Update of what happened the day before, discuss potential blocking steps

    • 10:15 12:30
      Hands-on 4/2-037 - TH meeting room

      4/2-037 - TH meeting room

      CERN

      18
      Show room on map

      Hands-on GPU TN: Run tensor network machine learning on GPU for the given data set
      Hands-on FPGA TN: Run tensor network machine learning inference on FPGA for the given data set
      Hands-on GPU NN: Run NN from GitHub for the given data set
      Hands-on FPGA NN: Run NN inference from GitHub model for the given data set

    • 12:30 14:00
      Lunch
    • 14:00 17:00
      Hands-on 4/2-037 - TH meeting room

      4/2-037 - TH meeting room

      CERN

      18
      Show room on map

      Hands-on GPU TN: Run tensor network machine learning on GPU for the given data set
      Hands-on FPGA TN: Run tensor network machine learning inference on FPGA for the given data set
      Hands-on GPU NN: Run NN from GitHub for the given data set
      Hands-on FPGA NN: Run NN inference from GitHub model for the given data set

    • 17:00 18:00
      CERN visit

      Small tour for whom is interested

    • 10:00 10:15
      Sync 4/S-030

      4/S-030

      CERN

      30
      Show room on map

      Update of what happened the day before, discuss potential blocking steps

    • 10:15 12:30
      Benchmark 4/S-030

      4/S-030

      CERN

      30
      Show room on map

      Hands-on GPU Benchmark: Run final data set on GPUs for TN-ML and NN
      Hands-on FPGA Benchmark: Run final data set on FPGA for inference for TN-ML and NN

    • 12:30 14:00
      Lunch
    • 14:00 18:00
      Final results / wrap-up / next steps 4/S-030

      4/S-030

      CERN

      30
      Show room on map

      Presentation of results and discussion of next steps
      Final results: ROC TN/NN
      Final results: FPGA-latency