%0 Conference Paper
%A Willsch, Dennis
%A Michielsen, Kristel
%A De Raedt, Hans
%T Benchmarking GPU Clusters with the Jülich Universal Quantum Computer Simulator
%M FZJ-2021-02813
%D 2021
%X Simulating quantum computers is a versatile approach to benchmark supercomputers with thousands of GPUs. In this session, quantum computer simulators are explained from a linear algebra perspective, using the Jülich Universal Quantum Computer Simulator (JUQCS) as an example. It is shown how the memory-, network-, and computation-intensive operations of JUQCS can be used to benchmark high-performance computers. In particular, the CUDA-aware MPI communication scheme is illustrated. We present results for benchmarking JUWELS Booster, a cluster with 3744 NVIDIA A100 Tensor Core GPUs.
%B GPU Technology Conference 2021
%C 12 Apr 2021 - 16 Apr 2021, San Jose (USA)
Y2 12 Apr 2021 - 16 Apr 2021
M2 San Jose, USA
%F PUB:(DE-HGF)6
%9 Conference Presentation
%U https://juser.fz-juelich.de/record/893758