AMD EPYC CPUs Power 3 of Top 10 Supercomputers on The Planet, Also Has 8 of The Top 10 Most Energy Efficient Supercomputers

AMD EPYC CPUs Power 3 of Top 10 Supercomputers on The Planet, Also Has 8 of The Top 10 Most Energy Efficient Supercomputers

3 years ago
Anonymous $drS9DEX_Sj

https://wccftech.com/amd-epyc-cpus-power-more-high-performance-energy-efficient-supercomputers-than-intel-xeon/

During the ISC High Performance 2021 seminar, it was revealed that not only are AMD's EPYC CPUs powering more of the top 10 supercomputers than Intel's Xeon chips but are also the leading in the energy efficiency charts.

According to the TOP500 and Green500 charts, out of the top 10 positions, AMD is powering three supercomputers which include Perlmutter (NERSC Lawrence Berkeley National Laboratory), Selene (NVIDIA Corporation), and JUWELS Booster Module (Forschungzentrum Julich FZH). All three of the supercomputers mentioned are powered by AMD's EPYC processors and based on various designs from HPE, BullSequana, and DGX. All of these supercomputers combined have a peak processing output of 172.2 PetaFlops which is expected to expand further in the coming years as deployment of more chips continues.

AMD EPYC CPUs Power 3 of Top 10 Supercomputers on The Planet, Also Has 8 of The Top 10 Most Energy Efficient Supercomputers

Jun 28, 2021, 2:29pm UTC
https://wccftech.com/amd-epyc-cpus-power-more-high-performance-energy-efficient-supercomputers-than-intel-xeon/ > During the ISC High Performance 2021 seminar, it was revealed that not only are AMD's EPYC CPUs powering more of the top 10 supercomputers than Intel's Xeon chips but are also the leading in the energy efficiency charts. > According to the TOP500 and Green500 charts, out of the top 10 positions, AMD is powering three supercomputers which include Perlmutter (NERSC Lawrence Berkeley National Laboratory), Selene (NVIDIA Corporation), and JUWELS Booster Module (Forschungzentrum Julich FZH). All three of the supercomputers mentioned are powered by AMD's EPYC processors and based on various designs from HPE, BullSequana, and DGX. All of these supercomputers combined have a peak processing output of 172.2 PetaFlops which is expected to expand further in the coming years as deployment of more chips continues.