At ISC High Performance 2018, AMD continued to make advances in deep learning and high-performance compute with the introduction of ready-to-deploy server platforms from AMAX, Exxact, Inventec and Supermicro, powered by Radeon Instinct and EPYC processors. These new solutions continue AMD’s recent momentum in compute, following the demonstration of 7nm Radeon “Vega” professional/datacenter graphics and the announcement of the next generation 7nm EPYC processor codenamed, “Rome,” shipping later this year.
The systems can easily be configured with the latest ROCm drivers and MIOpen libraries, giving customers the option of different deep learning frameworks such as TensorFlow or Caffe. New solutions include:
- AMAX ServMax P47 system: Ready-to-deploy TensorFlow or Caffe deep learning compute solution in a 2U AMD EPYC processor server with 4x Radeon Instinct MI25 compute GPUs
- Inventec P47G4: Optimal, deep learning solution featuring 2U single-socket AMD EPYC processor server with 4x Radeon Instinct MI25 compute GPUs
- Supermicro SuperServer 4029GP-TRT2 : Ultimate, single-root deep learning accelerated solution featuring a 4U Dual Intel Xeon processor server with 8x Radeon Instinct MI25 compute GPUs
- Exxact Tensor TS4-672702-AML: Fast and easy deep learning deployment on optimal single root complex design in a 4U dual Intel Xeon processor server with 8x Radeon Instinct MI25 compute GPUs
AMD also announced The National Institute for Nuclear Physics (INFN) in Italy has chosen the AMD EPYC 7351 processor to power its high-performance computing cluster. INFN is a leading European research institution, conducting theoretical and experimental research in the fields of subnuclear, nuclear, and astroparticle physics while offering access to its exceptional processing resources across Europe.
