Supermicro has introduced their AMD-optimised GPU Server and Blade Server fit with AMD EPYC 7002 CPUs to deliver high core count and memory capacity.

Blog | Newly Introduced Supermicro AMD-Optimised GPU Server and SuperBlade

June 5th 2020

 

The Supermicro A+ Server 4121GS-TNR is

AMD EPYC GPU Optimised

 
 

Maximized Performance And Cost Efficiency

 

The newly introduced Supermicro AS-4121GS-TNR is based upon dual AMD EPYC 7002 CPUs. This GPU server utilises AMD EPYC 7002 Rome to deliver high core counts and memory capacity per GPU.

  • Supporting up to 10x double width GPUs that direct connect for maximum performance.

The AMD EPYC 7002 Series Processors have set a high standard in architecture, performance and security for the modern data centre.

AMD EPYC Infinity Architecture
The processor is driven by AMD Infinity architecture as the first x86-architecture server processor based on 8nm process technology which is a hybrid, multi-die architecture. The system-on-chip (SoC) design eliminates the need for many external support chips, reducing capital and server design costs.

Leading Performance
The AMD EPYC 7002 Series offers up to 64 cores per SOC and Zen 2 features, surpassing the older 1st Gen AMD EPYC Processors. The new processors feature improved execution pipelines, higher clock rates, and up to 4x the shared Level 3 cache. As a result, performance is doubled and up to 4x the theoretical peak floating-point operations per second (FLOPS) compared to the older version.

Advanced Security
AMD EPYC processors feature advanced security features, named AMD Infinity Guard. This consists of AMD’s secure processor, Secure Encrypted Virtualization (SEV) and Secure Memory Encryption (SME). All of these innovative features assist in minimizing potential threat surfaces as software is booted, executed and accesses critical data.

 

 

The AS-4121GS-TNR is optimised for highly data-driven workloads within:

  • AI and Deep Learning
  • Virtualisation
  • Scientific Research
  • 3D Rendering
  • High Performance Computing

What You Can Expect

  • Multi-GPU optimized thermal architecture for high performance and reliability
  • Advanced GPU interconnect selections for ideal efficiency and reduced latency
  • Innovative GPU architecture for NVIDIA HGX-2 platform featuring NVLink and NVSwitch
Share