The ServMax™ P47v2 GPU compute platform features the latest AMD EPYC™ CPU and Vega-based Radeon Instinct™ GPU accelerators to deliver ultimate performance for machine learning, advanced rendering and HPC applications. The platform supports 4x GPUs, M.2 and U.2 NVMe for high performance storage and EDR InfiniBand options for low latency high bandwidth adapted from HPC to guarantee system-to-rack level scalability.

EPYC panel

AMD EPYC™ CPU

Flexibility. Performance. Security. Born in the cloud, the AMD EPYC system on chip delivers 33% better memory bandwidth, 60% more I/O and 45% more cores than the competitor to power data center applications. Whether you deploy bare metal, virtualized, or in the cloud, the AMD EPYC processor powers fast, responsive, and secure IT.


32 Cores per CPU

2.2GHz or 2 GHz


8 Memory Channels

DDR4 2666 MHz


128 PCIe Lanes

PCIe 3.0


2TB Memory

Single Socket Platform

Radeon Instinct ™ MI60 GPU

Combining 16,384 Stream Processors, AMD’s state of the art 32GB HBM2 GPU memory with ultra-high memory bandwidth and AMD’s newly designed high bandwidth cache (HBC), each Vega-based AMD Radeon Instinct MI60 accelerator cards provides 7.4TFLOPS of double precision, 14.8TFLOPs of single precision and 29.5TFLOPs of half precision performance, making it ideal for deep learning and HPC applications.


64 Next-Gen Compute Units

(NCu - 16384 Stream Processors)


29.5 TFLOPS

Peak Half Precision Compute (FP16)


1 TB/S

Memory Bandwidth

14.8 TFLOPS

Peak Single Precision Compute (FP32)


32 GB HBM2

High Bandwidth Cache

[SMART]RACK™ P47V2

One PetaFLOP of Performance for Machine
Intelligence and Virtualization

Live Chat