Inspur NF5488M5

Industry first 4U 8GPU NVSwitch-empowered AI Server

Product Overview

The growth of deep learning and compute workloads is driving hyperscalers to seek for dramatic acceleration. The Inspur NF5488M5 is the Industry first 4U 8GPU NVSwitch-empowered AI Server to deliver unparalleled AI performance of 1 petaFLOPS. It is tested and verified to be the ideal choice for hyperscale AI training and high performance computing workloads, including NLP, autonomous vehicles and more.

Extreme Performance

8x V100 GPU with NVSwitch to work as one GPU, delivering AI performance of 1 petaFLOPS•2x 28-core processors to provide leading general-purpose computing

2x 28-core processors to provide leading general-purpose computing

Easy to Deploy

6KW in 4U, a design make it easy to deploy in a wide spectrum of data center scenarios, especially for power-constrained racks

Flexible GPU cluster expansion over PCIe fabric

Energy-Efficient

Operate on 54VDC, a more power-efficient voltage for GPUs

A multi-layer cooling design and intelligent PID adjustment that provide industry-leading thermal management and control

Technical Specifications

Specifications

NF5488M5

GPU

8*NVIDIA® Tesla® V100 SXM3 GPU with NVSwitch

Performance

1 petaFLOPS

Processor

2*2nd Generation Intel® Xeon® Scalable Processor, 3*UPI

Memory

24 DDR4 2933MT/s ECC DIMMs, 12 channels

Storage

8*2.5“SATA (Includes 4*2.5” NVMe), 2*M.2

Interface

4*PCIe 3.0 x16 for 100G NIC, 1*PCIe 3.0 x8 for 50G NIC/NVMe

Cooling

Redundant hot swap fan, air cooling

PSU

2+2 Redundant, up to 6 KW in total

Chassis

4U W*H*D 448mm*175.5mm*850mm

Inspur NF5488M5