Service hotline: 400-898-1688

Search

Please enter the search content

LR

Online message

Service hotline

400-898-1688

Online message

Close

If you have anything to say, please leave a message and we will have someone to contact you at the first time. You can also call the hotline: 400-898-1688

name:

phone:

email:

Submit

Synopsis

Nettrix X640 G30 is a versatile GPU server integrating training and reasoning functions. Adopting dual-channel Intel® Xeon® scalable processors, the server is available for training eight GPUs or reasoning sixteen GPUs and gives performance support to intensive GPU computing in all respects in particular. Accommodating a variety of GPU topological structures and optimizing GPU interconnection for different applications and models, the product is the most efficient computing platform.

Product Highlights

1. Flexible architecture 2. Intensive and fast GPU computing

Product Features

Flexible configuration for optimized computing performance for various scenarios

● High-speed online processing: the server can balance the processing capabilities of CPU and GPU; high-performance online reasoning P2P model enabled by eight GPUs: besides, the server can feature deep learning to improve the efficiency of GPU Direct through interconnection of eight professional cards ● Intensive bitstream processing: with sync access of sixteen NVIDIA T4™Tensor Core GPUs, the server enables bitstream processing in over 100 channels ● HD processing: cost-effective HPC: with directly connected deployment of 4-8 cards, the server enables cost-effective computing based on CPU expansibility

Expansibility and optimized management

● Apart from intensive GPUs in full configuration, the server abounds with PCI-E interfaces for network, memory and other IO expansions. Available for U.2 hard drive, 10G-or-above TCP/IP network access, 20G IB network access, and distributed storage, the server enables rapid transmission and data storage in a high-end computing platform ● Characterized by optimized O&M efficiency, the professional management platform can recognize and monitor various popular GPUs on the market with functions of fan speed regulation, fault alarm, etc.

Intensive GPU computing

● Available for 2nd Gen Intel® Xeon® scalable processors based on a powerful X86 processing platform ● For training, the server can support eight NVIDIA V100™Tensor Core GPUs/NVIDIA Quadro® RTX 6000/ NVIDIA Quadro® RTX™8000 to work all day long at full speed ● For reasoning, the server can support on-line big data processing by at most 16NVIDIA T4™ Tensor Core GPUs

Technical Specifications

Model

4U rack server cabinet

Processor

Available for two 2nd Gen Intel® Xeon® scalable processors

Chipset

Intel C620 Series Chipset

Memory

Sixteen DDR4 memory slots, 2933 MHz; available for memory ECC, with expansion to 2TB memory at most

Storage controller

Available for directed mode, RAID0, RAID1, RAID10, RAID5, RAID50, RAID6, RAID60- and other storage modes; when the RAID controller is used, super capacitor of Cache protection and features as RAID level migration, disk roaming, self-diagno sis and Web remote setup are available

Local storage

Front: available for twenty-four 2.5-inch hard drives; Available for four U.2SSDs and two M.2 SATA SSDs

Network

Integrated 1G dual-port RJ45, various optional interfaces of 10G dual-port RJ45, 25G dual-port optical, 10G dual-port optical, etc.

PCI-Eexpandability

Available for twelve or twenty-two PCI-E expansion slots, expandable to OCP NIC (Including one special slot for RAID NIC)

GPUexpandability

Configuration 1: available for eight FHFL (full-height and full-length) dual-width GPUs or sixteen HHHL (half-height and half-length) single-width GPUs Configuration 2: available for four FHFL (full-height and full-length) dual-width GPUs or eight HHHL (half-height and half-length) single-width GPUs

List of available GPUs

Including but not limited to: NVIDIA® Tesla® V100/ NVIDIA® Tesla® V100S NVIDIA® Tesla® T4 NVIDIA Quadro® RTX 6000/8000 passive NVIDIA Quadro® P400/ P620/ P1000/ P2200/RTX 4000/ RTX 5000.SYNCII Please call Nettrix hotline 400-898-1688 for other GPUs

Other interfaces

One RJ-45 management interface at the rear of the cabinet Two USB3.0 interfaces at the front of the cabinet Two USB3.0 interfaces at the rear of the cabinet One VGA interface at the front of the cabinet One VGA interface at the rear of the cabinet One serial port at the rear of the cabinet

Power supply

1600W/2000W CRPS high-efficiency platinum hot-plug power supply, available for 3+1/2+2 power supply redundancy, voltage range: 100-240 V AC (Please refer to the power input value on the nameplate label of the computer host)

Fan module

4 hot-plug fan modules, available for N+1 fan redundancy

Management

Integrated BMC chip, available for advanced management of IPMI2.0, SOL, KVMOv er IP, virtual media, etc. with external one 1Gbps RJ45 management interface for enabling the NSCI function

Graphic card

Integrated display control unit, resolution: 1920x1080

Operating system

Microsoft® Windows Server, Red Hat Enterprise Linux, SUS ELinux Enterprise Server, CentOS, Ubuntu Vmware EXSI and others. Please call Nettrix hotline 400-898-1688 for more information

Working temperature

5℃-35℃ when working (please contact Nettrix for more information)

Product certification

CCC and others

Cabinet size

175.5mm (height) x 482mm (width) x 793mm (depth)

Weight under full configuration

58kg (excluding guide rail)
View All
Copyright Nettrix Information Industry co. LTD. jinicpb10200542 400-898-1688