Home > Cross Border Sourcing >

Gpu AI Server Enterprise HPE ProLiant Compute DL384 Gen12 NVIDIA GH200 NVL2 Free Compute Private Cloud Rack Mount

Product Categories

International Freight Forwarding Services [34]
Cross Border Sourcing [18]
China Air Freight Service [56]
China Sea Freight Services [47]
Middle East Shipping [26]
International Rail Freight [10]
Door To Door Shipping From China [38]
Road Freight From China [14]
International Packing Service [7]
International Warehousing Services [9]
Cargo Insurance Service [2]

SHENZHEN JUHE SUPPLY CHAIN CO.,LTD

Contact Now

Gpu AI Server Enterprise HPE ProLiant Compute DL384 Gen12 NVIDIA GH200 NVL2 Free Compute Private Cloud Rack Mount

Brand Name : HPE

Model Number : HPE DL384 Gen12

Place of Origin : America

MOQ : 1pcs

Price : $1950

Payment Terms : L/C, D/A, T/T, Western Union, MoneyGram

Supply Ability : unlimited

Delivery Time : 3 - 5days

Packaging Details : carton

Type : Rack

Processor Type : NVIDIA Grace CPU and Hopper GPU

Brand : HPE

Contact Now

MODEL	HPE ProLiant Compute DL384 Gen12
Processor Series	NVIDIA Grace CPU and Hopper GPU
Power supply type	Provides up to four HPE 1800W to 2200W Resilient Slot Titanium hot-swappable power supply kits for 2+0, 2+1, or 2+2 redundancy for a single GH200.Minimum of two power supplies. For dual GH200 and NVL2, 3+1 redundancy is possible and 4 power supplies are required at the same time, no other options are supported
expansion slot	Up to four high-speed PCIe Gen5 x16 devices, up to four PCIe full-height half-length (FHHL) slots, or up to two PCIe FHHLs and two OCP 3.0s
System Fan Function	6 hot-swappable redundant fans
Storage Controller	Hardware RAID controllers are not supported on this generation of M.2 or EDSFF NVMe drives
Infrastructure management	HPE iLO Standard (with embedded Intelligent Provisioning), HPE iLO Advanced
Supported Drives	Up to 8 EDSFF NVMe Gen5 Drives

Overview

HPE ProLiant Compute DL384 Gen12 is the first server from HPE enabled with NVIDIA GH200 NVL2, optimized for AI inferencing for large language models that require a large memory capacity and/or non-AI workloads like large scale simulation, EDA, weather forecasting, etc.

With up to 1.2TB of fast unified memory and 5TB/s bandwidth, the NVIDIA GH200 NVL2 can handle large language models fine tuning and inferencing with Retrieval Augmented Generation (RAG) and more users featuring twice the performance of the previous generation. The HPE ProLiant Compute DL384 Gen12 delivers the best performance per GPU in our HPE ProLiant portfolio. HPE ProLiant Compute DL384 is ideal for mixed or memory intensive workloads, whether you are doing AI or traditional HPC.

Building on HPE ProLiant as the legendary foundation, the HPE ProLiant Compute DL384 Gen12 delivers a consistent experience across the HPE ProLiant portfolio with HPE iLO management and firmware that delivers robust reliability and security with silicon Root of Trust technology from HPE.

HPE and NVIDIA are continuing to innovate to help our customers unlock next-gen scale-out accelerated computing for generative AI, with superchip performance for their Enterprise AI Factory.

Gpu AI Server Enterprise HPE ProLiant Compute DL384 Gen12 NVIDIA GH200 NVL2 Free Compute Private Cloud Rack Mount

What’s New

• The first HPE ProLiant rack mount server with the latest NVIDIA GH200 Grace Hopper™ Superchip

• Support for dual superchips with NVIDIA GH200 NVL2. NVLink between two GH200 for twice the memory and performance.

• Support for the latest NVIDIA InfiniBand,Ethernet, and Bluefield adapters ensures your AI fabric runs at top performance. • NVIDIA OVX™ certified for Artificial Intelligence workloads

• High-performance LLM inference to maximize data center utilization

• With up to 2X higher inference performance compared to H100

• HPE Silicon Root of Trust offers industry leading innovation based on the HPE zero trust architecture from edge to cloud, to protect your infrastructure, workloads, and data

Application: Gpu AI Server Enterprise HPE ProLiant Compute DL384 Gen12 NVIDIA GH200 NVL2 Free Compute Private Cloud Rack Mount

To simplify your experience, configuration of the server has been streamlined significantly. In brief we have set up only three core hardware kits- the server itself, a kit with a single GH200 superchip, and a kit with two GH200 superchips. The superchip kits include the enablement items needed to build any supported configuration of the server. Then we have a small number of configuration settings to choose that set whether you configure with the spare PCIe, OCP, or added drives. This is summarized in tables below:

Product Tags: