HPE ProLiant DL384 Gen12 GPU Servers
Grace Hopper in a 2U air-cooled chassis. The HPE ProLiant Compute DL384 Gen12 runs single or dual NVIDIA GH200 superchips — up to 144 Arm Neoverse V2 cores, 960GB LPDDR5X and 288GB HBM3e at 9.8TB/s — built for large language model inference and memory-intensive HPC.
Product Overview
The HPE ProLiant Compute DL384 Gen12 is the first HPE server enabled with the NVIDIA GH200 Grace Hopper Superchip, optimised for AI inferencing on large language models that need substantial memory capacity, and for non-AI workloads such as large-scale simulation, EDA and weather forecasting. Configured as a dual-superchip NVIDIA GH200 NVL2 system it provides 144 Arm Neoverse V2 cores, 960GB of LPDDR5X and 288GB of HBM3e with up to 9.8TB/s of memory bandwidth, the two superchips NVLink-connected for twice the memory and performance of a single GH200. NVIDIA NVLink-C2C delivers 1800GB/s of CPU-to-GPU bandwidth. The 2U air-cooled chassis supports up to eight EDSFF NVMe Gen5 E3.S drives, up to four full-height half-length PCIe Gen5 slots, and the latest NVIDIA InfiniBand, Ethernet and BlueField adapters. The platform is NVIDIA OVX certified.
Key Specifications
- Brand
- HPE
- Condition
- New
- Category
- HPE GPU Servers
- Stock
- 5 units
- Superchip
- NVIDIA GH200 Grace Hopper Superchip; single-superchip or dual-superchip NVIDIA GH200 NVL2 configuration with NVLink between the two GH200
- CPU Cores
- 72 Arm Neoverse V2 cores (GH200) or 144 Arm Neoverse V2 cores (GH200 NVL2); base frequency 3.1GHz, all-core SIMD frequency 3.0GHz