HPE ProLiant Compute DL384 Gen12

Categories: ,

Breakthrough AI Performance
• Industry-first rack server featuring the NVIDIA® GH200 Grace Hopper™ Superchip
• Dual Superchip configuration with NVIDIA® GH200 NVL2 technology
• 8 petaflops of AI performance in a single node – ideal for: Large language models (LLMs), Generative AI workloads, Advanced scientific computing

Unmatched GPU Capabilities
• 3.5× greater GPU memory capacity vs. NVIDIA® H100 Tensor Core GPUs
• 3× higher memory bandwidth for data-intensive AI applications
• Innovative architecture combining: 2× Grace CPUs, 2× Hopper GPUs, Unified memory space

Specifications

Processor

  • Processor type: Intel
  • Processor family: 6th Generation Intel® Xeon® Scalable Processors
  • Processor number: 2
  • Processor core available: 64 to 144 core, depending on processor
  • Processor speed: 2.4 GHz maximum, depending on processor

Server Memory

  • Memory type: HPE DDR5 Smart Memory
  • Maximum memory: 4 TB – RDIMM (2 TB per processor)
  • Memory slots: 32 DIMM slots
  • Memory protection features: RAS – Advanced ECC, online spare, mirroring, combined channel (lockstep) functionality, and HPE Fast Fault Tolerant Memory (ADDDC)

Server Cooling

System fan features: 4 dual-rotor and 8 single-rotor hot plug fans included

Server Expansions

Expansion slots6, for detailed descriptions refer to the QuickSpecs

Server Form Factor

Form factor: 4U Rack

Server Drives

Drive supported: SFF NMVE and EDSFF

Server Power

Power supply type: Up to 8 M-CRPS. Single 1+1 redundancy for system board. Dual 2+1 redundancy for GPUs.

Management Software

Infrastructure managementHPE iLO Standard with intelligent provisioning (embedded), HPE OneView Standard (requires download) (standard) HPE iLO Advanced, and HPE OneView Advanced (optional, requires licenses)

Security

SecurityOptional locking Bezel kit, Intrusion Detection kit, and embedded HPE TPM 2.0

Warranty

WarrantyServer warranty includes 3-year parts, 3-year labor, and 3-year on-site support with next business day response.