HPE ProLiant DL380a Gen11 2x Gaudi 2 PCIe Server

The HPE ProLiant DL380a Gen11 2x Gaudi 2 PCIe Server (P65395-B21) is a purpose-engineered 2U rack server designed to accelerate deep learning training and AI inference workloads at enterprise and hyperscale datacenter scale. The system integrates two Intel Gaudi 2 accelerator cards in a PCIe form factor, leveraging the Gaudi 2 HL-225H ASIC architecture with 96 GB of HBM2e memory per accelerator and 24 tensor processor cores per card, delivering high-throughput matrix computations suited to large language model training, computer vision, and generative AI pipelines.

Built on the HPE ProLiant DL380a Gen11 platform, the server pairs the dual Gaudi 2 accelerators with support for 4th Generation Intel Xeon Scalable processors and DDR5 system memory, providing the CPU-side computational bandwidth and memory capacity needed to feed AI accelerator workloads efficiently. The Gaudi 2 cards interconnect via on-card RDMA-capable 24-port 100 GbE RoCE v2 integration, enabling scale-out training across multiple nodes without requiring external InfiniBand fabric in many configurations. HPE iLO 6 management and the ProLiant Gen11 silicon root-of-trust security architecture are embedded throughout the platform.

Targeted at enterprise IT organizations, national AI research institutions, and datacenter operators across UAE, GCC, EMEA, and APAC, the DL380a Gen11 2x Gaudi 2 PCIe Server is optimized for organizations deploying the Intel Gaudi software ecosystem, including integration with PyTorch and TensorFlow via the Intel Gaudi software suite (SynapseAI), providing a vendor-supported software path for large-scale AI model development and production inference serving.

Ideal for

• Large language model (LLM) fine-tuning and pre-training using PyTorch with Intel SynapseAI SDK on enterprise-grade AI infrastructure
• Generative AI inference serving for text, image, and multimodal foundation models in production datacenter environments
• Computer vision model training including object detection, segmentation, and classification pipelines at scale
• High-performance deep learning research and AI benchmarking within national labs, universities, and enterprise AI centers of excellence
• Scale-out distributed training clusters leveraging Gaudi 2 on-card 100 GbE RoCE v2 interconnect across multiple DL380a Gen11 nodes
• AI-accelerated data analytics and recommendation system training for e-commerce, financial services, and telecommunications verticals

Technical specifications

Available from Omnixon Global. Submit an RFQ and our team will confirm configuration and availability for your order.