HPE NVIDIA L4 24GB GDDR6 PCIe Accelerator Graphics Card S0K89C

HPE NVIDIA L4 24GB GDDR6 PCIe Accelerator Graphics Card S0K89C

The HPE NVIDIA L4 24GB PCIe Accelerator is designed to bring efficient GPU acceleration to supported HPE compute platforms. Powered by the NVIDIA Ada Lovelace architecture, it supports enterprise workloads including artificial intelligence inference, video processing, virtual workstations, graphics, rendering and data analytics.

Versatile Enterprise Acceleration

The NVIDIA L4 combines fourth-generation Tensor Cores with third-generation RT Cores to accelerate a broad selection of AI, visual computing and graphics applications. Its 24GB of GDDR6 memory and 300 GB/s memory bandwidth provide capacity and throughput for data-intensive enterprise workloads.

Designed for AI and Video Processing

The accelerator supports mixed-precision processing for deep learning and machine learning applications. Two NVENC encoders, four NVDEC decoders and four JPEG decoders also provide dedicated hardware resources for video transcoding, streaming, visual effects and image-processing workflows.

Compact, Energy-Conscious Design

A half-height, half-length, single-slot form factor allows the L4 to fit into a range of supported HPE server configurations. The card operates within a maximum 72W power envelope and uses passive cooling, making correctly configured server airflow essential.

Key Features:

• NVIDIA Ada Lovelace architecture supports AI, graphics, video and virtualisation workloads.
• 24GB of GDDR6 memory provides capacity for demanding datasets and visual applications.
• 300 GB/s memory bandwidth supports efficient movement of data within the accelerator.
• PCIe Gen4 x16 connectivity provides a high-bandwidth connection to supported HPE servers.
• Low-profile, half-length, single-slot construction helps conserve valuable server space.
• 72W maximum power envelope supports efficient deployment in compatible data-centre and edge systems.
• Passive thermal design works with the airflow supplied by an approved server configuration.

Performance for Mixed Workloads:

*Tensor Core performance is shown with sparsity. NVIDIA states that these figures are halved without sparsity.

Supported Applications:

• Artificial intelligence and machine learning inference
• Generative AI workloads
• Computer vision and video analytics
• Video transcoding and streaming
• Rendering and visual computing
• Virtual desktops and virtual workstations
• Data science and data analytics

Technical Specifications:

HPE Server Compatibility

HPE lists the S0K89C for selected configurations across its ProLiant, Alletra, Apollo, edge and scale-up server families. These include supported models from the HPE ProLiant DL and ML Gen10 Plus, Gen11 and Gen12 ranges.

Important: GPU support can depend on the server model, processor platform, riser configuration, available PCIe slot, power arrangement, airflow requirements and any required enablement kit. Confirm compatibility in the current HPE QuickSpecs for your exact server configuration before ordering.

Add the HPE NVIDIA L4 accelerator to a compatible server to support AI, video, graphics and virtualisation workloads with a compact 24GB enterprise GPU.