Detailed Specifications
HPE 872323-001 NVIDIA Tesla P40 24 GB Data-Centre GPU
Overview
The HPE 872323-001 is a GPU accelerator module based on the NVIDIA Tesla P40 architecture. It features 24 GB of GDDR5 memory and 3,840 CUDA cores, delivering high compute performance for AI inference, machine learning, virtualization, and other data-centre workloads. Its passive (fanless) cooling design is optimized for integration into HPE ProLiant servers with sufficient chassis airflow.
Key Features
24 GB GDDR5 memory for large datasets and models
3,840 CUDA cores for heavy parallel computation
Passive (heatsink) design × no dedicated fan
PCIe 3.0 ×16 interface for server compatibility
Supports NVIDIA virtualization (vGPU) for shared GPU workloads
Technical Specifications
| Specification | Details |
|---|
| Brand | HPE / NVIDIA |
| Part Number | 872323-001 |
| Product Type | GPU Accelerator / Computational Module |
| GPU | NVIDIA Tesla P40 |
| Memory | 24 GB GDDR5 |
| CUDA Cores | 3,840 |
| Memory Bandwidth | ~346 GB/s |
| Interface | PCI Express 3.0 ×16 |
| Power Consumption | 250 W typical |
| Cooling | Passive (heatsink) |
| Peak Single-Precision Performance | ~12 TFLOPS |
| ECC Memory | Yes |
| Virtualization Support | NVIDIA vGPU / GRID |
| Compatible Servers | HPE ProLiant Gen9 / Gen10 servers |
| Dimensions | ~10.5″ × 4.4″ (L × H) |
| Weight | ~2.35 lb |
| Warranty | HPE standard parts warranty |
