Detailed Specifications
HPE P05913-001 NVIDIA Tesla V100 32 GB PCIe Accelerator
Overview
The HPE P05913-001 is a 32 GB Tesla V100 GPU accelerator built for PCIe, aimed at high-performance computing, deep learning, and parallel workloads. It delivers strong double-precision and tensor performance using NVIDIA's Volta GV100 architecture, making it ideal for AI training, simulations, and scientific computing. The card is passive-cooled and requires sufficient airflow in server systems, and is compatible with HPE ProLiant and Apollo platforms. It integrates with CUDA, cuDNN, and other NVIDIA software frameworks for compute-intensive environments.
Key Features
32 GB HBM2 memory with 4096-bit bus
640 tensor cores for deep learning workloads
PCI Express 3.0 x16 interface
Passive cooling for data center airflow environments
ECC memory support for data integrity
Technical Specifications
| Specification | Details |
|---|
| Brand | HPE / NVIDIA |
| Part Number | P05913-001 |
| Product Type | Computational GPU Accelerator |
| GPU | NVIDIA GV100 (Volta) |
| CUDA Cores | 5,120 |
| Tensor Cores | 640 |
| Memory Installed | 32 GB HBM2 |
| Memory Interface | 4,096-bit |
| Memory Bandwidth | ~900 GB/s |
| Bus Interface | PCI Express 3.0 x16 |
| Power Consumption | 250 W TDP |
| Form Factor | Dual-slot, full-height PCIe |
| Performance (FP64) | ~7 TFLOPS |
| Performance (FP32) | ~14 TFLOPS |
| Weight | ~2.64 lbs |
| Thermal Solution | Passive Heatsink |
| Supported APIs | CUDA, OpenCL, OpenACC |
| ECC Support | Yes |
