NVIDIA P40

Overview

The NVIDIA Tesla P40 is a high-performance GPU accelerator designed to deliver exceptional throughput for data-intensive workloads, including deep learning inference, high-performance computing (HPC), and virtualized graphics applications. Built on the NVIDIA Pascal™ architecture, it offers a significant leap in performance and efficiency compared to its predecessors.

Features

• High Throughput for Inference: Delivers up to 47 TOPS (Tera Operations Per Second) for INT8 operations, making it ideal for real-time deep learning inference tasks.
• Massive Memory Capacity: Equipped with 24 GB of GDDR5 memory, enabling the handling of large datasets and complex models with ease.
• Advanced Virtualization Support: Supports up to 24 virtual GPU instances (1 GB profile), facilitating the deployment of multiple virtual workstations or desktops in a shared environment.
• Dual-Slot PCIe 3.0 Interface: Ensures high bandwidth and compatibility with a wide range of servers and workstations.
• Passive Cooling Design: Designed for efficient heat dissipation, suitable for systems with adequate airflow.

Specifications

• GPU Architecture: NVIDIA Pascal™
• CUDA Cores: 3,840
• Memory: 24 GB GDDR5
• Memory Bandwidth: 346 GB/s
• Form Factor: Dual-slot, PCIe 3.0 x16
• Maximum Power Consumption: 250 W
• Thermal Solution: Passive cooling (requires adequate system airflow)
• Virtual GPU Profiles Supported: 1 GB, 2 GB, 3 GB, 4 GB, 6 GB, 8 GB, 12 GB, 24 GB
• Maximum Virtual GPU Instances: 24 (1 GB profile)
• Video Output Interface: DisplayPort or HDMI
• Ideal Use Cases: Deep learning inference, high-performance computing, virtualized graphics, and AI workloads.