NVIDIA Tesla P40 24GB GDDR5 3840 CUDA Cores AI Deep Learning LLM Inference Passive Data Center GPU Accelerator Server Graphics Card 699-2G610-0200-100

Unlock powerful AI inference, deep learning, machine learning, virtualization, and high-performance computing capabilities with the NVIDIA Tesla P40 Data Center GPU Accelerator. Built on NVIDIA's Pascal architecture, the Tesla P40 delivers exceptional compute performance with 3,840 CUDA cores and 24GB of GDDR5 memory, making it a popular choice for AI research, LLM inference, computer vision, data analytics, HPC, and GPU-accelerated enterprise workloads.

The 699-2G610-0200-100 is the standard NVIDIA enterprise version of the Tesla P40 and is widely deployed in data centers, AI clusters, Supermicro servers, and HPE GPU-accelerated platforms. Its passive cooling design is optimized for rack-mounted servers with high-airflow environments.

Specifications

• Manufacturer: NVIDIA
• Model: Tesla P40
• Part Number: 699-2G610-0200-100
• GPU Architecture: Pascal (GP102)
• CUDA Cores: 3,840
• Memory Capacity: 24GB GDDR5
• Memory Interface: 384-Bit
• Memory Bandwidth: Up to 346 GB/s
• Base Clock: 1,303 MHz
• Boost Clock: Up to 1,531 MHz
• Interface: PCI Express 3.0 x16
• Form Factor: Full-Height, Dual-Slot
• Cooling: Passive Heatsink
• Max Power Consumption: 250W
• ECC Memory Support: Yes
• Power Connector: 8-Pin EPS (CPU-Style Connector)
• Display Outputs: None (Compute Accelerator)

Performance Features

• 12 TFLOPS Single Precision Performance
• Up to 47 TOPS INT8 Performance
• Optimized for AI Inference
• Deep Learning Ready
• Large Language Model (LLM) Inference
• Computer Vision Processing
• Machine Learning Acceleration
• GPU Virtualization Support
• TensorRT Compatible
• CUDA Compute Platform Support

Ideal Applications

• Artificial Intelligence (AI)
• Deep Learning
• Machine Learning
• LLM Inference
• Stable Diffusion
• Computer Vision
• Data Analytics
• High Performance Computing (HPC)
• Scientific Computing
• Virtual Desktop Infrastructure (VDI)
• Virtualization
• Video Transcoding

Compatible Server Platforms

• HPE ProLiant DL380 Gen9
• HPE ProLiant DL380 Gen10
• HPE ProLiant DL385 Gen10
• HPE Apollo Systems
• Supermicro GPU Servers
• Dell PowerEdge R730
• Dell PowerEdge R740
• Dell PowerEdge R750
• Lenovo ThinkSystem SR650
• Cisco UCS Servers
• Other PCIe x16 servers supporting passive GPU accelerators

Compatible Part Numbers

• 699-2G610-0200-100
• 699-1G610-0300-100
• Tesla P40
• Tesla P41
• Q0V80A
• 870919-001
• 870919-B21
• 913985-001
• PG610
• NV-P40-24GB

Features

• Enterprise-grade NVIDIA Data Center GPU
• 24GB High-Capacity GDDR5 Memory
• 3,840 CUDA Cores
• Passive Cooling Design
• ECC Memory Capability
• AI and Deep Learning Optimized
• Large Memory Pool for Model Inference
• Designed for 24/7 Data Center Operation
• Supports NVIDIA CUDA, TensorRT, and vGPU Technologies

Condition

Tested and fully functional. Pulled from a working enterprise environment. May show minor cosmetic signs of handling that do not affect performance.

Package Includes

• 1x NVIDIA Tesla P40 24GB GDDR5 Passive GPU Accelerator (699-2G610-0200-100)

Other Related Part Numbers:

699-2G610-0200-100, 699-1G610-0300-100, Tesla P40, Tesla P41, Q0V80A, 870919-001, 870919-B21, 913985-001, PG610, NV-P40-24GB, 699-2G610-0000-000, 699-2G610-0100-000.

Note: The Tesla P40 uses a passive cooling solution and requires installation in a server chassis with adequate airflow. The 699-1G610-0300-100 and 699-2G610-0200-100 variants share the same Tesla P40 architecture, 24GB GDDR5 memory, and 3,840 CUDA cores, making them compatible for the same AI, inference, and data center workloads.