NVIDIA Tesla K80 Accelerator Card 24GB GDDR5 PCI-E 3.0 x16 Graphics Card

NVIDIA Tesla K80 Accelerator Card is a high-performance computing graphics card designed for data centers and computationally intensive applications. Equipped with 24GB GDDR5 memory and utilizing PCI-E 3.0 x16 connectivity, it provides powerful parallel processing with dual GPUs, enabling exceptional performance for deep learning, scientific computing, and engineering simulations. The K80 is built for efficient large-scale processing and is widely used in AI, machine learning, and high-performance computing environments.

GPU Configuration
Dual GPUs: The K80 has two GK210 GPUs on a single board, functioning as a dual-GPU setup.
CUDA Cores: 4, 992 total (2, 496 per GPU)
Clock Speed:
Base Clock: 562 MHz
Boost Clock: Up to 875 MHz (variable based on workload and temperature)

Memory
Total Memory: 24 GB GDDR5 (12 GB per GPU)
Memory Speed: 5.0 Gbps
Memory Bandwidth: 480 GB/s (240 GB/s per GPU)
Memory Interface Width: 384-bit per GPU

Performance
Single-Precision Performance: 8.73 TFLOPS (combined for both GPUs)
Double-Precision Performance: 2.91 TFLOPS (combined for both GPUs)

Power and Thermal
TDP (Thermal Design Power): 300W
Power Connectors: None on card; power provided through PCIe slot

Features and Capabilities
ECC Memory Support: Yes (Error-Correcting Code support for increased reliability)
GPU Boost: Yes, with dynamic boost to optimize performance based on thermal headroom
Compute Capability: 3.7
API Support: CUDA, OpenCL, DirectCompute