Items purchased through Cisco Outlet are remanufactured. Conditions may vary from the visual depictions on the site.
UCSC-GPU-P4-RF
NVIDIA P4 PCIE 75W 8GB REMANUFACTURED
Inventory quantities are subject to availability and are not guaranteed until the order is placed.
The UCSC-GPU-P4-RF is a professionally remanufactured Cisco UCS C-Series NVIDIA Tesla P4 GPU accelerator card for Cisco UCS rack servers, providing hardware-accelerated inference processing for artificial intelligence and machine learning applications. The NVIDIA Tesla P4 uses INT8 precision inference optimized silicon delivering up to 22 TOPS of INT8 inference throughput in a low-profile single-slot PCIe form factor, making it suitable for high-density GPU inference server deployments where multiple P4 cards can be installed per server for scalable AI inference capacity.
This remanufactured GPU accelerator has been tested for NVIDIA CUDA compute functionality, INT8 inference performance, PCIe interface stability, and Cisco UCS C-Series server compatibility. Each unit meets Cisco UCS GPU certification standards for production AI server deployments.
Key Features:
• NVIDIA Tesla P4 GPU with up to 22 TOPS INT8 inference throughput
• Low-profile single-slot PCIe form factor for high-density multi-GPU server deployments
• Optimized for AI inference including computer vision and NLP applications
• Cisco UCS C-Series certified for full hardware and thermal compatibility
• NVIDIA vGPU software support for virtualized GPU inference deployments
• Passive cooling design for data center server rack thermal management
This refurbished GPU is ideal for enterprise AI teams deploying inference-optimized compute on Cisco UCS C-Series servers for production machine learning workloads. By choosing a remanufactured Cisco UCS NVIDIA P4 GPU, you scale AI inference capacity at significantly lower per-GPU cost than new hardware.