Innova-2 Flex

Advanced Programmability Delivers Acceleration and Performance

The need to accelerate applications such as security and access to storage spans AI, media and entertainment, cloud, and more. The NVIDIA® Mellanox® Innova-2 Flex Open Programmable SmartNIC natively offers industry-leading accelerations, such as hardware support for RoCE, overlay networks, stateless offload engines, and NVIDIA GPUDirect®. Similarly, the on-board FPGA further increases flexibility by offloading the CPU of processes within custom applications.

Open Programmable SmartNIC

Maximizing network efficiency and scalability, the NVIDIA Mellanox Innova-2 Flex Open Programmable SmartNIC offers customers an open platform to develop custom-made offloads for a range of markets, including Storage, High-Performance Computing (HPC), Artificial Intelligence, Security, and more. By combining the advanced ConnectX®-5 InfiniBand and Ethernet network controller ASIC with a state-of-the-art FPGA on board creating the ideal infrastructure for developers to implement their own applications to achieve hardware-like performance with minimum time-to-market.

NVIDIA Mellanox Innova-2 Flex Open Programmable SmartNIC


Achieving Maximum Efficiency



The NVIDIA RoCE technology encapsulates packet transport over Ethernet and lowers CPU load to enable a high bandwidth and low latency network infrastructure for  networking and storage-intensive applications.



The groundbreaking NVIDIA ASAP² technology delivers innovative SR-IOV and VirtIO acceleration by offloading Open vSwitch datapath from the host’s CPU to the adapter to enable extreme performance and scalability.



ConnectX adapters leverage SR-IOV to separate access to physical resources and functions in virtualized environments. This diminishes I/O overhead and allows adapters to maintain near non-virtualized performance.

Key Features

  • Dual-port 25Gb/s Ethernet or dual-port EDR InfiniBand / 100Gb/s Ethernet
  • Complete Mellanox and Xilinx FPGA for “look-aside” accelerations
  • Low latency RDMA and RDMA over Converged Ethernet (RoCE)
  • OVS and Erasure Coding offloads
  • NVIDIA Mellanox PeerDirect® communication acceleration
  • End-to-end QoS and congestion control
  • Hardware-based I/O virtualization
  • RoHS compliant



  • Versatile adapter combining ability to accelerate computational tasks
  • Leverages single PCIe slot for providing “look-aside” accelerations
  • Suitable for FPGA as a service deployment
  • Suitable for advanced security and compression accelerations
  • Industry leading throughput and latency for data center applications
  • Efficient I/O consolidation, lowering data center costs and complexity
  • Open platform for easy development of user applications and deployment to the FPGA


  • Product Documents
  • Firmware
  • Software Driver

We're here to help you build the most efficient, high performance network.

SmartNIC Selector

Academy Online Courses

Ready to Purchase