Advantages of remote NVMe storage:
Remote storage over network can offers many advantages for your FPGA-based systems such as:
- Simplify FPGA hardware with compactness, lower power and heat, and with cost-competitive FPGA devices
- Accessing network and storage through with the same 10G/25G or 100G Ethernet interface
- Benefit from consolidated, secure and scalable storage for enhanced system performance and sharing resources
Why need NVMe/TCP Initiator IP Core on FPGA:
NVMe/TCP Initiator requires intensive computation power of CPU to handle NVMe-oF and TCP/IP protocol especially for the application that require very high performance NVMe storage access over 10G/25G and 100G network. NVMe/TCP Initiator IP Core on FPGA can drastically simplify NVMe Storage access for various such of application such as CPU offloading for Cloud & Data center and enabling remote NVMe storage access for FPGA-based systems.
Remote storage for AI application development
High-capacity and high-performance storage is essential for AI application development such as video surveillance, medical image analysis, signal or pattern analysis such as mobile network, radio, radar and lidar. These applications require vast amounts of data generated by their sensors and cameras to be stored and processed in real-time to ensure the quality of the training dataset. The NVMeTCP-IP core can simplify the development of data storage systems for these AI-based hardware applications.
Fully CPU offload NVMeTCP Initiator for Data Center
The FPGA-based accelerator delivers very low-latency and high throughput performance over 100G~400G line rate for infrastructure workloads in the data center. Together with the NVMeTCP-IP core, near line rate NVMe/TCP performance is possible. It frees up server CPUs from infrastructure tasks to mainly focus on running business-critical applications, improving overall throughput and latency for your data center workloads.
NVMeTCP-IP (NVMe/TCP 10G/25G IP Core) is the standalone host side NVMe Over Fabric (NVMe/TCP) controller with no CPU and external memory required. Enabling very high-performance remote access to NVMe-oF Storage Server by simple CPU <> FPGA DMA interface. Allowing Host CPU get access to the NVMe/TCP Target storage through FPGA Accelerator with very low CPU usage to maximize storage workload performance.