Disaggregation of FPGA accelerators using Eideticom NoLoad and NVMe-over-Fabrics

In this demo, Eideticom describes NoLoad which is a storage and compute acceleration platform targeting Xilinx FPGAs. NoLoad’s NVMe compatible interface provides the ability for acceleration functions to be accessed from remote clients via NVMe-over-Fabrics. The NoLoad architecture is then used to access Erasure Code accelerators across an RDMA-based NVME-over-Fabrics Ethernet network.