Virtualizing GPUs over TCP/IP introduces latency and context switches. Our RDMA communicator for GVirtuS cuts Matrix Multiplication time by 5.5x and SAXPY by 82%, proving Infiniband is the future of low-latency GPGPU remoting.
🌟 Follow us on LinkedIn! https://www.linkedin.com/company/clever-project/?viewAsMember=true
🌟 Check the updates from the website: www.cleverproject.eu
🌟 Full paper in: https://zenodo.org/records/14717622