Hello Gilles, Assuming you run one MPI task per unikernel, and two unikernels share > nothing, > it means that inter-node communication cannot be performed via shared > memory or kernel feature > (such as xpmem or knem). That also implies communication are likely using > the loopback interface > which is much slower.
Let me elaborate on it. I have one VM in which I have the unikernel. The VM has many cores and each core has a single thread. Each thread runs a MPI task. Each MPI task relies on shared-memory to communicate, i.e., VirtIO. This is a shared-nothing unikernel because each core runs an instance of the kernel that shares nothing with the other cores in terms of memory and resources. Regards, Matias.
