Would it be possible to run 2 SD instances in parallel on a single M1/M2 chip?

One on the GPU and another on the ML core?