Does Ray support message passing like MPI?

Ray Core
1

How severe does this issue affect your experience of using Ray?

  • Medium: It contributes to significant difficulty to complete my task, but I can work around it.

Hi! Recently I want to run a CFD application on Ray, it need to divide the grid into several sub-grids, and calculate on different workers.

But the questions is: during every calculation step of a sub-grid, it need to communicate with its neighbor, is there any way to send and receive message between workers? Or do I have to split the task and use ray.put() and ray.get()?

def sub_worker():
  fe[1:-1, 1:-1] = 0.5 * (hc[1:-1, 1:-1] + hc[1:-1, 2:]) * self.u[1:-1, 1:-1]
  fn[1:-1, 1:-1] = 0.5 * (hc[1:-1, 1:-1] + hc[2:, 1:-1]) * self.v[1:-1, 1:-1]
  # need to exchange data with neighbor grid
  fe = self.enforce_boundaries()
  fn = self.enforce_boundaries()
  # continue when data exchange are done
  dh_new[1:-1, 1:-1] = -(
              (fe[1:-1, 1:-1] - fe[1:-1, :-2]) / dx
              + (fn[1:-1, 1:-1] - fn[:-2, 1:-1]) / dy
          )
  # potential vorticity
   q = q.at[1:-1, 1:-1].mul(
        1.0 / (0.25 * (hc[1:-1, 1:-1] + hc[1:-1, 2:] + hc[2:, 1:-1] + hc[2:, 2:]))
    )
  # another data exchange
   q = enforce_boundaries()

   du_new = du.at[1:-1, 1:-1].set(
        -GRAVITY * (h[1:-1, 2:] - h[1:-1, 1:-1]) / dx
        + 0.5
        * (
            q[1:-1, 1:-1] * 0.5 * (fn[1:-1, 1:-1] + fn[1:-1, 2:])
            + q[:-2, 1:-1] * 0.5 * (fn[:-2, 1:-1] + fn[:-2, 2:])
        )
    )
2

I think you can use ray actor? Actors — Ray 3.0.0.dev0