Parallel Computing for Data Science

84 CHAPTER 4. SHARED MEMORY: R

# r e c e i v e from worker 2

y <− mpi . r e c v . Robj ( tag =0, so u rc e =2)

to run on process 5. By contrast, in a shared-memory environment, the

variables x and y would be shared, and the programmer would merely

write

y <− x

What a difference! Now that x and y are shared by the processes, we can

access them directly, making our code vastly simpler.

Note carefully that we are talking about human efficiency here, not machine

efficiency. Use of shared memory can greatly simplify our code, with far less

clutter, so that we can write and debug our program much faster than we

could in a message-passing environment. That doesn’t ...