Working with Multiple Hands
summary
This session shows mpi4py and mpiexec used to run the same Python process across cores or machines, demonstrates a 4–5 process hello world and a parallel area-under-curve integrator that aggregates results with broadcast/reduce, and notes that small-message buffering can make sends return before receipts. It uses a machine file to run on multiple IPs, highlights blocking send/recv behavior, reports collective ops working well, and warns that communication cost and nonparallelizable code often kill speedups.
AI-written summary · about
chapters (8)
- 0:00Start
- 7:29Terminologies
- 9:14MPI for Python?
- 17:21More Helpful Methods
- 23:00Adding Complexity: Collective Operations
- 24:36Calculating Area Revisited
- 25:42Distributed Programming with MPI
- 29:18Ready for a Trivia?
related talks
10mMPI Meets Python: How to Use MPI in the Python Universe
33mLet's Run Python on a Supercomputer!
30mParallel computing in Python: Current state and recent advances
30mParallelization of code in Python for beginners
34mUnderstanding of distributed processing in Python
31mPython Multithreading and Multiprocessing: Concurrency and Parallelism
10mRedlining Your CPU for Fun and Profit: Threading and Multiprocessing in Python
39mUnlocking the Parallel Universe: Subinterpreters and Free-Threading in...