Tomas Plachetka

Concurrent and Distributed Programming and Systems 2, 2022/2023

Thursday 9:50-13:00

The evaluation of work during the semester (homeworks) forms 50% of the final grade, at least 40% are required.

• Threads. Thread state diagram. Shared variables, atomicity of memory accesses, critical regions. Decker's and Peterson's algorithms, assumptions on memory accesses. POSIX threads (pthread), mutexes, condition variables.
  • Lecture notes
  • Open Group pthread reference
  • A pthread tutorial
  • Another pthread tutorial
  • Historical recommendation (when to use threads and when not, 1995)
• OCCAM
  • Lecture notes
  • D. Pountain: A tutorial introduction to OCCAM programming
  • G. Jones, M.Goldsmith: Programming in OCCAM 2
  • Inmos OCCAM libraries
  • KROC/OccamPI
• Channel vs. point-to-point message passing
  • Channel-Less Process Communication
• PVM, Parallel Virtual Machine
  • PVM home (ORNL)
  • PVM book
• MPI, Message Passing Interface
  • MPI Forum
  • MPICH
• HPC, High-Performance Computing
  • Top 500 HPC clusters
  • Beowulf clusters (low-cost parallel computing)
• Thread safety in communication libraries
  • Quasi Thread-Safe PVM and MPI

Homeworks

• HW1: Implement an OCCAM program for one-dimensional simulation of LIFE.
• HW2: Implement a PVM program for one-dimensional simulation of LIFE.
• HW3: Implement an MPI (MPICH) program for one-dimensional simulation of LIFE.
• HW4: Report programming environments used on Top500 HPC clusters (not necessarily through all 500 clusters)
• HW5: Implement a pingpong benchmark in MPI (MPICH).
• HW6: Implement a symmetrical threaded pingpong benchmark in MPI (MPICH). Run experiments to measure min, avg, max roundtrip time for 2 processes.