Parallel Programming — A1 (C++11 Threads)

C++11 multithreaded implementations of four workloads with per-thread timing and Slurm-ready builds. Focus: work partitioning, synchronization, cache locality, and false-sharing avoidance.

What’s implemented

• Monte Carlo π (pi_calculation_parallel.cpp)
  Parallel random-point generation (≈ n/T per thread), per-thread counts & timers, global reduction.
• Triangle Counting (triangle_counting_parallel.cpp)
  Vertex-range partitioning over CSR/CSC; thread-local counts → single reduction; unique triangles reported.
• PageRank w/ Locks (page_rank_parallel.cpp)
  Push model, double buffers, barrier between scatter/gather, std::mutex for shared updates; per-thread time + PR sum.
• PageRank w/ Atomics (page_rank_parallel_atomic.cpp)
  Lock-free accumulation via std::atomic/CAS; same iteration/barrier structure and metrics.

Techniques & practices

• C++11 std::thread, std::mutex, CustomBarrier, std::atomic
• Thread-local stats, padded structs to mitigate false sharing
• CSR/CSC traversal for sequential memory access
• Fine-grained timers for total & per-thread regions
• Slurm-friendly CLI: --nWorkers, --nPoints/--nIterations, --inputFile