How vchaitrad/pgc-lab-evaluation-1 works
This lab compares vector addition across serial CPU, OpenMP, and CUDA implementations, then presents execution-time comparisons. The main workflow is to run the CPU and GPU benchmarks for three array sizes, capture their timings, and generate a comparison chart. The OpenMP implementation is shown in source; CUDA behavior and run-to-result relationships are documented in the README, but the CUDA source was not excerpted. The chart script uses embedded timing values rather than reading the result files.
Main entry points: cpu_vs_gpu.cu CPU + CUDA · cpu_openmp.c Serial + OpenMP · plot_graph.py Timing chart generator · openmp_results.txt OpenMP timings
Components 7
Benchmarks
- Serial + OpenMP
cpu_openmp.c - CPU + CUDA
cpu_vs_gpu.cu - Tesla T4 GPU
Results
- OpenMP timings
openmp_results.txt - CUDA run timings
cuda_results.txt
Reporting
- Timing chart generator
plot_graph.py
Other components
- Lab user
Connections 6
- Lab user → Serial + OpenMP: runs benchmark
README.md - Lab user → CPU + CUDA: runs benchmark
README.md - CPU + CUDA → Tesla T4 GPU: runs kernel
README.md - Serial + OpenMP → OpenMP timings: writes timings
README.md - CPU + CUDA → CUDA run timings: writes timings
README.md - Timing chart generator → Lab user: saves chart
plot_graph.py