Goal: Benchmark and optimize PySCF workloads for quantum chemistry calculations.

Methods:

• CPU profiling with cProfile and line_profiler
• Thread scaling with OMP_NUM_THREADS
• Memory profiling
• Optional GPU comparison with GPU4PySCF

Workloads:

• RHF
• DFT
• FCIDUMP generation
• AO-to-MO integral transformation

Metrics:

• wall time
• SCF iteration time
• memory usage
• speedup
• scaling with basis size

Data:

• from CCCBDB experimental geometry data