Found during epic #324 Phase 17 (#355).
AMICANative(threads=...) sets OMP_NUM_THREADS, but the binary calls omp_set_num_threads(max_threads) from input.param (amica15.f90:95-100). So the binary always runs with max_threads, 10 by default. With threads=2 its stdout still reports "using 10 threads".
Expected: threads sets the binary's thread count, or the parameter is removed in favor of max_threads. The docs should say which one controls it.
benchmarks/reproduce_table1.py passes both threads and max_threads, so its runs used the requested count.
Found during epic #324 Phase 17 (#355).
AMICANative(threads=...)setsOMP_NUM_THREADS, but the binary callsomp_set_num_threads(max_threads)frominput.param(amica15.f90:95-100). So the binary always runs withmax_threads, 10 by default. Withthreads=2its stdout still reports "using 10 threads".Expected:
threadssets the binary's thread count, or the parameter is removed in favor ofmax_threads. The docs should say which one controls it.benchmarks/reproduce_table1.pypasses boththreadsandmax_threads, so its runs used the requested count.