generate.py will generate masks for multiple ranks to bind the ranks to cores for srun
generate.py 3:2:0:0 generates masks for 3 ranks, 2 threads per rank, with starting core 0 and without any stride.
generate.py 3:2:0:0 2:8 generates masks for 3 ranks, 2 threads per rank, with starting core 0 and without any stride, for each 8 cores, 2 times
generate.py 3:2:0:0 2:8 4:24 generates same above(generate.py 3:2:0:0 2:8) masks for each 24 cores, 4 times
generate.py 3:2:0:0 2:8 4:24 2:96 generates same above(generate.py 3:2:0:0 2:8 4:24) masks for each 96 cores, 2 times
generate.py 3:2:0:0 2:8 4:24 2:96 It can be used to mimic AMD 9454 using AMD 9654, 2 socket node
six cores per CCD on the first two CCDs of each NUMA node on 2 sockets of 9654
verify.py will converts hexadecimal mask values to integer core values
verify.py --cpu-bind=verbose,mask_cpu:0x00000003,0x0000000c,0x00000030
output: --cpu-bind=verbose,mask_cpu: [0-1], [2-3], [4-5]
verify.py --cpu-bind=mask_cpu:0x00000003,0x0000000c,0x00000030,0x00000300,0x00000c00,0x00003000
output: --cpu-bind=mask_cpu: [0-1], [2-3], [4-5], [8-9], [10-11], [12-13]