root@OnePlus3:/ # perf stat --repeat 10 hackbench -g 20 -l 1000
Running in process mode with 20 groups using 40 file descriptors each (== 800 tasks)
Each sender will pass 1000 messages of 100 bytes
Time: 11.244
Running in process mode with 20 groups using 40 file descriptors each (== 800 tasks)
Each sender will pass 1000 messages of 100 bytes
Time: 10.819
Running in process mode with 20 groups using 40 file descriptors each (== 800 tasks)
Each sender will pass 1000 messages of 100 bytes
Time: 11.234
Running in process mode with 20 groups using 40 file descriptors each (== 800 tasks)
Each sender will pass 1000 messages of 100 bytes
Time: 11.587
Running in process mode with 20 groups using 40 file descriptors each (== 800 tasks)
Each sender will pass 1000 messages of 100 bytes
Time: 11.320
Running in process mode with 20 groups using 40 file descriptors each (== 800 tasks)
Each sender will pass 1000 messages of 100 bytes
Time: 12.058
Running in process mode with 20 groups using 40 file descriptors each (== 800 tasks)
Each sender will pass 1000 messages of 100 bytes
Time: 11.242
Running in process mode with 20 groups using 40 file descriptors each (== 800 tasks)
Each sender will pass 1000 messages of 100 bytes
Time: 11.044
Running in process mode with 20 groups using 40 file descriptors each (== 800 tasks)
Each sender will pass 1000 messages of 100 bytes
Time: 11.155
Running in process mode with 20 groups using 40 file descriptors each (== 800 tasks)
Each sender will pass 1000 messages of 100 bytes
Time: 11.344

 Performance counter stats for 'hackbench -g 20 -l 1000' (10 runs):

      44942.575652 task-clock                #    3.875 CPUs utilized            ( +-  0.86% )
            726789 context-switches          #    0.016 M/sec                    ( +-  4.32% )
              6425 CPU-migrations            #    0.000 M/sec                    ( +-  4.88% )
             86711 page-faults               #    0.002 M/sec                    ( +-  0.15% )
       82509993897 cycles                    #    1.836 GHz                      ( +-  0.80% )
     <not counted> stalled-cycles-frontend
     <not counted> stalled-cycles-backend
       36576667060 instructions              #    0.44  insns per cycle          ( +-  0.42% )
     <not counted> branches
         109895762 branch-misses             #    0.00% of all branches          ( +-  0.88% )

      11.598168468 seconds time elapsed                                          ( +-  0.88% )