CPU scaling benchmark

W=0 W=1 W=2 W=3 W=4 W=6 W=8 W=9 W=10 W=12 W=16 iters: 100M 500M 1B 2B
workers
9 +1 main
iters total
500M
50000000/stream
elapsed
1172.6 ms
total CPU used
10330.05 ms
speedup
8.81×
vs serial
efficiency
88.1%
of 10× ideal
stream spawn ms spawned@ work start@ work end@ work ms reap wait ms
0 (main) 0 30.66 30.67 1154.56 1123.89 0
1 2.001 2.02 14.97 1055.13 1040.16 0.13
2 1.649 3.69 27.82 1072.11 1044.29 0.18
3 1.568 5.28 22.68 978.1 955.42 0.2
4 4.797 10.11 42.55 1142.48 1099.93 0.21
5 1.609 11.74 48.01 1082.64 1034.63 0.22
6 1.406 13.17 41.53 993.89 952.36 0.23
7 1.517 14.71 60.61 1057.55 996.94 0.24
8 1.432 16.16 42.82 1013.02 970.2 0.25
9 14.471 30.65 57.67 1169.9 1112.23 15.45
main
w1
w2
w3
w4
w5
w6
w7
w8
w9
    fork+handshake      CPU work      parent reap wait
what this measures
Each stream runs a tight integer LCG loop — working set is one CPU register, no memory access, no shared data. Speedup = sum(stream CPU time) / wall-clock elapsed. Efficiency = speedup / (workers+1). 100% efficiency means perfect linear scaling; less than 100% is the cost of serial fork setup, reap tail, SMT/core contention.