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
1150.78 ms
total CPU used
9741.56 ms
speedup
8.47×
vs serial
efficiency
84.7%
of 10× ideal
stream spawn ms spawned@ work start@ work end@ work ms reap wait ms
0 (main) 0 41.04 41.05 1145.93 1104.88 0
1 2.254 2.27 16.78 892.8 876.02 0.17
2 1.762 4.06 18.73 1046.28 1027.55 0.23
3 3.608 7.69 20.69 1091.9 1071.21 0.24
4 2.134 9.84 62.07 1148.02 1085.95 2.23
5 6.947 16.82 71.57 1060.7 989.13 0.25
6 6.014 22.84 53.99 752.19 698.2 0.28
7 1.805 24.67 49.63 845.25 795.62 0.29
8 12.227 36.91 70.95 1125.95 1055 0.31
9 4.082 41.02 96.52 1134.52 1038 0.33
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.