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
8 +1 main
iters total
500M
55555555/stream
elapsed
1176.75 ms
total CPU used
8924.95 ms
speedup
7.58×
vs serial
efficiency
84.2%
of 9× ideal
stream spawn ms spawned@ work start@ work end@ work ms reap wait ms
0 (main) 0 13.05 13.06 781.86 768.8 0
1 2.114 2.13 27.62 1121.43 1093.81 339.65
2 1.444 3.59 24.99 1099.21 1074.22 317.46
3 1.446 5.04 28.42 1050.06 1021.64 268.34
4 1.424 6.48 35.97 979.89 943.92 198.23
5 1.442 7.93 61.08 1173.97 1112.89 392.23
6 1.516 9.46 51.09 807.56 756.47 41.87
7 1.576 11.05 37.58 1108.14 1070.56 331
8 1.99 13.05 51.86 1134.5 1082.64 352.76
main
w1
w2
w3
w4
w5
w6
w7
w8
    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.