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
1158.93 ms
total CPU used
8841.72 ms
speedup
7.63×
vs serial
efficiency
84.8%
of 9× ideal
stream spawn ms spawned@ work start@ work end@ work ms reap wait ms
0 (main) 0 13.46 13.47 895.84 882.37 0
1 2.139 2.16 21.2 885.49 864.29 0.14
2 1.593 3.77 18.71 1143.64 1124.93 248
3 1.559 5.35 44.01 1034.16 990.15 143.82
4 1.511 6.88 51.77 1062.51 1010.74 169.78
5 1.451 8.35 27.4 829.32 801.92 0.2
6 1.542 9.91 41.76 1148.35 1106.59 253.1
7 1.943 11.86 45.19 1004.33 959.14 114.11
8 1.564 13.44 54.27 1155.86 1101.59 260.14
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.