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
1167.83 ms
total CPU used
9729.32 ms
speedup
8.33×
vs serial
efficiency
92.6%
of 9× ideal
stream spawn ms spawned@ work start@ work end@ work ms reap wait ms
0 (main) 0 28.27 28.28 1157.05 1128.77 0
1 1.974 1.99 18.74 1143.26 1124.52 0.13
2 1.503 3.52 22.45 1098.22 1075.77 0.2
3 1.415 4.95 21.53 1030.94 1009.41 0.22
4 4.311 9.28 32.88 1112.94 1080.06 0.23
5 1.625 10.92 39.05 1070.68 1031.63 0.24
6 1.712 12.65 41.97 1117.69 1075.72 0.25
7 1.539 14.21 40.29 1139.82 1099.53 0.26
8 14.03 28.26 61.07 1164.98 1103.91 8.03
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.