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
1176.72 ms
total CPU used
9147.93 ms
speedup
7.77×
vs serial
efficiency
86.3%
of 9× ideal
stream spawn ms spawned@ work start@ work end@ work ms reap wait ms
0 (main) 0 15.48 15.49 1006.3 990.81 0
1 2.004 2.02 19.79 1132.99 1113.2 126.82
2 1.592 3.63 20.68 1050.37 1029.69 54.22
3 1.508 5.15 28.99 1075.34 1046.35 69.14
4 1.484 6.65 50.57 1116.18 1065.61 110.03
5 1.56 8.23 40.61 1076.68 1036.07 74.39
6 1.454 9.7 62.97 1173.7 1110.73 167.56
7 4.289 14 47.66 1088.75 1041.09 84.26
8 1.453 15.46 42.3 756.68 714.38 0.13
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.