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
6 +1 main
iters total
100M
14285714/stream
elapsed
274.55 ms
total CPU used
1293.96 ms
speedup
4.71×
vs serial
efficiency
67.3%
of 7× ideal
stream spawn ms spawned@ work start@ work end@ work ms reap wait ms
0 (main) 0 20.65 20.66 251.81 231.15 0
1 2.395 2.41 16.76 207.29 190.53 0.13
2 1.887 4.32 17.97 164.42 146.45 0.19
3 1.854 6.21 19.49 148.86 129.37 0.2
4 2.063 8.29 61.31 270.95 209.64 19.29
5 10.045 18.36 44.36 247.89 203.53 0.22
6 2.266 20.64 48.68 231.97 183.29 0.25
main
w1
w2
w3
w4
w5
w6
    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.