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
1191.07 ms
total CPU used
9208.54 ms
speedup
7.73×
vs serial
efficiency
85.9%
of 9× ideal
stream spawn ms spawned@ work start@ work end@ work ms reap wait ms
0 (main) 0 19.62 19.63 1007.01 987.38 0
1 2.125 2.14 14.44 1037.93 1023.49 33.51
2 1.499 3.66 15.56 1147.48 1131.92 140.59
3 2.612 6.29 20.6 975.03 954.43 0.13
4 1.723 8.03 55.88 914.32 858.44 0.19
5 7.136 15.18 43.24 1188.1 1144.86 181.23
6 1.442 16.64 41.02 1025.72 984.7 18.86
7 1.476 18.13 40.6 1153.47 1112.87 146.64
8 1.457 19.61 43.24 1053.69 1010.45 51.61
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.