CPU scaling benchmark
workers
3 +1 main
iters total
500M
125000000/stream
elapsed
1143.83 ms
total CPU used
4465.26 ms
speedup
3.9×
vs serial
efficiency
97.5%
of 4× ideal
| stream | spawn ms | spawned@ | work start@ | work end@ | work ms | reap wait ms |
|---|---|---|---|---|---|---|
| 0 (main) | 0 | 5.04 | 5.05 | 1112.29 | 1107.24 | 0 |
| 1 | 1.895 | 1.91 | 14.88 | 1128.27 | 1113.39 | 16.15 |
| 2 | 1.552 | 3.48 | 15.99 | 1140.96 | 1124.97 | 28.79 |
| 3 | 1.54 | 5.03 | 16.97 | 1136.63 | 1119.66 | 24.49 |
main
w1
w2
w3
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.