CPU scaling benchmark
workers
3 +1 main
iters total
500M
125000000/stream
elapsed
1140.9 ms
total CPU used
4383.09 ms
speedup
3.84×
vs serial
efficiency
96%
of 4× ideal
| stream | spawn ms | spawned@ | work start@ | work end@ | work ms | reap wait ms |
|---|---|---|---|---|---|---|
| 0 (main) | 0 | 4.77 | 4.78 | 1086.15 | 1081.37 | 0 |
| 1 | 1.819 | 1.83 | 16.01 | 1104.23 | 1088.22 | 18.24 |
| 2 | 1.389 | 3.24 | 16.25 | 1138.04 | 1121.79 | 52.02 |
| 3 | 1.504 | 4.76 | 19.89 | 1111.6 | 1091.71 | 25.58 |
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.