CPU scaling benchmark
workers
3 +1 main
iters total
500M
125000000/stream
elapsed
1111.21 ms
total CPU used
4351.96 ms
speedup
3.92×
vs serial
efficiency
98%
of 4× ideal
| stream | spawn ms | spawned@ | work start@ | work end@ | work ms | reap wait ms |
|---|---|---|---|---|---|---|
| 0 (main) | 0 | 5.8 | 5.81 | 1090.56 | 1084.75 | 0 |
| 1 | 2.077 | 2.09 | 14.9 | 1103.28 | 1088.38 | 12.88 |
| 2 | 1.524 | 3.64 | 16.82 | 1103.64 | 1086.82 | 18.5 |
| 3 | 2.143 | 5.79 | 17.01 | 1109.02 | 1092.01 | 18.58 |
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.