📊 Performance Comparison Charts
Throughput Comparison
Items processed per second across different implementations
Latency Comparison
Average message latency with error bars
Because life is too precious to waste on sluggish thread synchronisation. Built in Iceland's long winter nights, fastcond understands that your threads should work efficiently so you can enjoy the important things: reading poetry under starry skies, sipping rum-spiked cocoa by the fireplace, or watching the aurora dance across the darkness.
Built using nothing but POSIX semaphores and Nordic engineering sensibilities, fastcond delivers performance that lets you spend less time debugging race conditions and more time living.
Two varieties of excellence:
Rather amusing fact: Both are faster than pthread. We've checked. Thoroughly. During those long Nordic nights when there's proper time for benchmarking.
Items processed per second across different implementations
Average message latency with error bars
Configuration: Producer-consumer queue test (10K items, 4 threads, queue size 10)
| Implementation | Throughput (items/sec) | Speedup vs pthread |
|---|---|---|
| fastcond_weak | 781,700.94 | +45.7% |
| fastcond_strong | 776,137.38 | +44.6% |
| pthread | 536,624.95 | baseline |
| Implementation | Avg Latency (μs) | Min (μs) | Max (μs) | Std Dev (μs) |
|---|---|---|---|---|
| fastcond_strong | 6.66 | 0.06 | 62.97 | 6.67 |
| fastcond_weak | 7.03 | 0.08 | 72.63 | 6.72 |
| pthread | 9.32 | 0.07 | 95.69 | 11.25 |
Configuration: Strong semantics test (10K items, queue size 5)
| Implementation | Throughput (items/sec) | Speedup vs pthread |
|---|---|---|
| fastcond_strong | 1,428,502.86 | +20.4% |
| pthread | 1,186,829.66 | baseline |
| Implementation | Avg Latency (μs) | Min (μs) | Max (μs) | Std Dev (μs) |
|---|---|---|---|---|
| fastcond_strong | 2.03 | 0.08 | 1017.29 | 37.92 |
| pthread | 2.40 | 0.08 | 1028.25 | 37.04 |