Joseph Doherty
|
1d4b87e5f9
|
docs: refresh benchmark comparison with increased message counts
Increase message counts across all 14 benchmark test files to reduce
run-to-run variance (e.g. PubSub 16B: 10K→50K, FanOut: 10K→15K,
SinglePub: 100K→500K, JS tests: 5K→25K). Rewrite benchmarks_comparison.md
with fresh numbers from two-batch runs. Key changes: multi 4x4 reached
parity (1.01x), fan-out improved to 0.84x, TLS pub/sub shows 4.70x .NET
advantage, previous small-count anomalies corrected.
|
2026-03-13 17:52:03 -04:00 |
|
Joseph Doherty
|
11e01b9026
|
perf: optimize MQTT cross-protocol path (0.30x → 0.78x Go)
Replace per-message async fire-and-forget with direct-buffer write loop
mirroring NatsClient pattern: SpinLock-guarded buffer append, double-
buffer swap, single WriteAsync per batch.
- MqttConnection: add _directBuf/_writeBuf + RunMqttWriteLoopAsync
- MqttConnection: add EnqueuePublishNoFlush (zero-alloc PUBLISH format)
- MqttPacketWriter: add WritePublishTo(Span<byte>) + MeasurePublish
- MqttTopicMapper: add NatsToMqttBytes with bounded ConcurrentDictionary
- MqttNatsClientAdapter: synchronous SendMessageNoFlush + SignalFlush
- Skip FlushAsync on plain TCP sockets (TCP auto-flushes)
|
2026-03-13 14:25:13 -04:00 |
|