Fibre SPMC benchmarks

one producer broadcasting to many consumers (every consumer receives every item)

Throughput in Melem/s: Million items per second per item sent, median of samples. The best result in each row is highlighted. In batched tables, the bracketed figure is the gain over the same implementation's single-item row. A dash means the implementation has no channel for that cell. Measured on MacBook M4 Pro (14 cores).

Only fibre appears here: no other library in the set has a comparable broadcast channel, and tokio::sync::broadcast drops items for lagging consumers rather than applying backpressure. Throughput is per item sent, so a 1×64 row does 64 times the receive work per unit shown.

sync

Capacity	P×C	fibre
1	1x1	0.246
1	1x16	0.032
1	1x64	0.010
128	1x1	14.9
128	1x16	0.016
128	1x64	0.005
1024	1x1	17.7
1024	1x16	0.029
1024	1x64	0.007

sync, batched (512 items per call)

Capacity	P×C	fibre
128	1x1	30.4 (2.0x)
128	1x16	3.77 (235.9x)
128	1x64	1.02 (203.8x)
1024	1x1	81.6 (4.6x)
1024	1x16	13.6 (469.6x)
1024	1x64	4.10 (585.1x)

Native batch: fibre send and receive.

async

Capacity	P×C	fibre
1	1x1	7.62
1	1x16	0.354
1	1x64	0.078
128	1x1	43.1
128	1x16	0.755
128	1x64	0.204
1024	1x1	20.4
1024	1x16	0.759
1024	1x64	0.212

async, batched (512 items per call)

Capacity	P×C	fibre
128	1x1	112.8 (2.6x)
128	1x16	24.0 (31.7x)
128	1x64	9.93 (48.7x)
1024	1x1	142.9 (7.0x)
1024	1x16	58.2 (76.7x)
1024	1x64	23.6 (111.1x)

Native batch: fibre send and receive.