Skip to content

perf: run handlers queued behind the reactor before polling again - #416

Draft
sgerbino wants to merge 1 commit into
developfrom
pr/recycling-free-list
Draft

sgerbino wants to merge 1 commit into
developfrom
pr/recycling-free-list

Conversation

@sgerbino

Copy link
Copy Markdown
Collaborator

When the reactor sentinel reached the front of the queue with handlers still behind it, the scheduler ran a non-blocking reactor poll. Those handlers were mostly posted by the pass that just ran, such as a coroutine an I/O completion woke, so the poll usually found nothing. On fan_out/fork_join/1 that was one empty epoll_wait(0) per lap.

The sentinel now moves behind the queued handlers once, then polls on its next turn regardless, so handlers that keep reposting still cannot starve I/O. poll() and poll_one() are unchanged.

testPostedWakeRunsBeforeNextPoll in the reactor fault suite counts the backend's wait calls to check that a posted wake runs before the next poll.

Pairs with cppalliance/capy pr/recycling-free-list, which lets the frame allocator cache a wide fan-out. CI resolves that capy branch by name. The benchmark A/B builds both sides against it, so the A/B measures only the scheduler change.

Local A/B on one pinned P-core, 5 interleaved runs of 2 s each, both changes vs develop:

Row develop this branch Δ
fork_join/1 188.6K 194.4K +3.1%
fork_join/64 3092 3182 +2.9%
nested/16 3004 3176 +5.7%
concurrent_parents/4 12.31K 12.73K +3.4%
concurrent_parents/16 10.97K 11.44K +4.3%
concurrent_parents_lockless/16 11.23K 11.62K +3.5%

io_context multithreaded/8 measured −0.9% locally, within noise. The CI run should settle it.

@cppalliance-bot

cppalliance-bot commented Nov 12, 2026 •

Copy link
Copy Markdown

GCOVR code coverage report https://416.corosio.prtest3.cppalliance.org/gcovr/index.html
LCOV code coverage report https://416.corosio.prtest3.cppalliance.org/genhtml/index.html
Coverage Diff Report https://416.corosio.prtest3.cppalliance.org/diff-report/index.html

Build time: 2026-11-13 06:22:12 UTC

@sgerbino
sgerbino force-pushed the pr/recycling-free-list branch from 7c6bd6d to 1e59c16 Compare November 12, 2026 15:07
@cppalliance-bot

cppalliance-bot commented Nov 12, 2026 •

Copy link
Copy Markdown

An automated preview of the documentation is available at https://416.corosio.prtest3.cppalliance.org/index.html

If more commits are pushed to the pull request, the docs will rebuild at the same URL.

2026-11-12 19:25:00 UTC

@github-actions

github-actions Bot commented Nov 12, 2026 •

Copy link
Copy Markdown

Benchmark report

13bcf933adf8 (base) vs 97fbd80fe670 (head)

  • ⚠️ linux — 3 flagged (epoll, uring)
  • ✅ windows — clean (iocp)
  • ⚠️ macos — 17 flagged (kqueue)

⚠️ Flagged

Platform Backend Benchmark Metric Base Head Δ Noise
linux epoll local_socket_latency/concurrent_lockless/4 latency_mean_ns 27.62 µs 26.46 µs 🟢▲ +4.51% 1.30%
linux epoll local_socket_latency/pingpong/64 latency_mean_ns 7.08 µs 6.73 µs 🟢▲ +5.00% 0.86%
linux epoll socket_throughput/bidirectional/1048576 bytes_per_sec 3.24 GB/s 3.37 GB/s 🟢▲ +4.08% 0.88%
macos kqueue accept_churn/burst/10 ops_per_sec 4.58K ops/s 4.82K ops/s 🟢▲ +5.49% 1.79%
macos kqueue accept_churn/burst_lockless/10 ops_per_sec 4.65K ops/s 4.86K ops/s 🟢▲ +4.73% 0.88%
macos kqueue fan_out/fork_join/1 ops_per_sec 49.04K ops/s 50.28K ops/s 🟢▲ +2.60% 0.65%
macos kqueue fan_out/fork_join/4 ops_per_sec 23.46K ops/s 24.07K ops/s 🟢▲ +2.71% 0.56%
macos kqueue fan_out/fork_join_lockless/1 ops_per_sec 50.31K ops/s 51.36K ops/s 🟢▲ +2.25% 0.56%
macos kqueue fan_out/fork_join_lockless/4 ops_per_sec 23.59K ops/s 24.18K ops/s 🟢▲ +2.60% 0.80%
macos kqueue local_socket_latency/concurrent/1 latency_mean_ns 1.83 µs 1.72 µs 🟢▲ +6.18% 0.40%
macos kqueue local_socket_latency/concurrent_lockless/1 latency_mean_ns 1.82 µs 1.70 µs 🟢▲ +6.14% 0.47%
macos kqueue local_socket_latency/concurrent_lockless/4 latency_mean_ns 6.80 µs 6.62 µs 🟢▲ +2.61% 0.22%
macos kqueue local_socket_latency/pingpong/1 latency_mean_ns 1.83 µs 1.72 µs 🟢▲ +5.83% 0.34%
macos kqueue local_socket_latency/pingpong/1024 latency_mean_ns 1.92 µs 1.81 µs 🟢▲ +5.30% 0.46%
macos kqueue local_socket_latency/pingpong/64 latency_mean_ns 1.83 µs 1.71 µs 🟢▲ +6.30% 0.50%
macos kqueue local_socket_latency/pingpong_lockless/1 latency_mean_ns 1.81 µs 1.70 µs 🟢▲ +6.00% 0.33%
macos kqueue local_socket_latency/pingpong_lockless/1024 latency_mean_ns 1.91 µs 1.80 µs 🟢▲ +6.11% 0.57%
macos kqueue local_socket_latency/pingpong_lockless/64 latency_mean_ns 1.81 µs 1.70 µs 🟢▲ +6.01% 0.49%
macos kqueue local_socket_throughput/unidirectional/1024 bytes_per_sec 1.28 GB/s 1.32 GB/s 🟢▲ +3.47% 0.40%
macos kqueue local_socket_throughput/unidirectional_lockless/1024 bytes_per_sec 1.28 GB/s 1.33 GB/s 🟢▲ +3.34% 0.40%
linux — full results (226 benchmarks, 5 iterations, 1.5s each)

accept_churn

Benchmark Backend Metric Base Head Δ Noise
burst/10 epoll ops_per_sec 2.01K ops/s 2.01K ops/s 🟢▲ +0.47% 0.95%
burst/10 uring ops_per_sec 2.40K ops/s 2.40K ops/s 🔴▼ -0.09% 0.37%
burst/100 epoll ops_per_sec 204.7 ops/s 203.9 ops/s ⚪ 0.00% 0.92%
burst/100 uring ops_per_sec 247.5 ops/s 248.1 ops/s 🟢▲ +0.01% 0.57%
burst_lockless/10 epoll ops_per_sec 2.04K ops/s 2.03K ops/s 🟢▲ +0.17% 1.16%
burst_lockless/10 uring ops_per_sec 2.41K ops/s 2.42K ops/s 🔴▼ -0.11% 0.97%
burst_lockless/100 epoll ops_per_sec 207.5 ops/s 206.1 ops/s 🔴▼ -0.16% 1.13%
burst_lockless/100 uring ops_per_sec 248.9 ops/s 248.7 ops/s 🔴▼ -0.12% 0.60%
concurrent/1 epoll ops_per_sec 15.84K ops/s 15.98K ops/s 🟢▲ +0.96% 0.61%
concurrent/1 uring ops_per_sec 17.18K ops/s 17.25K ops/s 🔴▼ -0.08% 1.35%
concurrent/16 epoll ops_per_sec 15.93K ops/s 15.98K ops/s 🟢▲ +0.22% 0.58%
concurrent/16 uring ops_per_sec 18.78K ops/s 18.88K ops/s 🔴▼ -0.04% 1.23%
concurrent/4 epoll ops_per_sec 15.87K ops/s 15.99K ops/s 🟢▲ +0.80% 0.44%
concurrent/4 uring ops_per_sec 17.79K ops/s 17.85K ops/s 🔴▼ -0.08% 1.33%
sequential epoll ops_per_sec 15.87K ops/s 16.06K ops/s 🟢▲ +0.84% 1.03%
sequential uring ops_per_sec 17.21K ops/s 17.24K ops/s 🔴▼ -0.21% 1.43%
sequential_lockless epoll ops_per_sec 16.05K ops/s 16.25K ops/s 🟢▲ +0.98% 1.01%
sequential_lockless uring ops_per_sec 17.37K ops/s 17.42K ops/s 🔴▼ -0.05% 1.31%

fan_out

Benchmark Backend Metric Base Head Δ Noise
concurrent_parents/1 epoll ops_per_sec 3.86K ops/s 3.86K ops/s 🔴▼ -0.16% 0.98%
concurrent_parents/1 uring ops_per_sec 4.00K ops/s 3.99K ops/s 🔴▼ -0.17% 0.17%
concurrent_parents/16 epoll ops_per_sec 3.80K ops/s 3.80K ops/s 🟢▲ +0.46% 1.18%
concurrent_parents/16 uring ops_per_sec 3.93K ops/s 3.94K ops/s 🔴▼ -0.06% 0.88%
concurrent_parents/4 epoll ops_per_sec 3.80K ops/s 3.80K ops/s 🟢▲ +0.22% 0.95%
concurrent_parents/4 uring ops_per_sec 3.96K ops/s 3.96K ops/s 🟢▲ +0.07% 0.23%
concurrent_parents_lockless/1 epoll ops_per_sec 3.93K ops/s 3.93K ops/s 🔴▼ -0.15% 1.23%
concurrent_parents_lockless/1 uring ops_per_sec 4.04K ops/s 4.06K ops/s 🔴▼ -0.02% 1.04%
concurrent_parents_lockless/16 epoll ops_per_sec 3.84K ops/s 3.83K ops/s 🔴▼ -0.17% 1.36%
concurrent_parents_lockless/16 uring ops_per_sec 3.96K ops/s 3.98K ops/s 🟢▲ +0.04% 0.92%
concurrent_parents_lockless/4 epoll ops_per_sec 3.85K ops/s 3.85K ops/s 🔴▼ -0.05% 1.29%
concurrent_parents_lockless/4 uring ops_per_sec 4.00K ops/s 4.01K ops/s 🔴▼ -0.13% 1.04%
fork_join/1 epoll ops_per_sec 56.86K ops/s 57.69K ops/s 🟢▲ +2.09% 1.33%
fork_join/1 uring ops_per_sec 54.34K ops/s 54.57K ops/s 🟢▲ +0.19% 0.51%
fork_join/16 epoll ops_per_sec 3.87K ops/s 3.85K ops/s 🔴▼ -0.48% 1.03%
fork_join/16 uring ops_per_sec 4.01K ops/s 4.01K ops/s 🔴▼ -0.04% 0.95%
fork_join/4 epoll ops_per_sec 15.29K ops/s 15.31K ops/s 🟢▲ +0.33% 1.26%
fork_join/4 uring ops_per_sec 15.05K ops/s 15.02K ops/s 🔴▼ -0.14% 0.95%
fork_join/64 epoll ops_per_sec 949.0 ops/s 946.4 ops/s 🔴▼ -0.38% 1.10%
fork_join/64 uring ops_per_sec 997.3 ops/s 995.6 ops/s 🔴▼ -0.14% 0.95%
fork_join_lockless/1 epoll ops_per_sec 57.29K ops/s 58.55K ops/s 🟢▲ +2.15% 0.86%
fork_join_lockless/1 uring ops_per_sec 55.87K ops/s 56.22K ops/s 🟢▲ +0.29% 1.13%
fork_join_lockless/16 epoll ops_per_sec 3.88K ops/s 3.89K ops/s 🔴▼ -0.01% 0.74%
fork_join_lockless/16 uring ops_per_sec 4.05K ops/s 4.06K ops/s 🟢▲ +0.18% 0.94%
fork_join_lockless/4 epoll ops_per_sec 15.33K ops/s 15.48K ops/s 🟢▲ +0.68% 0.95%
fork_join_lockless/4 uring ops_per_sec 15.25K ops/s 15.27K ops/s 🔴▼ -0.22% 0.97%
fork_join_lockless/64 epoll ops_per_sec 953.7 ops/s 951.6 ops/s 🔴▼ -0.10% 0.24%
fork_join_lockless/64 uring ops_per_sec 1.00K ops/s 1.00K ops/s 🔴▼ -0.08% 0.14%
nested/16 epoll ops_per_sec 940.9 ops/s 938.9 ops/s 🔴▼ -0.20% 0.32%
nested/16 uring ops_per_sec 988.9 ops/s 986.2 ops/s 🔴▼ -0.19% 0.22%
nested/4 epoll ops_per_sec 3.81K ops/s 3.81K ops/s 🔴▼ -0.12% 0.27%
nested/4 uring ops_per_sec 3.98K ops/s 3.97K ops/s 🔴▼ -0.31% 0.21%
nested_lockless/16 epoll ops_per_sec 954.0 ops/s 953.1 ops/s 🔴▼ -0.04% 1.10%
nested_lockless/16 uring ops_per_sec 997.8 ops/s 998.8 ops/s 🟢▲ +0.15% 0.19%
nested_lockless/4 epoll ops_per_sec 3.87K ops/s 3.86K ops/s 🔴▼ -0.23% 0.50%
nested_lockless/4 uring ops_per_sec 4.03K ops/s 4.03K ops/s 🔴▼ -0.16% 0.27%

http_server

Benchmark Backend Metric Base Head Δ Noise
concurrent/1 epoll ops_per_sec 56.88K ops/s 57.26K ops/s 🟢▲ +0.63% 1.02%
concurrent/1 uring ops_per_sec 53.36K ops/s 53.07K ops/s 🔴▼ -0.54% 1.04%
concurrent/16 epoll ops_per_sec 59.85K ops/s 60.16K ops/s 🟢▲ +0.16% 0.93%
concurrent/16 uring ops_per_sec 63.49K ops/s 63.59K ops/s 🟢▲ +0.03% 0.97%
concurrent/32 epoll ops_per_sec 59.44K ops/s 59.74K ops/s 🟢▲ +0.30% 0.98%
concurrent/32 uring ops_per_sec 63.44K ops/s 63.59K ops/s 🔴▼ -0.22% 1.05%
concurrent/4 epoll ops_per_sec 58.98K ops/s 59.59K ops/s 🟢▲ +0.94% 1.12%
concurrent/4 uring ops_per_sec 61.34K ops/s 61.05K ops/s 🔴▼ -0.48% 0.98%
multithread/1 epoll ops_per_sec 59.34K ops/s 59.51K ops/s 🟢▲ +0.07% 0.53%
multithread/1 uring ops_per_sec 63.51K ops/s 63.42K ops/s 🔴▼ -0.47% 1.03%
multithread/16 epoll ops_per_sec 163.7K ops/s 164.1K ops/s 🟢▲ +0.12% 0.28%
multithread/16 uring ops_per_sec 100.8K ops/s 101.0K ops/s 🟢▲ +0.25% 0.55%
multithread/2 epoll ops_per_sec 87.61K ops/s 87.92K ops/s 🟢▲ +0.46% 0.51%
multithread/2 uring ops_per_sec 75.22K ops/s 75.36K ops/s 🟢▲ +0.28% 0.56%
multithread/4 epoll ops_per_sec 161.5K ops/s 161.1K ops/s 🟢▲ +0.11% 0.81%
multithread/4 uring ops_per_sec 120.5K ops/s 120.5K ops/s 🔴▼ -0.19% 0.51%
multithread/8 epoll ops_per_sec 162.8K ops/s 162.4K ops/s 🔴▼ -0.37% 0.38%
multithread/8 uring ops_per_sec 111.8K ops/s 112.1K ops/s 🟢▲ +0.37% 0.37%
single_conn epoll ops_per_sec 57.01K ops/s 57.43K ops/s 🟢▲ +1.72% 1.26%
single_conn uring ops_per_sec 53.35K ops/s 52.78K ops/s 🔴▼ -0.75% 1.14%
single_conn_lockless epoll ops_per_sec 57.76K ops/s 58.33K ops/s 🟢▲ +2.36% 1.25%
single_conn_lockless uring ops_per_sec 55.17K ops/s 54.79K ops/s 🔴▼ -0.47% 0.91%

local_socket_latency

Benchmark Backend Metric Base Head Δ Noise
concurrent/1 epoll latency_mean_ns 7.23 µs 6.89 µs 🟢▲ +5.18% 4.52%
concurrent/1 uring latency_mean_ns 5.84 µs 5.81 µs 🟢▲ +0.38% 5.46%
concurrent/16 epoll latency_mean_ns 109.87 µs 107.80 µs 🟢▲ +1.93% 0.65%
concurrent/16 uring latency_mean_ns 93.39 µs 93.37 µs 🟢▲ +0.28% 0.32%
concurrent/4 epoll latency_mean_ns 27.82 µs 27.13 µs 🟢▲ +2.63% 1.86%
concurrent/4 uring latency_mean_ns 23.08 µs 23.20 µs 🟢▲ +0.11% 1.74%
concurrent_lockless/1 epoll latency_mean_ns 7.27 µs 6.69 µs 🟢▲ +5.13% 5.03%
concurrent_lockless/1 uring latency_mean_ns 5.67 µs 5.77 µs 🟢▲ +0.33% 5.33%
concurrent_lockless/16 epoll latency_mean_ns 108.74 µs 106.43 µs 🟢▲ +1.90% 0.92%
concurrent_lockless/16 uring latency_mean_ns 92.81 µs 92.69 µs 🟢▲ +0.32% 0.41%
concurrent_lockless/4 epoll latency_mean_ns 27.62 µs 26.46 µs 🟢▲ +4.51% 1.30%
concurrent_lockless/4 uring latency_mean_ns 23.23 µs 23.20 µs 🟢▲ +0.20% 1.60%
pingpong/1 epoll latency_mean_ns 7.16 µs 6.83 µs 🟢▲ +4.81% 3.42%
pingpong/1 uring latency_mean_ns 5.89 µs 6.00 µs 🔴▼ -2.03% 5.83%
pingpong/1024 epoll latency_mean_ns 7.58 µs 7.19 µs 🟢▲ +4.89% 4.37%
pingpong/1024 uring latency_mean_ns 6.31 µs 6.55 µs 🔴▼ -2.21% 5.57%
pingpong/64 epoll latency_mean_ns 7.08 µs 6.73 µs 🟢▲ +5.00% 0.86%
pingpong/64 uring latency_mean_ns 5.78 µs 5.65 µs 🟢▲ +0.79% 4.50%
pingpong_lockless/1 epoll latency_mean_ns 7.19 µs 6.91 µs 🟢▲ +3.98% 5.64%
pingpong_lockless/1 uring latency_mean_ns 5.74 µs 5.84 µs 🔴▼ -1.88% 6.44%
pingpong_lockless/1024 epoll latency_mean_ns 7.59 µs 7.20 µs 🟢▲ +4.12% 5.06%
pingpong_lockless/1024 uring latency_mean_ns 6.44 µs 6.52 µs 🟢▲ +0.76% 4.73%
pingpong_lockless/64 epoll latency_mean_ns 6.96 µs 6.95 µs 🟢▲ +4.45% 5.26%
pingpong_lockless/64 uring latency_mean_ns 5.90 µs 5.89 µs 🟢▲ +0.52% 5.75%

local_socket_throughput

Benchmark Backend Metric Base Head Δ Noise
bidirectional/1024 epoll bytes_per_sec 306.7 MB/s 313.3 MB/s 🟢▲ +2.40% 1.09%
bidirectional/1024 uring bytes_per_sec 354.5 MB/s 352.0 MB/s 🔴▼ -1.40% 1.28%
bidirectional/1048576 epoll bytes_per_sec 3.25 GB/s 3.31 GB/s 🟢▲ +1.91% 1.99%
bidirectional/1048576 uring bytes_per_sec 3.53 GB/s 3.53 GB/s 🟢▲ +2.05% 2.12%
bidirectional/16384 epoll bytes_per_sec 3.08 GB/s 3.10 GB/s 🔴▼ -0.48% 2.63%
bidirectional/16384 uring bytes_per_sec 2.94 GB/s 3.05 GB/s 🟢▲ +3.08% 1.62%
bidirectional/262144 epoll bytes_per_sec 3.27 GB/s 3.23 GB/s 🔴▼ -1.45% 1.84%
bidirectional/262144 uring bytes_per_sec 3.54 GB/s 3.46 GB/s 🔴▼ -0.90% 3.88%
bidirectional/4096 epoll bytes_per_sec 1.01 GB/s 1.02 GB/s 🟢▲ +1.43% 1.57%
bidirectional/4096 uring bytes_per_sec 1.12 GB/s 1.13 GB/s 🟢▲ +1.95% 0.62%
bidirectional/65536 epoll bytes_per_sec 3.03 GB/s 2.97 GB/s 🔴▼ -1.27% 2.21%
bidirectional/65536 uring bytes_per_sec 3.45 GB/s 3.46 GB/s 🔴▼ -0.17% 2.45%
bidirectional_lockless/1024 epoll bytes_per_sec 310.9 MB/s 311.6 MB/s 🟢▲ +0.83% 1.69%
bidirectional_lockless/1024 uring bytes_per_sec 353.6 MB/s 353.4 MB/s 🔴▼ -0.49% 1.05%
bidirectional_lockless/1048576 epoll bytes_per_sec 3.25 GB/s 3.31 GB/s 🟢▲ +2.55% 2.16%
bidirectional_lockless/1048576 uring bytes_per_sec 3.53 GB/s 3.58 GB/s 🟢▲ +3.89% 2.71%
bidirectional_lockless/16384 epoll bytes_per_sec 3.14 GB/s 3.09 GB/s 🟢▲ +0.54% 4.73%
bidirectional_lockless/16384 uring bytes_per_sec 2.99 GB/s 2.99 GB/s 🔴▼ -0.18% 1.82%
bidirectional_lockless/262144 epoll bytes_per_sec 3.25 GB/s 3.27 GB/s 🟢▲ +1.38% 2.25%
bidirectional_lockless/262144 uring bytes_per_sec 3.55 GB/s 3.48 GB/s 🔴▼ -1.70% 3.31%
bidirectional_lockless/4096 epoll bytes_per_sec 1.01 GB/s 1.02 GB/s 🟢▲ +1.38% 1.03%
bidirectional_lockless/4096 uring bytes_per_sec 1.12 GB/s 1.13 GB/s 🟢▲ +0.91% 0.85%
bidirectional_lockless/65536 epoll bytes_per_sec 3.02 GB/s 2.99 GB/s 🔴▼ -0.95% 1.95%
bidirectional_lockless/65536 uring bytes_per_sec 3.47 GB/s 3.48 GB/s 🔴▼ -0.19% 2.45%
unidirectional/1024 epoll bytes_per_sec 297.5 MB/s 307.7 MB/s 🟢▲ +3.62% 1.33%
unidirectional/1024 uring bytes_per_sec 353.2 MB/s 351.5 MB/s 🔴▼ -0.66% 1.02%
unidirectional/1048576 epoll bytes_per_sec 3.32 GB/s 3.39 GB/s 🟢▲ +2.05% 1.78%
unidirectional/1048576 uring bytes_per_sec 3.49 GB/s 3.54 GB/s 🟢▲ +3.63% 2.18%
unidirectional/16384 epoll bytes_per_sec 3.08 GB/s 3.12 GB/s 🟢▲ +2.25% 1.39%
unidirectional/16384 uring bytes_per_sec 3.01 GB/s 3.02 GB/s 🟢▲ +1.27% 2.59%
unidirectional/262144 epoll bytes_per_sec 3.35 GB/s 3.31 GB/s 🔴▼ -2.66% 2.96%
unidirectional/262144 uring bytes_per_sec 3.56 GB/s 3.48 GB/s 🔴▼ -1.41% 3.81%
unidirectional/4096 epoll bytes_per_sec 986.0 MB/s 1.01 GB/s 🟢▲ +1.65% 1.16%
unidirectional/4096 uring bytes_per_sec 1.13 GB/s 1.13 GB/s 🟢▲ +0.40% 0.58%
unidirectional/65536 epoll bytes_per_sec 3.28 GB/s 3.24 GB/s 🔴▼ -1.32% 2.03%
unidirectional/65536 uring bytes_per_sec 3.46 GB/s 3.54 GB/s 🟢▲ +2.26% 1.49%
unidirectional_lockless/1024 epoll bytes_per_sec 304.0 MB/s 309.7 MB/s 🟢▲ +2.30% 1.07%
unidirectional_lockless/1024 uring bytes_per_sec 355.3 MB/s 354.3 MB/s 🔴▼ -0.74% 0.72%
unidirectional_lockless/1048576 epoll bytes_per_sec 3.35 GB/s 3.43 GB/s 🟢▲ +3.42% 1.65%
unidirectional_lockless/1048576 uring bytes_per_sec 3.51 GB/s 3.57 GB/s 🟢▲ +4.34% 2.86%
unidirectional_lockless/16384 epoll bytes_per_sec 3.09 GB/s 3.10 GB/s 🔴▼ -0.76% 4.49%
unidirectional_lockless/16384 uring bytes_per_sec 3.03 GB/s 3.03 GB/s 🟢▲ +1.38% 3.30%
unidirectional_lockless/262144 epoll bytes_per_sec 3.38 GB/s 3.36 GB/s 🔴▼ -3.78% 2.96%
unidirectional_lockless/262144 uring bytes_per_sec 3.57 GB/s 3.50 GB/s 🔴▼ -1.00% 4.99%
unidirectional_lockless/4096 epoll bytes_per_sec 997.7 MB/s 1.02 GB/s 🟢▲ +1.73% 0.73%
unidirectional_lockless/4096 uring bytes_per_sec 1.12 GB/s 1.13 GB/s 🟢▲ +0.99% 1.38%
unidirectional_lockless/65536 epoll bytes_per_sec 3.27 GB/s 3.27 GB/s 🔴▼ -0.38% 1.45%
unidirectional_lockless/65536 uring bytes_per_sec 3.47 GB/s 3.52 GB/s 🟢▲ +0.04% 2.75%

socket_latency

Benchmark Backend Metric Base Head Δ Noise
concurrent/1 epoll latency_mean_ns 14.73 µs 14.43 µs 🟢▲ +2.37% 0.96%
concurrent/1 uring latency_mean_ns 13.56 µs 13.59 µs 🟢▲ +0.10% 1.31%
concurrent/16 epoll latency_mean_ns 232.32 µs 229.46 µs 🟢▲ +1.22% 1.09%
concurrent/16 uring latency_mean_ns 218.99 µs 220.67 µs 🔴▼ -0.07% 1.42%
concurrent/4 epoll latency_mean_ns 58.01 µs 57.33 µs 🟢▲ +1.41% 0.90%
concurrent/4 uring latency_mean_ns 54.44 µs 54.78 µs 🟢▲ +0.36% 1.48%
concurrent_lockless/1 epoll latency_mean_ns 14.64 µs 14.24 µs 🟢▲ +2.77% 1.21%
concurrent_lockless/1 uring latency_mean_ns 13.53 µs 13.61 µs 🟢▲ +0.28% 1.35%
concurrent_lockless/16 epoll latency_mean_ns 231.30 µs 227.57 µs 🟢▲ +1.23% 1.26%
concurrent_lockless/16 uring latency_mean_ns 219.36 µs 220.37 µs 🔴▼ -0.09% 1.16%
concurrent_lockless/4 epoll latency_mean_ns 57.69 µs 56.62 µs 🟢▲ +1.52% 1.38%
concurrent_lockless/4 uring latency_mean_ns 54.27 µs 54.68 µs 🔴▼ -0.08% 1.43%
pingpong/1 epoll latency_mean_ns 14.78 µs 14.31 µs 🟢▲ +2.73% 1.19%
pingpong/1 uring latency_mean_ns 13.55 µs 13.53 µs 🟢▲ +0.27% 1.25%
pingpong/1024 epoll latency_mean_ns 15.05 µs 14.67 µs 🟢▲ +2.83% 1.13%
pingpong/1024 uring latency_mean_ns 13.90 µs 13.87 µs 🟢▲ +0.20% 1.18%
pingpong/64 epoll latency_mean_ns 14.79 µs 14.38 µs 🟢▲ +2.83% 1.16%
pingpong/64 uring latency_mean_ns 13.62 µs 13.60 µs 🟢▲ +0.22% 1.21%
pingpong_lockless/1 epoll latency_mean_ns 14.55 µs 14.29 µs 🟢▲ +2.22% 1.02%
pingpong_lockless/1 uring latency_mean_ns 13.55 µs 13.48 µs 🟢▲ +0.64% 0.99%
pingpong_lockless/1024 epoll latency_mean_ns 14.92 µs 14.65 µs 🟢▲ +2.35% 1.18%
pingpong_lockless/1024 uring latency_mean_ns 13.77 µs 13.84 µs 🔴▼ -0.28% 1.48%
pingpong_lockless/64 epoll latency_mean_ns 14.63 µs 14.40 µs 🟢▲ +2.11% 1.13%
pingpong_lockless/64 uring latency_mean_ns 13.58 µs 13.54 µs 🟢▲ +0.46% 1.25%

socket_throughput

Benchmark Backend Metric Base Head Δ Noise
bidirectional/1024 epoll bytes_per_sec 133.9 MB/s 135.0 MB/s 🟢▲ +1.08% 1.15%
bidirectional/1024 uring bytes_per_sec 150.4 MB/s 150.5 MB/s 🟢▲ +0.01% 0.81%
bidirectional/1048576 epoll bytes_per_sec 3.24 GB/s 3.37 GB/s 🟢▲ +4.08% 0.88%
bidirectional/1048576 uring bytes_per_sec 3.32 GB/s 3.33 GB/s 🔴▼ -0.28% 1.61%
bidirectional/16384 epoll bytes_per_sec 1.62 GB/s 1.64 GB/s 🟢▲ +1.60% 1.77%
bidirectional/16384 uring bytes_per_sec 1.61 GB/s 1.61 GB/s 🔴▼ -0.23% 0.84%
bidirectional/262144 epoll bytes_per_sec 3.25 GB/s 3.25 GB/s 🟢▲ +0.30% 0.76%
bidirectional/262144 uring bytes_per_sec 3.27 GB/s 3.28 GB/s 🟢▲ +0.99% 1.24%
bidirectional/4096 epoll bytes_per_sec 505.6 MB/s 509.2 MB/s 🟢▲ +1.63% 1.31%
bidirectional/4096 uring bytes_per_sec 522.8 MB/s 523.0 MB/s ⚪ 0.00% 0.89%
bidirectional/65536 epoll bytes_per_sec 2.62 GB/s 2.62 GB/s 🟢▲ +0.66% 2.72%
bidirectional/65536 uring bytes_per_sec 2.86 GB/s 2.89 GB/s 🟢▲ +1.78% 2.04%
bidirectional_lockless/1024 epoll bytes_per_sec 135.3 MB/s 134.9 MB/s 🟢▲ +0.67% 1.24%
bidirectional_lockless/1024 uring bytes_per_sec 150.4 MB/s 150.2 MB/s 🔴▼ -0.14% 0.38%
bidirectional_lockless/1048576 epoll bytes_per_sec 3.28 GB/s 3.38 GB/s 🟢▲ +3.70% 1.67%
bidirectional_lockless/1048576 uring bytes_per_sec 3.30 GB/s 3.30 GB/s 🔴▼ -0.39% 1.16%
bidirectional_lockless/16384 epoll bytes_per_sec 1.63 GB/s 1.63 GB/s 🔴▼ -0.18% 2.17%
bidirectional_lockless/16384 uring bytes_per_sec 1.61 GB/s 1.61 GB/s 🔴▼ -0.15% 0.81%
bidirectional_lockless/262144 epoll bytes_per_sec 3.23 GB/s 3.25 GB/s 🟢▲ +0.41% 1.14%
bidirectional_lockless/262144 uring bytes_per_sec 3.23 GB/s 3.22 GB/s 🔴▼ -0.43% 0.90%
bidirectional_lockless/4096 epoll bytes_per_sec 510.4 MB/s 510.0 MB/s 🟢▲ +0.54% 1.41%
bidirectional_lockless/4096 uring bytes_per_sec 522.6 MB/s 522.7 MB/s 🟢▲ +0.13% 0.46%
bidirectional_lockless/65536 epoll bytes_per_sec 2.60 GB/s 2.60 GB/s 🔴▼ -0.45% 3.09%
bidirectional_lockless/65536 uring bytes_per_sec 2.88 GB/s 2.87 GB/s 🟢▲ +0.02% 1.78%
multithread/2 epoll bytes_per_sec 3.69 GB/s 3.65 GB/s 🔴▼ -0.50% 1.17%
multithread/2 uring bytes_per_sec 4.22 GB/s 4.21 GB/s 🔴▼ -0.42% 0.89%
multithread/4 epoll bytes_per_sec 5.74 GB/s 6.00 GB/s 🟢▲ +4.71% 1.70%
multithread/4 uring bytes_per_sec 6.75 GB/s 6.70 GB/s 🔴▼ -1.15% 0.85%
multithread/8 epoll bytes_per_sec 6.33 GB/s 6.40 GB/s 🟢▲ +0.80% 0.62%
multithread/8 uring bytes_per_sec 6.86 GB/s 6.82 GB/s 🔴▼ -0.63% 0.50%
unidirectional/1024 epoll bytes_per_sec 126.4 MB/s 126.2 MB/s 🟢▲ +0.29% 1.34%
unidirectional/1024 uring bytes_per_sec 148.3 MB/s 148.4 MB/s 🟢▲ +0.14% 0.23%
unidirectional/1048576 epoll bytes_per_sec 3.37 GB/s 3.37 GB/s 🟢▲ +0.43% 0.80%
unidirectional/1048576 uring bytes_per_sec 3.39 GB/s 3.41 GB/s 🟢▲ +0.69% 0.81%
unidirectional/16384 epoll bytes_per_sec 1.54 GB/s 1.54 GB/s 🔴▼ -0.40% 1.99%
unidirectional/16384 uring bytes_per_sec 1.71 GB/s 1.67 GB/s 🔴▼ -1.47% 4.04%
unidirectional/262144 epoll bytes_per_sec 3.32 GB/s 3.34 GB/s 🟢▲ +0.04% 1.28%
unidirectional/262144 uring bytes_per_sec 3.22 GB/s 3.26 GB/s 🟢▲ +0.87% 1.83%
unidirectional/4096 epoll bytes_per_sec 472.5 MB/s 471.9 MB/s 🔴▼ -0.52% 1.41%
unidirectional/4096 uring bytes_per_sec 523.7 MB/s 523.2 MB/s ⚪ 0.00% 1.26%
unidirectional/65536 epoll bytes_per_sec 2.74 GB/s 2.67 GB/s 🔴▼ -3.37% 4.77%
unidirectional/65536 uring bytes_per_sec 2.80 GB/s 2.85 GB/s 🟢▲ +0.46% 3.39%
unidirectional_lockless/1024 epoll bytes_per_sec 126.0 MB/s 127.5 MB/s 🟢▲ +0.65% 1.23%
unidirectional_lockless/1024 uring bytes_per_sec 149.5 MB/s 148.6 MB/s 🔴▼ -0.29% 1.01%
unidirectional_lockless/1048576 epoll bytes_per_sec 3.31 GB/s 3.35 GB/s 🟢▲ +1.64% 1.40%
unidirectional_lockless/1048576 uring bytes_per_sec 3.40 GB/s 3.39 GB/s 🔴▼ -0.71% 1.06%
unidirectional_lockless/16384 epoll bytes_per_sec 1.55 GB/s 1.57 GB/s 🟢▲ +1.29% 1.07%
unidirectional_lockless/16384 uring bytes_per_sec 1.71 GB/s 1.72 GB/s 🟢▲ +0.06% 0.25%
unidirectional_lockless/262144 epoll bytes_per_sec 3.35 GB/s 3.36 GB/s 🟢▲ +0.89% 0.94%
unidirectional_lockless/262144 uring bytes_per_sec 3.29 GB/s 3.27 GB/s 🔴▼ -0.61% 0.87%
unidirectional_lockless/4096 epoll bytes_per_sec 471.1 MB/s 477.7 MB/s 🟢▲ +1.62% 1.16%
unidirectional_lockless/4096 uring bytes_per_sec 520.7 MB/s 519.1 MB/s 🔴▼ -0.21% 0.46%
unidirectional_lockless/65536 epoll bytes_per_sec 2.80 GB/s 2.75 GB/s 🔴▼ -0.04% 3.91%
unidirectional_lockless/65536 uring bytes_per_sec 2.77 GB/s 2.79 GB/s 🟢▲ +0.02% 2.24%
windows — full results (77 benchmarks, 5 iterations, 1.5s each)

accept_churn

Benchmark Backend Metric Base Head Δ Noise
burst/10 iocp ops_per_sec 1.55K ops/s 1.56K ops/s 🟢▲ +0.14% 0.68%
burst/100 iocp ops_per_sec 156.3 ops/s 156.9 ops/s 🟢▲ +0.50% 0.81%
burst_lockless/10 iocp ops_per_sec 1.56K ops/s 1.56K ops/s 🔴▼ -0.48% 0.32%
burst_lockless/100 iocp ops_per_sec 159.9 ops/s 158.7 ops/s 🔴▼ -1.08% 0.96%
concurrent/1 iocp ops_per_sec 12.57K ops/s 12.58K ops/s 🟢▲ +0.20% 0.46%
concurrent/16 iocp ops_per_sec 13.49K ops/s 13.43K ops/s 🔴▼ -0.21% 0.56%
concurrent/4 iocp ops_per_sec 13.23K ops/s 13.18K ops/s 🔴▼ -0.41% 0.27%
sequential iocp ops_per_sec 12.48K ops/s 12.53K ops/s 🟢▲ +0.29% 0.46%
sequential_lockless iocp ops_per_sec 12.54K ops/s 12.56K ops/s 🟢▲ +0.33% 0.52%

fan_out

Benchmark Backend Metric Base Head Δ Noise
concurrent_parents/1 iocp ops_per_sec 7.36K ops/s 7.35K ops/s 🔴▼ -0.01% 0.38%
concurrent_parents/16 iocp ops_per_sec 6.83K ops/s 6.83K ops/s 🔴▼ -0.20% 1.56%
concurrent_parents/4 iocp ops_per_sec 7.27K ops/s 7.27K ops/s 🟢▲ +0.17% 0.78%
concurrent_parents_lockless/1 iocp ops_per_sec 7.38K ops/s 7.36K ops/s 🔴▼ -0.21% 0.46%
concurrent_parents_lockless/16 iocp ops_per_sec 6.83K ops/s 6.86K ops/s 🟢▲ +0.34% 1.20%
concurrent_parents_lockless/4 iocp ops_per_sec 7.26K ops/s 7.30K ops/s 🟢▲ +0.47% 0.78%
fork_join/1 iocp ops_per_sec 112.3K ops/s 112.1K ops/s 🔴▼ -0.24% 0.49%
fork_join/16 iocp ops_per_sec 7.36K ops/s 7.35K ops/s 🔴▼ -0.16% 0.44%
fork_join/4 iocp ops_per_sec 29.23K ops/s 29.05K ops/s 🔴▼ -0.66% 0.56%
fork_join/64 iocp ops_per_sec 1.81K ops/s 1.83K ops/s 🟢▲ +0.35% 1.93%
fork_join_lockless/1 iocp ops_per_sec 112.4K ops/s 111.7K ops/s 🔴▼ -0.51% 0.34%
fork_join_lockless/16 iocp ops_per_sec 7.37K ops/s 7.35K ops/s 🔴▼ -0.37% 0.25%
fork_join_lockless/4 iocp ops_per_sec 29.22K ops/s 29.10K ops/s 🔴▼ -0.38% 0.41%
fork_join_lockless/64 iocp ops_per_sec 1.82K ops/s 1.82K ops/s 🔴▼ -0.13% 0.91%
nested/16 iocp ops_per_sec 1.79K ops/s 1.80K ops/s 🟢▲ +0.25% 0.85%
nested/4 iocp ops_per_sec 7.23K ops/s 7.22K ops/s 🔴▼ -0.07% 0.27%
nested_lockless/16 iocp ops_per_sec 1.79K ops/s 1.80K ops/s 🔴▼ -0.12% 0.91%
nested_lockless/4 iocp ops_per_sec 7.22K ops/s 7.22K ops/s 🔴▼ -0.04% 0.24%

http_server

Benchmark Backend Metric Base Head Δ Noise
concurrent/1 iocp ops_per_sec 111.2K ops/s 111.5K ops/s 🟢▲ +0.13% 0.41%
concurrent/16 iocp ops_per_sec 108.5K ops/s 108.4K ops/s 🔴▼ -0.08% 0.18%
concurrent/32 iocp ops_per_sec 107.7K ops/s 107.8K ops/s 🔴▼ -0.13% 0.30%
concurrent/4 iocp ops_per_sec 109.3K ops/s 109.3K ops/s 🔴▼ -0.05% 0.35%
multithread/1 iocp ops_per_sec 107.5K ops/s 107.6K ops/s 🟢▲ +0.39% 0.68%
multithread/16 iocp ops_per_sec 264.0K ops/s 264.0K ops/s 🔴▼ -0.20% 0.42%
multithread/2 iocp ops_per_sec 181.0K ops/s 178.3K ops/s 🔴▼ -1.25% 2.76%
multithread/4 iocp ops_per_sec 223.0K ops/s 224.1K ops/s 🟢▲ +0.54% 0.60%
multithread/8 iocp ops_per_sec 264.4K ops/s 264.1K ops/s 🔴▼ -0.20% 0.47%
single_conn iocp ops_per_sec 111.6K ops/s 111.3K ops/s 🔴▼ -0.26% 0.34%
single_conn_lockless iocp ops_per_sec 111.4K ops/s 111.6K ops/s 🟢▲ +0.20% 0.39%

socket_latency

Benchmark Backend Metric Base Head Δ Noise
concurrent/1 iocp latency_mean_ns 7.65 µs 7.62 µs 🟢▲ +0.20% 0.31%
concurrent/16 iocp latency_mean_ns 129.60 µs 128.25 µs 🔴▼ -2.68% 6.05%
concurrent/4 iocp latency_mean_ns 30.69 µs 30.53 µs 🟢▲ +0.71% 0.37%
concurrent_lockless/1 iocp latency_mean_ns 7.63 µs 7.61 µs 🟢▲ +0.09% 0.28%
concurrent_lockless/16 iocp latency_mean_ns 122.38 µs 122.53 µs 🔴▼ -0.09% 0.34%
concurrent_lockless/4 iocp latency_mean_ns 30.64 µs 30.45 µs 🟢▲ +0.68% 0.28%
pingpong/1 iocp latency_mean_ns 7.63 µs 7.61 µs 🟢▲ +0.52% 0.56%
pingpong/1024 iocp latency_mean_ns 7.73 µs 7.72 µs 🔴▼ -0.11% 0.34%
pingpong/64 iocp latency_mean_ns 7.65 µs 7.63 µs 🟢▲ +0.42% 0.35%
pingpong_lockless/1 iocp latency_mean_ns 7.62 µs 7.62 µs 🔴▼ -0.18% 0.36%
pingpong_lockless/1024 iocp latency_mean_ns 7.71 µs 7.72 µs 🔴▼ -0.19% 0.46%
pingpong_lockless/64 iocp latency_mean_ns 7.63 µs 7.63 µs 🟢▲ +0.45% 0.38%

socket_throughput

Benchmark Backend Metric Base Head Δ Noise
bidirectional/1024 iocp bytes_per_sec 279.6 MB/s 280.4 MB/s 🟢▲ +0.31% 0.69%
bidirectional/1048576 iocp bytes_per_sec 5.10 GB/s 4.89 GB/s 🔴▼ -3.29% 10.99%
bidirectional/16384 iocp bytes_per_sec 3.73 GB/s 3.73 GB/s 🟢▲ +0.23% 0.67%
bidirectional/262144 iocp bytes_per_sec 9.49 GB/s 9.35 GB/s 🔴▼ -1.54% 1.60%
bidirectional/4096 iocp bytes_per_sec 1.07 GB/s 1.07 GB/s 🔴▼ -0.22% 0.70%
bidirectional/65536 iocp bytes_per_sec 9.04 GB/s 9.07 GB/s ⚪ 0.00% 1.03%
bidirectional_lockless/1024 iocp bytes_per_sec 279.9 MB/s 280.4 MB/s 🟢▲ +0.36% 0.56%
bidirectional_lockless/1048576 iocp bytes_per_sec 5.42 GB/s 5.18 GB/s 🔴▼ -4.11% 8.35%
bidirectional_lockless/16384 iocp bytes_per_sec 3.66 GB/s 3.71 GB/s 🔴▼ -0.40% 4.04%
bidirectional_lockless/262144 iocp bytes_per_sec 9.56 GB/s 9.50 GB/s 🔴▼ -0.52% 1.91%
bidirectional_lockless/4096 iocp bytes_per_sec 1.08 GB/s 1.08 GB/s ⚪ 0.00% 0.78%
bidirectional_lockless/65536 iocp bytes_per_sec 8.71 GB/s 8.96 GB/s 🔴▼ -0.48% 8.88%
multithread/2 iocp bytes_per_sec 12.38 GB/s 12.62 GB/s 🟢▲ +0.92% 3.55%
multithread/4 iocp bytes_per_sec 16.55 GB/s 16.74 GB/s 🔴▼ -0.79% 4.02%
multithread/8 iocp bytes_per_sec 18.43 GB/s 18.49 GB/s 🟢▲ +0.44% 0.79%
unidirectional/1024 iocp bytes_per_sec 284.4 MB/s 284.0 MB/s 🔴▼ -0.33% 0.76%
unidirectional/1048576 iocp bytes_per_sec 4.09 GB/s 3.99 GB/s 🔴▼ -4.72% 5.21%
unidirectional/16384 iocp bytes_per_sec 3.77 GB/s 3.80 GB/s 🟢▲ +0.63% 1.07%
unidirectional/262144 iocp bytes_per_sec 8.49 GB/s 8.51 GB/s 🟢▲ +1.35% 3.92%
unidirectional/4096 iocp bytes_per_sec 1.09 GB/s 1.09 GB/s 🟢▲ +0.20% 0.59%
unidirectional/65536 iocp bytes_per_sec 9.28 GB/s 9.17 GB/s 🔴▼ -0.55% 3.57%
unidirectional_lockless/1024 iocp bytes_per_sec 284.7 MB/s 284.3 MB/s 🔴▼ -0.25% 0.61%
unidirectional_lockless/1048576 iocp bytes_per_sec 4.62 GB/s 4.58 GB/s 🔴▼ -0.65% 6.21%
unidirectional_lockless/16384 iocp bytes_per_sec 3.77 GB/s 3.79 GB/s 🟢▲ +0.70% 1.37%
unidirectional_lockless/262144 iocp bytes_per_sec 9.66 GB/s 9.53 GB/s 🟢▲ +0.01% 2.23%
unidirectional_lockless/4096 iocp bytes_per_sec 1.09 GB/s 1.09 GB/s 🔴▼ -0.09% 0.76%
unidirectional_lockless/65536 iocp bytes_per_sec 9.24 GB/s 9.34 GB/s 🟢▲ +0.94% 2.16%
macos — full results (113 benchmarks, 5 iterations, 1.5s each)

accept_churn

Benchmark Backend Metric Base Head Δ Noise
burst/10 kqueue ops_per_sec 4.58K ops/s 4.82K ops/s 🟢▲ +5.49% 1.79%
burst/100 kqueue ops_per_sec 464.9 ops/s 468.7 ops/s 🟢▲ +1.17% 0.95%
burst_lockless/10 kqueue ops_per_sec 4.65K ops/s 4.86K ops/s 🟢▲ +4.73% 0.88%
burst_lockless/100 kqueue ops_per_sec 468.7 ops/s 473.2 ops/s 🟢▲ +0.45% 0.80%
concurrent/1 kqueue ops_per_sec 20.76K ops/s 20.64K ops/s 🔴▼ -0.57% 0.66%
concurrent/16 kqueue ops_per_sec 37.63K ops/s 37.73K ops/s 🟢▲ +0.37% 0.35%
concurrent/4 kqueue ops_per_sec 29.39K ops/s 29.45K ops/s 🟢▲ +0.10% 0.62%
sequential kqueue ops_per_sec 21.06K ops/s 20.92K ops/s 🟢▲ +0.09% 1.18%
sequential_lockless kqueue ops_per_sec 20.73K ops/s 20.85K ops/s 🔴▼ -0.05% 1.64%

fan_out

Benchmark Backend Metric Base Head Δ Noise
concurrent_parents/1 kqueue ops_per_sec 8.16K ops/s 8.18K ops/s 🟢▲ +0.15% 1.40%
concurrent_parents/16 kqueue ops_per_sec 7.15K ops/s 7.71K ops/s 🟢▲ +39.86% 20.33%
concurrent_parents/4 kqueue ops_per_sec 8.38K ops/s 8.32K ops/s 🔴▼ -3.50% 2.42%
concurrent_parents_lockless/1 kqueue ops_per_sec 8.25K ops/s 8.27K ops/s 🔴▼ -0.09% 1.68%
concurrent_parents_lockless/16 kqueue ops_per_sec 7.17K ops/s 7.75K ops/s 🟢▲ +41.42% 21.30%
concurrent_parents_lockless/4 kqueue ops_per_sec 8.47K ops/s 8.40K ops/s 🔴▼ -3.61% 2.40%
fork_join/1 kqueue ops_per_sec 49.04K ops/s 50.28K ops/s 🟢▲ +2.60% 0.65%
fork_join/16 kqueue ops_per_sec 8.16K ops/s 8.19K ops/s 🟢▲ +0.18% 1.21%
fork_join/4 kqueue ops_per_sec 23.46K ops/s 24.07K ops/s 🟢▲ +2.71% 0.56%
fork_join/64 kqueue ops_per_sec 2.14K ops/s 2.11K ops/s 🔴▼ -3.78% 2.39%
fork_join_lockless/1 kqueue ops_per_sec 50.31K ops/s 51.36K ops/s 🟢▲ +2.25% 0.56%
fork_join_lockless/16 kqueue ops_per_sec 8.28K ops/s 8.29K ops/s 🔴▼ -0.22% 1.30%
fork_join_lockless/4 kqueue ops_per_sec 23.59K ops/s 24.18K ops/s 🟢▲ +2.60% 0.80%
fork_join_lockless/64 kqueue ops_per_sec 2.14K ops/s 2.13K ops/s 🟢▲ +0.30% 2.40%
nested/16 kqueue ops_per_sec 2.11K ops/s 2.10K ops/s 🔴▼ -0.46% 2.65%
nested/4 kqueue ops_per_sec 8.07K ops/s 8.16K ops/s 🟢▲ +0.32% 0.88%
nested_lockless/16 kqueue ops_per_sec 2.11K ops/s 2.11K ops/s 🔴▼ -2.63% 2.77%
nested_lockless/4 kqueue ops_per_sec 8.14K ops/s 8.25K ops/s 🟢▲ +0.48% 1.51%

http_server

Benchmark Backend Metric Base Head Δ Noise
concurrent/1 kqueue ops_per_sec 49.64K ops/s 49.64K ops/s 🔴▼ -0.12% 0.25%
concurrent/16 kqueue ops_per_sec 130.6K ops/s 131.2K ops/s 🟢▲ +0.66% 0.54%
concurrent/32 kqueue ops_per_sec 129.9K ops/s 129.9K ops/s 🔴▼ -0.06% 1.21%
concurrent/4 kqueue ops_per_sec 90.08K ops/s 90.29K ops/s 🔴▼ -0.05% 1.01%
multithread/1 kqueue ops_per_sec 130.0K ops/s 129.9K ops/s 🔴▼ -0.01% 1.60%
multithread/16 kqueue ops_per_sec 132.5K ops/s 132.6K ops/s 🔴▼ -0.08% 0.69%
multithread/2 kqueue ops_per_sec 184.3K ops/s 185.1K ops/s 🟢▲ +0.79% 0.59%
multithread/4 kqueue ops_per_sec 162.4K ops/s 162.4K ops/s 🟢▲ +0.18% 0.93%
multithread/8 kqueue ops_per_sec 137.7K ops/s 137.7K ops/s 🔴▼ -0.07% 1.21%
single_conn kqueue ops_per_sec 49.69K ops/s 49.61K ops/s 🔴▼ -0.08% 0.22%
single_conn_lockless kqueue ops_per_sec 49.92K ops/s 49.89K ops/s 🔴▼ -0.02% 0.12%

local_socket_latency

Benchmark Backend Metric Base Head Δ Noise
concurrent/1 kqueue latency_mean_ns 1.83 µs 1.72 µs 🟢▲ +6.18% 0.40%
concurrent/16 kqueue latency_mean_ns 27.32 µs 26.87 µs 🟢▲ +1.53% 0.30%
concurrent/4 kqueue latency_mean_ns 6.87 µs 6.66 µs 🟢▲ +2.52% 0.90%
concurrent_lockless/1 kqueue latency_mean_ns 1.82 µs 1.70 µs 🟢▲ +6.14% 0.47%
concurrent_lockless/16 kqueue latency_mean_ns 27.10 µs 26.77 µs 🟢▲ +1.16% 0.61%
concurrent_lockless/4 kqueue latency_mean_ns 6.80 µs 6.62 µs 🟢▲ +2.61% 0.22%
pingpong/1 kqueue latency_mean_ns 1.83 µs 1.72 µs 🟢▲ +5.83% 0.34%
pingpong/1024 kqueue latency_mean_ns 1.92 µs 1.81 µs 🟢▲ +5.30% 0.46%
pingpong/64 kqueue latency_mean_ns 1.83 µs 1.71 µs 🟢▲ +6.30% 0.50%
pingpong_lockless/1 kqueue latency_mean_ns 1.81 µs 1.70 µs 🟢▲ +6.00% 0.33%
pingpong_lockless/1024 kqueue latency_mean_ns 1.91 µs 1.80 µs 🟢▲ +6.11% 0.57%
pingpong_lockless/64 kqueue latency_mean_ns 1.81 µs 1.70 µs 🟢▲ +6.01% 0.49%

local_socket_throughput

Benchmark Backend Metric Base Head Δ Noise
bidirectional/1024 kqueue bytes_per_sec 1.30 GB/s 1.32 GB/s 🟢▲ +1.58% 0.47%
bidirectional/1048576 kqueue bytes_per_sec 3.89 GB/s 3.89 GB/s 🟢▲ +0.26% 0.84%
bidirectional/16384 kqueue bytes_per_sec 3.91 GB/s 3.91 GB/s 🔴▼ -0.24% 0.91%
bidirectional/262144 kqueue bytes_per_sec 3.90 GB/s 3.89 GB/s 🔴▼ -0.14% 0.75%
bidirectional/4096 kqueue bytes_per_sec 2.94 GB/s 2.93 GB/s 🔴▼ -0.50% 0.51%
bidirectional/65536 kqueue bytes_per_sec 3.89 GB/s 3.91 GB/s 🟢▲ +0.56% 0.67%
bidirectional_lockless/1024 kqueue bytes_per_sec 1.31 GB/s 1.33 GB/s 🟢▲ +1.54% 0.67%
bidirectional_lockless/1048576 kqueue bytes_per_sec 3.97 GB/s 3.98 GB/s 🟢▲ +0.20% 0.92%
bidirectional_lockless/16384 kqueue bytes_per_sec 3.99 GB/s 3.98 GB/s 🔴▼ -0.75% 0.70%
bidirectional_lockless/262144 kqueue bytes_per_sec 3.95 GB/s 3.96 GB/s 🟢▲ +0.46% 1.01%
bidirectional_lockless/4096 kqueue bytes_per_sec 2.98 GB/s 2.99 GB/s 🟢▲ +0.34% 0.48%
bidirectional_lockless/65536 kqueue bytes_per_sec 3.96 GB/s 3.98 GB/s 🟢▲ +0.04% 0.89%
unidirectional/1024 kqueue bytes_per_sec 1.28 GB/s 1.32 GB/s 🟢▲ +3.47% 0.40%
unidirectional/1048576 kqueue bytes_per_sec 3.54 GB/s 3.53 GB/s 🔴▼ -0.03% 0.59%
unidirectional/16384 kqueue bytes_per_sec 3.55 GB/s 3.55 GB/s 🔴▼ -0.27% 0.51%
unidirectional/262144 kqueue bytes_per_sec 3.55 GB/s 3.54 GB/s 🔴▼ -0.11% 0.39%
unidirectional/4096 kqueue bytes_per_sec 2.73 GB/s 2.73 GB/s 🔴▼ -0.55% 0.43%
unidirectional/65536 kqueue bytes_per_sec 3.55 GB/s 3.53 GB/s 🔴▼ -0.33% 0.56%
unidirectional_lockless/1024 kqueue bytes_per_sec 1.28 GB/s 1.33 GB/s 🟢▲ +3.34% 0.40%
unidirectional_lockless/1048576 kqueue bytes_per_sec 3.60 GB/s 3.59 GB/s 🔴▼ -0.40% 0.22%
unidirectional_lockless/16384 kqueue bytes_per_sec 3.63 GB/s 3.61 GB/s 🔴▼ -0.74% 0.64%
unidirectional_lockless/262144 kqueue bytes_per_sec 3.61 GB/s 3.60 GB/s 🔴▼ -0.26% 0.31%
unidirectional_lockless/4096 kqueue bytes_per_sec 2.77 GB/s 2.76 GB/s 🔴▼ -0.17% 0.70%
unidirectional_lockless/65536 kqueue bytes_per_sec 3.62 GB/s 3.60 GB/s 🔴▼ -0.33% 0.60%

socket_latency

Benchmark Backend Metric Base Head Δ Noise
concurrent/1 kqueue latency_mean_ns 18.71 µs 18.68 µs 🟢▲ +0.11% 0.46%
concurrent/16 kqueue latency_mean_ns 120.56 µs 120.64 µs 🟢▲ +0.08% 1.01%
concurrent/4 kqueue latency_mean_ns 43.15 µs 43.23 µs 🟢▲ +0.18% 1.08%
concurrent_lockless/1 kqueue latency_mean_ns 18.65 µs 18.62 µs 🟢▲ +0.16% 0.28%
concurrent_lockless/16 kqueue latency_mean_ns 120.63 µs 120.83 µs 🔴▼ -0.27% 1.20%
concurrent_lockless/4 kqueue latency_mean_ns 42.58 µs 42.55 µs 🔴▼ -0.02% 0.37%
pingpong/1 kqueue latency_mean_ns 15.45 µs 15.31 µs 🟢▲ +0.31% 1.31%
pingpong/1024 kqueue latency_mean_ns 19.02 µs 19.08 µs 🔴▼ -0.22% 0.52%
pingpong/64 kqueue latency_mean_ns 18.70 µs 18.68 µs ⚪ 0.00% 0.21%
pingpong_lockless/1 kqueue latency_mean_ns 15.34 µs 15.30 µs 🟢▲ +0.30% 0.10%
pingpong_lockless/1024 kqueue latency_mean_ns 19.00 µs 19.10 µs 🔴▼ -0.45% 0.20%
pingpong_lockless/64 kqueue latency_mean_ns 18.65 µs 18.60 µs 🟢▲ +0.22% 0.18%

socket_throughput

Benchmark Backend Metric Base Head Δ Noise
bidirectional/1024 kqueue bytes_per_sec 472.4 MB/s 477.5 MB/s 🟢▲ +1.39% 1.01%
bidirectional/1048576 kqueue bytes_per_sec 7.88 GB/s 6.84 GB/s 🔴▼ -12.61% 6.90%
bidirectional/16384 kqueue bytes_per_sec 4.52 GB/s 4.55 GB/s 🟢▲ +0.40% 1.65%
bidirectional/262144 kqueue bytes_per_sec 8.75 GB/s 9.06 GB/s 🟢▲ +0.01% 7.16%
bidirectional/4096 kqueue bytes_per_sec 1.83 GB/s 1.82 GB/s 🔴▼ -0.04% 0.95%
bidirectional/65536 kqueue bytes_per_sec 7.62 GB/s 7.59 GB/s 🔴▼ -0.18% 1.52%
bidirectional_lockless/1024 kqueue bytes_per_sec 476.6 MB/s 478.0 MB/s 🟢▲ +0.76% 0.57%
bidirectional_lockless/1048576 kqueue bytes_per_sec 7.81 GB/s 6.90 GB/s 🔴▼ -12.31% 4.13%
bidirectional_lockless/16384 kqueue bytes_per_sec 4.56 GB/s 4.58 GB/s 🟢▲ +0.56% 1.89%
bidirectional_lockless/262144 kqueue bytes_per_sec 8.78 GB/s 8.99 GB/s 🟢▲ +1.13% 3.37%
bidirectional_lockless/4096 kqueue bytes_per_sec 1.83 GB/s 1.83 GB/s 🟢▲ +0.10% 0.70%
bidirectional_lockless/65536 kqueue bytes_per_sec 7.55 GB/s 7.60 GB/s 🟢▲ +0.67% 2.31%
multithread/2 kqueue bytes_per_sec 7.55 GB/s 7.37 GB/s 🔴▼ -0.35% 4.69%
multithread/4 kqueue bytes_per_sec 3.66 GB/s 3.59 GB/s 🟢▲ +0.02% 4.70%
multithread/8 kqueue bytes_per_sec 3.21 GB/s 3.35 GB/s 🟢▲ +4.31% 10.59%
unidirectional/1024 kqueue bytes_per_sec 461.0 MB/s 460.0 MB/s 🔴▼ -0.21% 2.04%
unidirectional/1048576 kqueue bytes_per_sec 9.00 GB/s 8.75 GB/s 🟢▲ +0.06% 5.37%
unidirectional/16384 kqueue bytes_per_sec 4.52 GB/s 4.47 GB/s 🔴▼ -1.23% 1.96%
unidirectional/262144 kqueue bytes_per_sec 8.75 GB/s 8.94 GB/s 🟢▲ +1.78% 1.59%
unidirectional/4096 kqueue bytes_per_sec 1.81 GB/s 1.80 GB/s ⚪ 0.00% 1.75%
unidirectional/65536 kqueue bytes_per_sec 7.30 GB/s 7.41 GB/s 🟢▲ +2.01% 2.53%
unidirectional_lockless/1024 kqueue bytes_per_sec 478.6 MB/s 480.8 MB/s 🟢▲ +1.03% 2.13%
unidirectional_lockless/1048576 kqueue bytes_per_sec 9.05 GB/s 8.81 GB/s 🔴▼ -0.80% 4.66%
unidirectional_lockless/16384 kqueue bytes_per_sec 4.45 GB/s 4.49 GB/s 🟢▲ +0.44% 3.25%
unidirectional_lockless/262144 kqueue bytes_per_sec 8.93 GB/s 9.11 GB/s 🟢▲ +2.27% 3.24%
unidirectional_lockless/4096 kqueue bytes_per_sec 1.83 GB/s 1.83 GB/s 🟢▲ +0.22% 0.99%
unidirectional_lockless/65536 kqueue bytes_per_sec 7.44 GB/s 7.44 GB/s 🟢▲ +0.24% 1.85%

Run & raw JSON artifacts · flag rule: |median Δ| > max(2%, 3×CV of the noisier side) · advisory only

@sgerbino
sgerbino force-pushed the pr/recycling-free-list branch from 1e59c16 to f4b0d0d Compare November 12, 2026 19:16
@coderabbitai

coderabbitai Bot commented Nov 12, 2026

Copy link
Copy Markdown

Important

Draft PR not reviewed

Draft PRs are not automatically reviewed by default.

  • Trigger a manual review

To automatically review draft PRs, update your CodeRabbit configuration:

reviews:
  auto_review:
    drafts: true
  • Autofix · Keep fixing CodeRabbit findings and required CI, and resolving merge conflicts

When the reactor sentinel reached the front with handlers still queued,
the scheduler polled the reactor without blocking. Those handlers were
mostly posted by the pass that just ran, such as a coroutine an I/O
completion woke, so the poll usually found nothing.

The sentinel now moves behind the queued handlers once and polls on
its next turn regardless, so handlers that keep reposting still cannot
starve I/O. restart() forgets a deferral still pending, so a run
stopped with the sentinel moved back doesn't cost the next run its own.
poll() and poll_one() are unchanged.
@sgerbino
sgerbino force-pushed the pr/recycling-free-list branch from f4b0d0d to 97fbd80 Compare November 13, 2026 05:57
@sgerbino sgerbino added the benchmark Triggers benchmark suite label Nov 16, 2026
@github-actions github-actions Bot removed the benchmark Triggers benchmark suite label Nov 16, 2026

This branch has not been deployed

No deployments
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

Status: Backlog

Development

Successfully merging this pull request may close these issues.

2 participants