[WTX] WebSocket evaluation by c410-f3r · Pull Request #953 · MDA2AV/HttpArena

c410-f3r · 2026-06-29T11:02:13Z

Do you guys mind if an execute a couple of perf runs?

Tokio, the canonical Rust's runtime, is not scalable when a single runtime is involved. See https://medium.com/@fujita.tomonori/scalable-server-design-in-rust-with-tokio-4c81a5f350a3.

Therefore, the following scenarios will be reproduced to evaluate the best outcome.

Runtime per thread created in a spawn, build, new_current_thread
Runtime per thread created in a spawn, build, new_multi_thread
Runtime per thread created in a spawn, build_local, new_current_thread
Runtime per thread created in a thread, build, new_current_thread
Runtime per thread created in a thread, build, new_multi_thread
Runtime per thread created in a thread, build_local, new_current_thread

It will probably take less than a hour to finish everything.

MDA2AV · 2026-06-29T11:10:37Z

No problem but do one benchmark command at a time(only fire the next after the previous one finished) because a new run prompt cancels any previous pending

c410-f3r · 2026-06-29T11:12:18Z

Thank you @MDA2AV !

c410-f3r · 2026-06-29T11:13:27Z

Runtime per thread created in a thread, build_local, new_current_thread

/benchmark -f wtx-ws

github-actions · 2026-06-29T11:13:36Z

👋 /benchmark request received. A collaborator will review and approve the run.

github-actions · 2026-06-29T11:31:29Z

Benchmark Results

Framework: wtx-ws | Test: all tests

Test	Conn	RPS	CPU	Mem	Δ RPS	Δ Mem
echo-ws	512	4,041,238	6375.1%	43MiB	~0%	+26.5%
echo-ws	4096	4,052,778	6179.1%	110MiB	-2.8%	+34.1%
echo-ws	16384	3,663,086	6476.0%	358MiB	-4.0%	+47.9%
echo-ws-pipeline	512	5,532,242	6526.3%	42MiB	-1.8%	+50.0%
echo-ws-pipeline	4096	5,944,497	6633.9%	112MiB	-1.2%	+45.5%
echo-ws-pipeline	16384	5,733,603	6622.5%	354MiB	-1.7%	+50.0%

Full log

  Thread Stats   Avg      p50      p90      p99    p99.9
    Latency   1.48ms   1.47ms   2.31ms   3.03ms   3.25ms

  27540461 frames sent in 5.00s, 27540545 frames received
  Throughput: 5.51M req/s
  Bandwidth:  36.76MB/s
  WS upgrades: 512
  WS frames:   27540545
  Latency samples: 27540466 / 27540545 responses (100.0%)
[info] CPU 6566.3% | Mem 39MiB

[run 3/3]
gcannon v0.5.3 [WS]
  Target:    localhost:8080/ws
  Threads:   64
  Conns:     512 (8/thread)
  Pipeline:  16
  Req/conn:  unlimited (keep-alive)
  Expected:  200
  Duration:  5s


  Thread Stats   Avg      p50      p90      p99    p99.9
    Latency   1.48ms   1.44ms   2.24ms   2.98ms   3.26ms

  27661159 frames sent in 5.00s, 27661214 frames received
  Throughput: 5.53M req/s
  Bandwidth:  36.92MB/s
  WS upgrades: 512
  WS frames:   27661214
  Latency samples: 27661122 / 27661214 responses (100.0%)
[info] CPU 6526.3% | Mem 42MiB

=== Best: 5532242 req/s (CPU: 6526.3%, Mem: 42MiB) ===
[info] saved results/echo-ws-pipeline/512/wtx-ws.json
httparena-bench-wtx-ws
httparena-bench-wtx-ws

==============================================
=== wtx-ws / echo-ws-pipeline / 4096c (tool=gcannon) ===
==============================================
[info] ws-only framework — skipping HTTP probe (sleep 2s for startup)

[run 1/3]
gcannon v0.5.3 [WS]
  Target:    localhost:8080/ws
  Threads:   64
  Conns:     4096 (64/thread)
  Pipeline:  16
  Req/conn:  unlimited (keep-alive)
  Expected:  200
  Duration:  5s


  Thread Stats   Avg      p50      p90      p99    p99.9
    Latency   11.24ms   10.90ms   12.80ms   22.80ms   29.60ms

  28856076 frames sent in 5.00s, 28793694 frames received
  Throughput: 5.76M req/s
  Bandwidth:  38.53MB/s
  WS upgrades: 4096
  WS frames:   28793694
  Latency samples: 28793694 / 28793694 responses (100.0%)
[info] CPU 6061.7% | Mem 106MiB

[run 2/3]
gcannon v0.5.3 [WS]
  Target:    localhost:8080/ws
  Threads:   64
  Conns:     4096 (64/thread)
  Pipeline:  16
  Req/conn:  unlimited (keep-alive)
  Expected:  200
  Duration:  5s


  Thread Stats   Avg      p50      p90      p99    p99.9
    Latency   11.02ms   11.00ms   12.50ms   13.40ms   13.70ms

  29725515 frames sent in 5.00s, 29722487 frames received
  Throughput: 5.94M req/s
  Bandwidth:  39.68MB/s
  WS upgrades: 4096
  WS frames:   29722487
  Latency samples: 29722399 / 29722487 responses (100.0%)
[info] CPU 6633.9% | Mem 112MiB

[run 3/3]
gcannon v0.5.3 [WS]
  Target:    localhost:8080/ws
  Threads:   64
  Conns:     4096 (64/thread)
  Pipeline:  16
  Req/conn:  unlimited (keep-alive)
  Expected:  200
  Duration:  5s


  Thread Stats   Avg      p50      p90      p99    p99.9
    Latency   11.26ms   10.90ms   12.60ms   22.70ms   27.20ms

  29090416 frames sent in 5.00s, 29084648 frames received
  Throughput: 5.82M req/s
  Bandwidth:  38.83MB/s
  WS upgrades: 4096
  WS frames:   29084648
  Latency samples: 29084588 / 29084648 responses (100.0%)
[info] CPU 6195.3% | Mem 113MiB

=== Best: 5944497 req/s (CPU: 6633.9%, Mem: 112MiB) ===
[info] saved results/echo-ws-pipeline/4096/wtx-ws.json
httparena-bench-wtx-ws
httparena-bench-wtx-ws

==============================================
=== wtx-ws / echo-ws-pipeline / 16384c (tool=gcannon) ===
==============================================
[info] ws-only framework — skipping HTTP probe (sleep 2s for startup)

[run 1/3]
gcannon v0.5.3 [WS]
  Target:    localhost:8080/ws
  Threads:   64
  Conns:     16384 (256/thread)
  Pipeline:  16
  Req/conn:  unlimited (keep-alive)
  Expected:  200
  Duration:  5s


  Thread Stats   Avg      p50      p90      p99    p99.9
    Latency   46.40ms   45.40ms   50.10ms   89.40ms   103.90ms

  27435792 frames sent in 5.00s, 27173730 frames received
  Throughput: 5.43M req/s
  Bandwidth:  36.67MB/s
  WS upgrades: 16384
  WS frames:   27173730
  Latency samples: 27173730 / 27173730 responses (100.0%)
[info] CPU 5842.9% | Mem 350MiB

[run 2/3]
gcannon v0.5.3 [WS]
  Target:    localhost:8080/ws
  Threads:   64
  Conns:     16384 (256/thread)
  Pipeline:  16
  Req/conn:  unlimited (keep-alive)
  Expected:  200
  Duration:  5s


  Thread Stats   Avg      p50      p90      p99    p99.9
    Latency   45.28ms   45.70ms   50.20ms   52.60ms   53.30ms

  28930077 frames sent in 5.00s, 28668018 frames received
  Throughput: 5.73M req/s
  Bandwidth:  38.67MB/s
  WS upgrades: 16384
  WS frames:   28668018
  Latency samples: 28668018 / 28668018 responses (100.0%)
[info] CPU 6622.5% | Mem 354MiB

[run 3/3]
gcannon v0.5.3 [WS]
  Target:    localhost:8080/ws
  Threads:   64
  Conns:     16384 (256/thread)
  Pipeline:  16
  Req/conn:  unlimited (keep-alive)
  Expected:  200
  Duration:  5s


  Thread Stats   Avg      p50      p90      p99    p99.9
    Latency   45.41ms   45.70ms   48.80ms   52.00ms   52.80ms

  28826656 frames sent in 5.00s, 28564594 frames received
  Throughput: 5.71M req/s
  Bandwidth:  38.53MB/s
  WS upgrades: 16384
  WS frames:   28564594
  Latency samples: 28564594 / 28564594 responses (100.0%)
[info] CPU 6363.0% | Mem 356MiB

=== Best: 5733603 req/s (CPU: 6622.5%, Mem: 354MiB) ===
[info] saved results/echo-ws-pipeline/16384/wtx-ws.json
httparena-bench-wtx-ws
httparena-bench-wtx-ws
[info] rebuilding site/data/*.json
[updated] /home/diogo/actions-runner/_work/HttpArena/HttpArena/site/data/frameworks.json
[updated] /home/diogo/actions-runner/_work/HttpArena/HttpArena/site/data/echo-ws-16384.json
[updated] /home/diogo/actions-runner/_work/HttpArena/HttpArena/site/data/echo-ws-4096.json
[updated] /home/diogo/actions-runner/_work/HttpArena/HttpArena/site/data/echo-ws-512.json
[updated] /home/diogo/actions-runner/_work/HttpArena/HttpArena/site/data/echo-ws-pipeline-16384.json
[updated] /home/diogo/actions-runner/_work/HttpArena/HttpArena/site/data/echo-ws-pipeline-4096.json
[updated] /home/diogo/actions-runner/_work/HttpArena/HttpArena/site/data/echo-ws-pipeline-512.json
[updated] /home/diogo/actions-runner/_work/HttpArena/HttpArena/site/data/current.json
[info] done
[info] restoring loopback MTU to 65536

c410-f3r · 2026-06-29T11:36:17Z

Runtime per thread created in a thread, build, new_multi_thread

/benchmark -f wtx-ws

github-actions · 2026-06-29T11:36:25Z

👋 /benchmark request received. A collaborator will review and approve the run.

github-actions · 2026-06-29T11:41:07Z

Benchmark Results

Framework: wtx-ws | Test: all tests

Test	Conn	RPS	CPU	Mem	Δ RPS	Δ Mem
echo-ws	512	3,853,307	6409.9%	239MiB	-4.7%	+602.9%
echo-ws	4096	4,063,085	6324.8%	306MiB	-2.6%	+273.2%
echo-ws	16384	3,735,800	6225.1%	609MiB	-2.1%	+151.7%
echo-ws-pipeline	512	5,485,623	6538.2%	231MiB	-2.6%	+725.0%
echo-ws-pipeline	4096	5,719,510	6425.2%	339MiB	-4.9%	+340.3%
echo-ws-pipeline	16384	5,675,472	6593.2%	621MiB	-2.7%	+163.1%

Full log

  Thread Stats   Avg      p50      p90      p99    p99.9
    Latency   1.49ms    388us   1.59ms   23.70ms   33.10ms

  27369483 frames sent in 5.00s, 27369528 frames received
  Throughput: 5.47M req/s
  Bandwidth:  36.53MB/s
  WS upgrades: 512
  WS frames:   27369528
  Latency samples: 27369456 / 27369528 responses (100.0%)
[info] CPU 6528.7% | Mem 235MiB

[run 3/3]
gcannon v0.5.3 [WS]
  Target:    localhost:8080/ws
  Threads:   64
  Conns:     512 (8/thread)
  Pipeline:  16
  Req/conn:  unlimited (keep-alive)
  Expected:  200
  Duration:  5s


  Thread Stats   Avg      p50      p90      p99    p99.9
    Latency   1.49ms    389us   1.59ms   23.20ms   32.20ms

  27370899 frames sent in 5.00s, 27370949 frames received
  Throughput: 5.47M req/s
  Bandwidth:  36.53MB/s
  WS upgrades: 512
  WS frames:   27370949
  Latency samples: 27370846 / 27370949 responses (100.0%)
[info] CPU 6488.3% | Mem 238MiB

=== Best: 5485623 req/s (CPU: 6538.2%, Mem: 231MiB) ===
[info] saved results/echo-ws-pipeline/512/wtx-ws.json
httparena-bench-wtx-ws
httparena-bench-wtx-ws

==============================================
=== wtx-ws / echo-ws-pipeline / 4096c (tool=gcannon) ===
==============================================
[info] ws-only framework — skipping HTTP probe (sleep 2s for startup)

[run 1/3]
gcannon v0.5.3 [WS]
  Target:    localhost:8080/ws
  Threads:   64
  Conns:     4096 (64/thread)
  Pipeline:  16
  Req/conn:  unlimited (keep-alive)
  Expected:  200
  Duration:  5s


  Thread Stats   Avg      p50      p90      p99    p99.9
    Latency   11.23ms   4.47ms   19.50ms   120.60ms   155.30ms

  28650783 frames sent in 5.00s, 28597550 frames received
  Throughput: 5.72M req/s
  Bandwidth:  38.25MB/s
  WS upgrades: 4096
  WS frames:   28597550
  Latency samples: 28597142 / 28597550 responses (100.0%)
[info] CPU 6425.2% | Mem 339MiB

[run 2/3]
gcannon v0.5.3 [WS]
  Target:    localhost:8080/ws
  Threads:   64
  Conns:     4096 (64/thread)
  Pipeline:  16
  Req/conn:  unlimited (keep-alive)
  Expected:  200
  Duration:  5s


  Thread Stats   Avg      p50      p90      p99    p99.9
    Latency   11.41ms   3.94ms   20.40ms   125.10ms   162.10ms

  28551648 frames sent in 5.00s, 28548626 frames received
  Throughput: 5.71M req/s
  Bandwidth:  38.11MB/s
  WS upgrades: 4096
  WS frames:   28548626
  Latency samples: 28548472 / 28548626 responses (100.0%)
[info] CPU 6559.8% | Mem 350MiB

[run 3/3]
gcannon v0.5.3 [WS]
  Target:    localhost:8080/ws
  Threads:   64
  Conns:     4096 (64/thread)
  Pipeline:  16
  Req/conn:  unlimited (keep-alive)
  Expected:  200
  Duration:  5s


  Thread Stats   Avg      p50      p90      p99    p99.9
    Latency   11.42ms   3.50ms   19.60ms   133.10ms   173.30ms

  28505759 frames sent in 5.00s, 28478162 frames received
  Throughput: 5.69M req/s
  Bandwidth:  38.06MB/s
  WS upgrades: 4096
  WS frames:   28478162
  Latency samples: 28478055 / 28478162 responses (100.0%)
[info] CPU 6406.0% | Mem 357MiB

=== Best: 5719510 req/s (CPU: 6425.2%, Mem: 339MiB) ===
[info] saved results/echo-ws-pipeline/4096/wtx-ws.json
httparena-bench-wtx-ws
httparena-bench-wtx-ws

==============================================
=== wtx-ws / echo-ws-pipeline / 16384c (tool=gcannon) ===
==============================================
[info] ws-only framework — skipping HTTP probe (sleep 2s for startup)

[run 1/3]
gcannon v0.5.3 [WS]
  Target:    localhost:8080/ws
  Threads:   64
  Conns:     16384 (256/thread)
  Pipeline:  16
  Req/conn:  unlimited (keep-alive)
  Expected:  200
  Duration:  5s


  Thread Stats   Avg      p50      p90      p99    p99.9
    Latency   38.77ms   24.70ms   93.30ms   211.40ms   341.90ms

  28598800 frames sent in 5.00s, 28336749 frames received
  Throughput: 5.67M req/s
  Bandwidth:  38.23MB/s
  WS upgrades: 16382
  WS frames:   28336749
  Latency samples: 28336749 / 28336749 responses (100.0%)
[info] CPU 6354.6% | Mem 594MiB

[run 2/3]
gcannon v0.5.3 [WS]
  Target:    localhost:8080/ws
  Threads:   64
  Conns:     16384 (256/thread)
  Pipeline:  16
  Req/conn:  unlimited (keep-alive)
  Expected:  200
  Duration:  5s


  Thread Stats   Avg      p50      p90      p99    p99.9
    Latency   45.58ms   28.90ms   110.00ms   236.00ms   361.30ms

  28639442 frames sent in 5.00s, 28377364 frames received
  Throughput: 5.67M req/s
  Bandwidth:  38.28MB/s
  WS upgrades: 16384
  WS frames:   28377364
  Latency samples: 28377364 / 28377364 responses (100.0%)
[info] CPU 6593.2% | Mem 621MiB

[run 3/3]
gcannon v0.5.3 [WS]
  Target:    localhost:8080/ws
  Threads:   64
  Conns:     16384 (256/thread)
  Pipeline:  16
  Req/conn:  unlimited (keep-alive)
  Expected:  200
  Duration:  5s


  Thread Stats   Avg      p50      p90      p99    p99.9
    Latency   45.64ms   28.90ms   110.20ms   233.90ms   363.30ms

  28567905 frames sent in 5.00s, 28305819 frames received
  Throughput: 5.66M req/s
  Bandwidth:  38.18MB/s
  WS upgrades: 16384
  WS frames:   28305819
  Latency samples: 28305819 / 28305819 responses (100.0%)
[info] CPU 6360.8% | Mem 639MiB

=== Best: 5675472 req/s (CPU: 6593.2%, Mem: 621MiB) ===
[info] saved results/echo-ws-pipeline/16384/wtx-ws.json
httparena-bench-wtx-ws
httparena-bench-wtx-ws
[info] rebuilding site/data/*.json
[updated] /home/diogo/actions-runner/_work/HttpArena/HttpArena/site/data/frameworks.json
[updated] /home/diogo/actions-runner/_work/HttpArena/HttpArena/site/data/echo-ws-16384.json
[updated] /home/diogo/actions-runner/_work/HttpArena/HttpArena/site/data/echo-ws-4096.json
[updated] /home/diogo/actions-runner/_work/HttpArena/HttpArena/site/data/echo-ws-512.json
[updated] /home/diogo/actions-runner/_work/HttpArena/HttpArena/site/data/echo-ws-pipeline-16384.json
[updated] /home/diogo/actions-runner/_work/HttpArena/HttpArena/site/data/echo-ws-pipeline-4096.json
[updated] /home/diogo/actions-runner/_work/HttpArena/HttpArena/site/data/echo-ws-pipeline-512.json
[updated] /home/diogo/actions-runner/_work/HttpArena/HttpArena/site/data/current.json
[info] done
[info] restoring loopback MTU to 65536

c410-f3r · 2026-06-29T11:54:01Z

Runtime per thread created in a thread, build, new_current_thread

/benchmark -f wtx-ws

github-actions · 2026-06-29T11:54:09Z

👋 /benchmark request received. A collaborator will review and approve the run.

github-actions · 2026-06-29T11:59:02Z

Benchmark Results

Framework: wtx-ws | Test: all tests

Test	Conn	RPS	CPU	Mem	Δ RPS	Δ Mem
echo-ws	512	4,033,988	6336.8%	40MiB	-0.2%	+17.6%
echo-ws	4096	4,088,350	6477.1%	116MiB	-2.0%	+41.5%
echo-ws	16384	3,694,075	6481.6%	357MiB	-3.2%	+47.5%
echo-ws-pipeline	512	5,537,457	6478.5%	34MiB	-1.7%	+21.4%
echo-ws-pipeline	4096	5,909,514	6655.3%	112MiB	-1.7%	+45.5%
echo-ws-pipeline	16384	5,598,118	6627.5%	354MiB	-4.0%	+50.0%

Full log

  Thread Stats   Avg      p50      p90      p99    p99.9
    Latency   1.49ms   1.47ms   2.29ms   3.30ms   5.26ms

  27493208 frames sent in 5.00s, 27493296 frames received
  Throughput: 5.50M req/s
  Bandwidth:  36.70MB/s
  WS upgrades: 512
  WS frames:   27493296
  Latency samples: 27506909 / 27493296 responses (100.0%)
[info] CPU 6564.7% | Mem 39MiB

[run 3/3]
gcannon v0.5.3 [WS]
  Target:    localhost:8080/ws
  Threads:   64
  Conns:     512 (8/thread)
  Pipeline:  16
  Req/conn:  unlimited (keep-alive)
  Expected:  200
  Duration:  5s


  Thread Stats   Avg      p50      p90      p99    p99.9
    Latency   1.50ms   1.47ms   2.32ms   2.99ms   5.02ms

  27221937 frames sent in 5.00s, 27222005 frames received
  Throughput: 5.44M req/s
  Bandwidth:  36.34MB/s
  WS upgrades: 512
  WS frames:   27222005
  Latency samples: 27221881 / 27222005 responses (100.0%)
[info] CPU 6398.2% | Mem 42MiB

=== Best: 5537457 req/s (CPU: 6478.5%, Mem: 34MiB) ===
[info] saved results/echo-ws-pipeline/512/wtx-ws.json
httparena-bench-wtx-ws
httparena-bench-wtx-ws

==============================================
=== wtx-ws / echo-ws-pipeline / 4096c (tool=gcannon) ===
==============================================
[info] ws-only framework — skipping HTTP probe (sleep 2s for startup)

[run 1/3]
gcannon v0.5.3 [WS]
  Target:    localhost:8080/ws
  Threads:   64
  Conns:     4096 (64/thread)
  Pipeline:  16
  Req/conn:  unlimited (keep-alive)
  Expected:  200
  Duration:  5s


  Thread Stats   Avg      p50      p90      p99    p99.9
    Latency   11.10ms   10.90ms   13.10ms   14.60ms   25.40ms

  29195377 frames sent in 5.00s, 29132976 frames received
  Throughput: 5.82M req/s
  Bandwidth:  38.98MB/s
  WS upgrades: 4096
  WS frames:   29132976
  Latency samples: 29132976 / 29132976 responses (100.0%)
[info] CPU 6287.2% | Mem 105MiB

[run 2/3]
gcannon v0.5.3 [WS]
  Target:    localhost:8080/ws
  Threads:   64
  Conns:     4096 (64/thread)
  Pipeline:  16
  Req/conn:  unlimited (keep-alive)
  Expected:  200
  Duration:  5s


  Thread Stats   Avg      p50      p90      p99    p99.9
    Latency   11.09ms   11.00ms   13.20ms   14.60ms   24.30ms

  29550464 frames sent in 5.00s, 29547574 frames received
  Throughput: 5.91M req/s
  Bandwidth:  39.44MB/s
  WS upgrades: 4094
  WS frames:   29547574
  Latency samples: 29547415 / 29547574 responses (100.0%)
[info] CPU 6655.3% | Mem 112MiB

[run 3/3]
gcannon v0.5.3 [WS]
  Target:    localhost:8080/ws
  Threads:   64
  Conns:     4096 (64/thread)
  Pipeline:  16
  Req/conn:  unlimited (keep-alive)
  Expected:  200
  Duration:  5s


  Thread Stats   Avg      p50      p90      p99    p99.9
    Latency   11.13ms   11.00ms   13.10ms   14.60ms   25.00ms

  29443758 frames sent in 5.00s, 29441753 frames received
  Throughput: 5.89M req/s
  Bandwidth:  39.30MB/s
  WS upgrades: 4096
  WS frames:   29441753
  Latency samples: 29441550 / 29441753 responses (100.0%)
[info] CPU 6397.8% | Mem 114MiB

=== Best: 5909514 req/s (CPU: 6655.3%, Mem: 112MiB) ===
[info] saved results/echo-ws-pipeline/4096/wtx-ws.json
httparena-bench-wtx-ws
httparena-bench-wtx-ws

==============================================
=== wtx-ws / echo-ws-pipeline / 16384c (tool=gcannon) ===
==============================================
[info] ws-only framework — skipping HTTP probe (sleep 2s for startup)

[run 1/3]
gcannon v0.5.3 [WS]
  Target:    localhost:8080/ws
  Threads:   64
  Conns:     16384 (256/thread)
  Pipeline:  16
  Req/conn:  unlimited (keep-alive)
  Expected:  200
  Duration:  5s


  Thread Stats   Avg      p50      p90      p99    p99.9
    Latency   45.76ms   45.20ms   50.10ms   53.60ms   86.10ms

  27900038 frames sent in 5.00s, 27641743 frames received
  Throughput: 5.53M req/s
  Bandwidth:  37.29MB/s
  WS upgrades: 16384
  WS frames:   27641743
  Latency samples: 27641743 / 27641743 responses (100.0%)
[info] CPU 6163.6% | Mem 349MiB

[run 2/3]
gcannon v0.5.3 [WS]
  Target:    localhost:8080/ws
  Threads:   64
  Conns:     16384 (256/thread)
  Pipeline:  16
  Req/conn:  unlimited (keep-alive)
  Expected:  200
  Duration:  5s


  Thread Stats   Avg      p50      p90      p99    p99.9
    Latency   46.26ms   45.80ms   50.80ms   75.40ms   96.00ms

  28252653 frames sent in 5.00s, 27990591 frames received
  Throughput: 5.60M req/s
  Bandwidth:  37.76MB/s
  WS upgrades: 16384
  WS frames:   27990591
  Latency samples: 27990591 / 27990591 responses (100.0%)
[info] CPU 6627.5% | Mem 354MiB

[run 3/3]
gcannon v0.5.3 [WS]
  Target:    localhost:8080/ws
  Threads:   64
  Conns:     16384 (256/thread)
  Pipeline:  16
  Req/conn:  unlimited (keep-alive)
  Expected:  200
  Duration:  5s


  Thread Stats   Avg      p50      p90      p99    p99.9
    Latency   46.38ms   46.00ms   51.20ms   79.20ms   98.90ms

  28206846 frames sent in 5.00s, 27945175 frames received
  Throughput: 5.59M req/s
  Bandwidth:  37.70MB/s
  WS upgrades: 16384
  WS frames:   27945175
  Latency samples: 27945175 / 27945175 responses (100.0%)
[info] CPU 6238.9% | Mem 359MiB

=== Best: 5598118 req/s (CPU: 6627.5%, Mem: 354MiB) ===
[info] saved results/echo-ws-pipeline/16384/wtx-ws.json
httparena-bench-wtx-ws
httparena-bench-wtx-ws
[info] rebuilding site/data/*.json
[updated] /home/diogo/actions-runner/_work/HttpArena/HttpArena/site/data/frameworks.json
[updated] /home/diogo/actions-runner/_work/HttpArena/HttpArena/site/data/echo-ws-16384.json
[updated] /home/diogo/actions-runner/_work/HttpArena/HttpArena/site/data/echo-ws-4096.json
[updated] /home/diogo/actions-runner/_work/HttpArena/HttpArena/site/data/echo-ws-512.json
[updated] /home/diogo/actions-runner/_work/HttpArena/HttpArena/site/data/echo-ws-pipeline-16384.json
[updated] /home/diogo/actions-runner/_work/HttpArena/HttpArena/site/data/echo-ws-pipeline-4096.json
[updated] /home/diogo/actions-runner/_work/HttpArena/HttpArena/site/data/echo-ws-pipeline-512.json
[updated] /home/diogo/actions-runner/_work/HttpArena/HttpArena/site/data/current.json
[info] done
[info] restoring loopback MTU to 65536

c410-f3r · 2026-06-29T12:25:47Z

Runtime per thread created in a thread, build_local, new_current_thread (spawn_local)

/benchmark -f wtx-ws

github-actions · 2026-06-29T12:25:55Z

👋 /benchmark request received. A collaborator will review and approve the run.

github-actions · 2026-06-29T12:30:06Z

Benchmark Results

Framework: wtx-ws | Test: all tests

Test	Conn	RPS	CPU	Mem	Δ RPS	Δ Mem
echo-ws	512	4,057,881	6295.8%	42MiB	+0.4%	+23.5%
echo-ws	4096	4,125,126	6477.5%	117MiB	-1.1%	+42.7%
echo-ws	16384	3,628,785	6074.4%	359MiB	-4.9%	+48.3%
echo-ws-pipeline	512	5,509,412	6537.0%	35MiB	-2.2%	+25.0%
echo-ws-pipeline	4096	5,862,595	6298.8%	114MiB	-2.5%	+48.1%
echo-ws-pipeline	16384	5,713,723	6622.0%	357MiB	-2.0%	+51.3%

Full log


  27473203 frames sent in 5.00s, 27473261 frames received
  Throughput: 5.49M req/s
  Bandwidth:  36.67MB/s
  WS upgrades: 512
  WS frames:   27473261
  Latency samples: 27473196 / 27473261 responses (100.0%)
[info] CPU 6589.8% | Mem 39MiB

[run 3/3]
gcannon v0.5.3 [WS]
  Target:    localhost:8080/ws
  Threads:   64
  Conns:     512 (8/thread)
  Pipeline:  16
  Req/conn:  unlimited (keep-alive)
  Expected:  200
  Duration:  5s


  Thread Stats   Avg      p50      p90      p99    p99.9
    Latency   1.50ms   1.52ms   2.22ms   2.72ms   4.85ms

  27343384 frames sent in 5.00s, 27343447 frames received
  Throughput: 5.47M req/s
  Bandwidth:  36.51MB/s
  WS upgrades: 512
  WS frames:   27343447
  Latency samples: 27343352 / 27343447 responses (100.0%)
[info] CPU 6424.4% | Mem 43MiB

=== Best: 5509412 req/s (CPU: 6537.0%, Mem: 35MiB) ===
[info] saved results/echo-ws-pipeline/512/wtx-ws.json
httparena-bench-wtx-ws
httparena-bench-wtx-ws

==============================================
=== wtx-ws / echo-ws-pipeline / 4096c (tool=gcannon) ===
==============================================
[info] ws-only framework — skipping HTTP probe (sleep 2s for startup)

[run 1/3]
gcannon v0.5.3 [WS]
  Target:    localhost:8080/ws
  Threads:   64
  Conns:     4096 (64/thread)
  Pipeline:  16
  Req/conn:  unlimited (keep-alive)
  Expected:  200
  Duration:  5s


  Thread Stats   Avg      p50      p90      p99    p99.9
    Latency   11.26ms   10.90ms   13.00ms   24.00ms   37.50ms

  28920241 frames sent in 5.00s, 28856835 frames received
  Throughput: 5.77M req/s
  Bandwidth:  38.61MB/s
  WS upgrades: 4096
  WS frames:   28856835
  Latency samples: 28856835 / 28856835 responses (100.0%)
[info] CPU 6114.4% | Mem 106MiB

[run 2/3]
gcannon v0.5.3 [WS]
  Target:    localhost:8080/ws
  Threads:   64
  Conns:     4096 (64/thread)
  Pipeline:  16
  Req/conn:  unlimited (keep-alive)
  Expected:  200
  Duration:  5s


  Thread Stats   Avg      p50      p90      p99    p99.9
    Latency   11.24ms   11.00ms   13.20ms   23.50ms   28.30ms

  29131968 frames sent in 5.00s, 29127910 frames received
  Throughput: 5.82M req/s
  Bandwidth:  38.88MB/s
  WS upgrades: 4096
  WS frames:   29127911
  Latency samples: 29127814 / 29127910 responses (100.0%)

  WARNING: 18446744073709551615/29127910 responses (63330132761703.6%) had unexpected status (expected 2xx)
[info] CPU 6476.5% | Mem 113MiB

[run 3/3]
gcannon v0.5.3 [WS]
  Target:    localhost:8080/ws
  Threads:   64
  Conns:     4096 (64/thread)
  Pipeline:  16
  Req/conn:  unlimited (keep-alive)
  Expected:  200
  Duration:  5s


  Thread Stats   Avg      p50      p90      p99    p99.9
    Latency   11.17ms   11.00ms   13.30ms   19.20ms   26.30ms

  29318044 frames sent in 5.00s, 29312979 frames received
  Throughput: 5.86M req/s
  Bandwidth:  39.13MB/s
  WS upgrades: 4096
  WS frames:   29312979
  Latency samples: 29314331 / 29312979 responses (100.0%)
[info] CPU 6298.8% | Mem 114MiB

=== Best: 5862595 req/s (CPU: 6298.8%, Mem: 114MiB) ===
[info] saved results/echo-ws-pipeline/4096/wtx-ws.json
httparena-bench-wtx-ws
httparena-bench-wtx-ws

==============================================
=== wtx-ws / echo-ws-pipeline / 16384c (tool=gcannon) ===
==============================================
[info] ws-only framework — skipping HTTP probe (sleep 2s for startup)

[run 1/3]
gcannon v0.5.3 [WS]
  Target:    localhost:8080/ws
  Threads:   64
  Conns:     16384 (256/thread)
  Pipeline:  16
  Req/conn:  unlimited (keep-alive)
  Expected:  200
  Duration:  5s


  Thread Stats   Avg      p50      p90      p99    p99.9
    Latency   45.07ms   45.20ms   49.90ms   52.10ms   62.70ms

  28297619 frames sent in 5.00s, 28035574 frames received
  Throughput: 5.61M req/s
  Bandwidth:  37.82MB/s
  WS upgrades: 16384
  WS frames:   28035574
  Latency samples: 28035574 / 28035574 responses (100.0%)
[info] CPU 6116.5% | Mem 349MiB

[run 2/3]
gcannon v0.5.3 [WS]
  Target:    localhost:8080/ws
  Threads:   64
  Conns:     16384 (256/thread)
  Pipeline:  16
  Req/conn:  unlimited (keep-alive)
  Expected:  200
  Duration:  5s


  Thread Stats   Avg      p50      p90      p99    p99.9
    Latency   45.58ms   45.00ms   49.00ms   67.20ms   95.30ms

  28830666 frames sent in 5.00s, 28568615 frames received
  Throughput: 5.71M req/s
  Bandwidth:  38.53MB/s
  WS upgrades: 16384
  WS frames:   28568615
  Latency samples: 28568615 / 28568615 responses (100.0%)
[info] CPU 6622.0% | Mem 357MiB

[run 3/3]
gcannon v0.5.3 [WS]
  Target:    localhost:8080/ws
  Threads:   64
  Conns:     16384 (256/thread)
  Pipeline:  16
  Req/conn:  unlimited (keep-alive)
  Expected:  200
  Duration:  5s


  Thread Stats   Avg      p50      p90      p99    p99.9
    Latency   46.33ms   45.60ms   50.40ms   86.60ms   98.20ms

  28248076 frames sent in 5.00s, 27990105 frames received
  Throughput: 5.60M req/s
  Bandwidth:  37.76MB/s
  WS upgrades: 16384
  WS frames:   27990105
  Latency samples: 27990105 / 27990105 responses (100.0%)
[info] CPU 6083.2% | Mem 359MiB

=== Best: 5713723 req/s (CPU: 6622.0%, Mem: 357MiB) ===
[info] saved results/echo-ws-pipeline/16384/wtx-ws.json
httparena-bench-wtx-ws
httparena-bench-wtx-ws
[info] rebuilding site/data/*.json
[updated] /home/diogo/actions-runner/_work/HttpArena/HttpArena/site/data/frameworks.json
[updated] /home/diogo/actions-runner/_work/HttpArena/HttpArena/site/data/echo-ws-16384.json
[updated] /home/diogo/actions-runner/_work/HttpArena/HttpArena/site/data/echo-ws-4096.json
[updated] /home/diogo/actions-runner/_work/HttpArena/HttpArena/site/data/echo-ws-512.json
[updated] /home/diogo/actions-runner/_work/HttpArena/HttpArena/site/data/echo-ws-pipeline-16384.json
[updated] /home/diogo/actions-runner/_work/HttpArena/HttpArena/site/data/echo-ws-pipeline-4096.json
[updated] /home/diogo/actions-runner/_work/HttpArena/HttpArena/site/data/echo-ws-pipeline-512.json
[updated] /home/diogo/actions-runner/_work/HttpArena/HttpArena/site/data/current.json
[info] done
[info] restoring loopback MTU to 65536

c410-f3r · 2026-06-29T12:50:35Z

Runtime per thread created in a thread, build_local, new_current_thread (core_affinity)

/benchmark -f wtx-ws

github-actions · 2026-06-29T12:50:45Z

👋 /benchmark request received. A collaborator will review and approve the run.

github-actions · 2026-06-29T12:55:06Z

Benchmark Results

Framework: wtx-ws | Test: all tests

Test	Conn	RPS	CPU	Mem	Δ RPS	Δ Mem
echo-ws	512	4,013,004	6399.1%	43MiB	-0.7%	+26.5%
echo-ws	4096	4,096,964	6465.0%	117MiB	-1.8%	+42.7%
echo-ws	16384	3,728,981	6468.1%	360MiB	-2.3%	+48.8%
echo-ws-pipeline	512	5,458,371	6597.2%	34MiB	-3.1%	+21.4%
echo-ws-pipeline	4096	5,929,358	6709.6%	111MiB	-1.4%	+44.2%
echo-ws-pipeline	16384	5,750,213	6709.1%	358MiB	-1.4%	+51.7%

Full log

  Thread Stats   Avg      p50      p90      p99    p99.9
    Latency   1.50ms   1.50ms   2.16ms   2.96ms   3.14ms

  27184762 frames sent in 5.00s, 27184806 frames received
  Throughput: 5.44M req/s
  Bandwidth:  36.29MB/s
  WS upgrades: 512
  WS frames:   27184806
  Latency samples: 27184682 / 27184806 responses (100.0%)
[info] CPU 6631.2% | Mem 39MiB

[run 3/3]
gcannon v0.5.3 [WS]
  Target:    localhost:8080/ws
  Threads:   64
  Conns:     512 (8/thread)
  Pipeline:  16
  Req/conn:  unlimited (keep-alive)
  Expected:  200
  Duration:  5s


  Thread Stats   Avg      p50      p90      p99    p99.9
    Latency   1.51ms   1.48ms   2.54ms   2.97ms   3.15ms

  27141177 frames sent in 5.00s, 27141241 frames received
  Throughput: 5.43M req/s
  Bandwidth:  36.23MB/s
  WS upgrades: 512
  WS frames:   27141241
  Latency samples: 27141136 / 27141241 responses (100.0%)
[info] CPU 6602.3% | Mem 42MiB

=== Best: 5458371 req/s (CPU: 6597.2%, Mem: 34MiB) ===
[info] saved results/echo-ws-pipeline/512/wtx-ws.json
httparena-bench-wtx-ws
httparena-bench-wtx-ws

==============================================
=== wtx-ws / echo-ws-pipeline / 4096c (tool=gcannon) ===
==============================================
[info] ws-only framework — skipping HTTP probe (sleep 2s for startup)

[run 1/3]
gcannon v0.5.3 [WS]
  Target:    localhost:8080/ws
  Threads:   64
  Conns:     4096 (64/thread)
  Pipeline:  16
  Req/conn:  unlimited (keep-alive)
  Expected:  200
  Duration:  5s


  Thread Stats   Avg      p50      p90      p99    p99.9
    Latency   11.03ms   11.00ms   12.60ms   14.40ms   14.70ms

  29450208 frames sent in 5.00s, 29387839 frames received
  Throughput: 5.88M req/s
  Bandwidth:  39.32MB/s
  WS upgrades: 4096
  WS frames:   29387839
  Latency samples: 29387839 / 29387839 responses (100.0%)
[info] CPU 6463.1% | Mem 106MiB

[run 2/3]
gcannon v0.5.3 [WS]
  Target:    localhost:8080/ws
  Threads:   64
  Conns:     4096 (64/thread)
  Pipeline:  16
  Req/conn:  unlimited (keep-alive)
  Expected:  200
  Duration:  5s


  Thread Stats   Avg      p50      p90      p99    p99.9
    Latency   11.05ms   11.00ms   13.00ms   13.90ms   14.20ms

  29648759 frames sent in 5.00s, 29646793 frames received
  Throughput: 5.93M req/s
  Bandwidth:  39.57MB/s
  WS upgrades: 4096
  WS frames:   29646793
  Latency samples: 29646690 / 29646793 responses (100.0%)
[info] CPU 6709.6% | Mem 111MiB

[run 3/3]
gcannon v0.5.3 [WS]
  Target:    localhost:8080/ws
  Threads:   64
  Conns:     4096 (64/thread)
  Pipeline:  16
  Req/conn:  unlimited (keep-alive)
  Expected:  200
  Duration:  5s


  Thread Stats   Avg      p50      p90      p99    p99.9
    Latency   11.06ms   11.00ms   12.90ms   14.60ms   14.90ms

  29608747 frames sent in 5.00s, 29602661 frames received
  Throughput: 5.92M req/s
  Bandwidth:  39.52MB/s
  WS upgrades: 4096
  WS frames:   29602661
  Latency samples: 29602559 / 29602661 responses (100.0%)
[info] CPU 6555.8% | Mem 113MiB

=== Best: 5929358 req/s (CPU: 6709.6%, Mem: 111MiB) ===
[info] saved results/echo-ws-pipeline/4096/wtx-ws.json
httparena-bench-wtx-ws
httparena-bench-wtx-ws

==============================================
=== wtx-ws / echo-ws-pipeline / 16384c (tool=gcannon) ===
==============================================
[info] ws-only framework — skipping HTTP probe (sleep 2s for startup)

[run 1/3]
gcannon v0.5.3 [WS]
  Target:    localhost:8080/ws
  Threads:   64
  Conns:     16384 (256/thread)
  Pipeline:  16
  Req/conn:  unlimited (keep-alive)
  Expected:  200
  Duration:  5s


  Thread Stats   Avg      p50      p90      p99    p99.9
    Latency   45.02ms   45.20ms   49.20ms   53.20ms   54.00ms

  28212199 frames sent in 5.00s, 27950138 frames received
  Throughput: 5.59M req/s
  Bandwidth:  37.71MB/s
  WS upgrades: 16384
  WS frames:   27950138
  Latency samples: 27950138 / 27950138 responses (100.0%)
[info] CPU 6237.6% | Mem 350MiB

[run 2/3]
gcannon v0.5.3 [WS]
  Target:    localhost:8080/ws
  Threads:   64
  Conns:     16384 (256/thread)
  Pipeline:  16
  Req/conn:  unlimited (keep-alive)
  Expected:  200
  Duration:  5s


  Thread Stats   Avg      p50      p90      p99    p99.9
    Latency   44.96ms   45.40ms   48.70ms   52.80ms   53.60ms

  29013128 frames sent in 5.00s, 28751066 frames received
  Throughput: 5.75M req/s
  Bandwidth:  38.78MB/s
  WS upgrades: 16384
  WS frames:   28751066
  Latency samples: 28751066 / 28751066 responses (100.0%)
[info] CPU 6709.1% | Mem 358MiB

[run 3/3]
gcannon v0.5.3 [WS]
  Target:    localhost:8080/ws
  Threads:   64
  Conns:     16384 (256/thread)
  Pipeline:  16
  Req/conn:  unlimited (keep-alive)
  Expected:  200
  Duration:  5s


  Thread Stats   Avg      p50      p90      p99    p99.9
    Latency   45.19ms   45.40ms   49.20ms   53.10ms   53.90ms

  28931718 frames sent in 5.00s, 28669667 frames received
  Throughput: 5.73M req/s
  Bandwidth:  38.67MB/s
  WS upgrades: 16384
  WS frames:   28669667
  Latency samples: 28669667 / 28669667 responses (100.0%)
[info] CPU 6439.9% | Mem 358MiB

=== Best: 5750213 req/s (CPU: 6709.1%, Mem: 358MiB) ===
[info] saved results/echo-ws-pipeline/16384/wtx-ws.json
httparena-bench-wtx-ws
httparena-bench-wtx-ws
[info] rebuilding site/data/*.json
[updated] /home/diogo/actions-runner/_work/HttpArena/HttpArena/site/data/frameworks.json
[updated] /home/diogo/actions-runner/_work/HttpArena/HttpArena/site/data/echo-ws-16384.json
[updated] /home/diogo/actions-runner/_work/HttpArena/HttpArena/site/data/echo-ws-4096.json
[updated] /home/diogo/actions-runner/_work/HttpArena/HttpArena/site/data/echo-ws-512.json
[updated] /home/diogo/actions-runner/_work/HttpArena/HttpArena/site/data/echo-ws-pipeline-16384.json
[updated] /home/diogo/actions-runner/_work/HttpArena/HttpArena/site/data/echo-ws-pipeline-4096.json
[updated] /home/diogo/actions-runner/_work/HttpArena/HttpArena/site/data/echo-ws-pipeline-512.json
[updated] /home/diogo/actions-runner/_work/HttpArena/HttpArena/site/data/current.json
[info] done
[info] restoring loopback MTU to 65536

c410-f3r · 2026-06-29T12:57:14Z

OK. I guess this is it.

The creation of a new runtime inside an existing task creates conflict, which is expected.
The use of multi-thread runtimes in a per-thread runtime explodes memory usage and also doesn't help with performance, which is also expected

Thanks for lending a bit of computing power from your 64 cores machine :)

c410-f3r requested review from Kaliumhexacyanoferrat and MDA2AV as code owners June 29, 2026 11:02

c410-f3r force-pushed the misc branch from 70d1b19 to 17e295a Compare June 29, 2026 11:06

Test 1

7ae57f7

c410-f3r force-pushed the misc branch from 17e295a to 7ae57f7 Compare June 29, 2026 11:07

Test 2

69cbc3d

Test 3

20d259b

Test 4

4c94de8

c410-f3r force-pushed the misc branch from 39d02db to 4c94de8 Compare June 29, 2026 12:20

c410-f3r added 2 commits June 29, 2026 09:43

Test 5

719bb52

Update

0ee4dc3

c410-f3r closed this Jun 29, 2026

Conversation

c410-f3r commented Jun 29, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

MDA2AV commented Jun 29, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

c410-f3r commented Jun 29, 2026

Uh oh!

c410-f3r commented Jun 29, 2026

Uh oh!

github-actions Bot commented Jun 29, 2026

Uh oh!

github-actions Bot commented Jun 29, 2026

Benchmark Results

Uh oh!

c410-f3r commented Jun 29, 2026

Uh oh!

github-actions Bot commented Jun 29, 2026

Uh oh!

github-actions Bot commented Jun 29, 2026

Benchmark Results

Uh oh!

c410-f3r commented Jun 29, 2026

Uh oh!

github-actions Bot commented Jun 29, 2026

Uh oh!

github-actions Bot commented Jun 29, 2026

Benchmark Results

Uh oh!

c410-f3r commented Jun 29, 2026

Uh oh!

github-actions Bot commented Jun 29, 2026

Uh oh!

github-actions Bot commented Jun 29, 2026

Benchmark Results

Uh oh!

c410-f3r commented Jun 29, 2026

Uh oh!

github-actions Bot commented Jun 29, 2026

Uh oh!

github-actions Bot commented Jun 29, 2026

Benchmark Results

Uh oh!

c410-f3r commented Jun 29, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

c410-f3r commented Jun 29, 2026 •

edited

Loading

MDA2AV commented Jun 29, 2026 •

edited

Loading