Skip to content

deps(cargo): bump zstd from 0.13.3 to 0.14.0 - #212

Merged
github-actions[bot] merged 1 commit into
mainfrom
dependabot/cargo/zstd-0.14.0
Sep 14, 2026
Merged

github-actions[bot] merged 1 commit into
mainfrom
dependabot/cargo/zstd-0.14.0

Conversation

@dependabot

@dependabot dependabot Bot commented on behalf of github Sep 14, 2026

Copy link
Copy Markdown
Contributor

Bumps zstd from 0.13.3 to 0.14.0.

Commits
  • 648acb4 Avoid let...else in the seekable callbacks
  • e1152c1 Bump versions for the next release
  • 7caed6e Derive the usual traits on ResetDirective
  • bf7b1f7 Refuse to reuse a context an error may have left undefined
  • 1565618 Check the target, not the host, for MSVC
  • 8315a62 Return Ok(0) from Read::read for an empty buffer
  • a7cfa93 Keep the std gate on the Cursor WriteBuf impl
  • 9bf1692 Say why the Send and Sync impls hold
  • 681bcc3 Don't truncate Cursor positions on 32-bit targets
  • d5a1fdd Don't hand out a &mut Seekable from AdvancedSeekable
  • Additional commits viewable in compare view

Dependabot compatibility score

Dependabot will resolve any conflicts with this PR as long as you don't alter it yourself. You can also trigger a rebase manually by commenting @dependabot rebase.


Dependabot commands and options

You can trigger Dependabot actions by commenting on this PR:

  • @dependabot rebase will rebase this PR
  • @dependabot recreate will recreate this PR, overwriting any edits that have been made to it
  • @dependabot show <dependency name> ignore conditions will show all of the ignore conditions of the specified dependency
  • @dependabot ignore this major version will close this PR and stop Dependabot creating any more for this major version (unless you reopen the PR or upgrade to it yourself)
  • @dependabot ignore this minor version will close this PR and stop Dependabot creating any more for this minor version (unless you reopen the PR or upgrade to it yourself)
  • @dependabot ignore this dependency will close this PR and stop Dependabot creating any more for this dependency (unless you reopen the PR or upgrade to it yourself)

Bumps [zstd](https://github.com/gyscos/zstd-rs) from 0.13.3 to 0.14.0.
- [Release notes](https://github.com/gyscos/zstd-rs/releases)
- [Commits](gyscos/zstd-rs@v0.13.3...v0.14.0)

---
updated-dependencies:
- dependency-name: zstd
  dependency-version: 0.14.0
  dependency-type: direct:production
  update-type: version-update:semver-minor
...

Signed-off-by: dependabot[bot] <support@github.com>
@dependabot @github

dependabot Bot commented on behalf of github Sep 14, 2026

Copy link
Copy Markdown
Contributor Author

Assignees

The following users could not be added as assignees: lord-patpak. Either the username does not exist or it does not have the correct permissions to be added as an assignee.

Please fix the above issues or remove invalid values from dependabot.yml.

@dependabot dependabot Bot added dependencies Pull requests that update a dependency file rust Pull requests that update rust code labels Sep 14, 2026
@github-actions
github-actions Bot enabled auto-merge (squash) September 14, 2026 06:42
@github-actions

Copy link
Copy Markdown
Contributor

Benchmark gate

Baseline: rolling window of 12 run(s) on AMD EPYC 9V45 96-Core Processor. Current arm: median of 6 repetition(s).

No baseline run matched this runner's CPU model. All verdicts are advisory.

Result: PASS (with tail advisories)

Adaptive_CmdVel@16B — 50000 samples
metric baseline current change band (<=) verdict
median 40.0 ns 40.0 ns +0.0% 65.0 ns INCONCLUSIVE
p95 50.0 ns 60.0 ns +20.0% 100.0 ns INCONCLUSIVE
p99 69.0 ns 70.0 ns +1.4% 169.0 ns INCONCLUSIVE
p99.9 80.0 ns 100.0 ns +25.0% 330.0 ns INCONCLUSIVE
p99.99 — 120.0 ns — — N/A (samples)
max 14.97 µs 3.54 µs -76.4% 29.93 µs INCONCLUSIVE
max_jitter 14.96 µs 3.52 µs -76.4% 29.92 µs INCONCLUSIVE
p99/median 1.73x 1.75x +1.4% 2.23x INCONCLUSIVE
p99.9/median 2.00x 2.50x +25.0% 3.00x INCONCLUSIVE
ns/msg 170.3 ns 139.0 ns -18.4% 220.3 ns INCONCLUSIVE
Adaptive_Imu@304B — 50000 samples
metric baseline current change band (<=) verdict
median 40.0 ns 40.0 ns +0.0% 65.0 ns INCONCLUSIVE
p95 60.0 ns 70.0 ns +16.7% 110.0 ns INCONCLUSIVE
p99 90.0 ns 80.0 ns -11.1% 190.0 ns INCONCLUSIVE
p99.9 110.0 ns 100.0 ns -9.1% 360.0 ns INCONCLUSIVE
p99.99 — 230.0 ns — — N/A (samples)
max 12.82 µs 9.84 µs -23.2% 28.35 µs INCONCLUSIVE
max_jitter 12.80 µs 9.82 µs -23.2% 28.33 µs INCONCLUSIVE
p99/median 2.25x 2.00x -11.1% 2.81x INCONCLUSIVE
p99.9/median 2.75x 2.50x -9.1% 3.75x INCONCLUSIVE
ns/msg 270.3 ns 220.4 ns -18.5% 324.3 ns INCONCLUSIVE
Adaptive_JointCommand@928B — 50000 samples
metric baseline current change band (<=) verdict
median 80.0 ns 70.0 ns -12.5% 105.0 ns IMPROVED
p95 100.0 ns 100.0 ns +0.0% 150.0 ns INCONCLUSIVE
p99 130.0 ns 120.0 ns -7.7% 230.0 ns INCONCLUSIVE
p99.9 160.0 ns 150.0 ns -6.2% 410.0 ns INCONCLUSIVE
p99.99 — 260.0 ns — — N/A (samples)
max 14.04 µs 8.26 µs -41.1% 30.73 µs INCONCLUSIVE
max_jitter 13.99 µs 8.23 µs -41.2% 30.69 µs INCONCLUSIVE
p99/median 1.62x 1.71x +5.5% 2.12x INCONCLUSIVE
p99.9/median 2.00x 2.14x +7.1% 3.00x INCONCLUSIVE
ns/msg 335.0 ns 270.7 ns -19.2% 402.0 ns INCONCLUSIVE
Adaptive_LaserScan@1480B — 10000 samples
metric baseline current change band (<=) verdict
median 89.0 ns 60.0 ns -32.6% 114.0 ns IMPROVED
p95 110.0 ns 90.0 ns -18.2% 160.0 ns IMPROVED
p99 150.0 ns 110.0 ns -26.7% 250.0 ns IMPROVED
p99.9 170.0 ns 130.0 ns -23.5% 420.0 ns INCONCLUSIVE
p99.99 — 200.0 ns — — N/A (samples)
max 10.68 µs 8.22 µs -23.0% 39.88 µs INCONCLUSIVE
max_jitter 10.62 µs 8.19 µs -22.9% 39.82 µs INCONCLUSIVE
p99/median 1.69x 1.67x -1.1% 2.19x INCONCLUSIVE
p99.9/median 1.91x 2.17x +13.4% 2.91x INCONCLUSIVE
ns/msg 689.0 ns 422.5 ns -38.7% 826.8 ns IMPROVED
Topic_cross_process_oneway_paced_stream@1024B — 200000 samples
metric baseline current change band (<=) verdict
median 270.5 ns 180.0 ns -33.5% 377.2 ns IMPROVED
p95 345.5 ns 220.0 ns -36.3% 627.2 ns IMPROVED
p99 597.5 ns 731.0 ns +22.3% 1.05 µs INCONCLUSIVE
p99.9 3.91 µs 12.29 µs +214.3% 15.96 µs INCONCLUSIVE
p99.99 15.83 µs 139.86 µs +783.6% 32.55 µs REGRESSED
max 39.66 µs 181.84 µs +358.4% 109.35 µs REGRESSED
max_jitter 39.37 µs 181.73 µs +361.6% 108.15 µs REGRESSED
p99/median 2.00x 3.17x +58.6% 3.95x INCONCLUSIVE
p99.9/median 14.57x 74.06x +408.5% 65.14x INCONCLUSIVE
ns/msg 10.00 µs 10.00 µs +0.0% 12.00 µs INCONCLUSIVE
Topic_cross_process_oneway_paced_stream@64B — 200000 samples
metric baseline current change band (<=) verdict
median 175.5 ns 135.0 ns -23.1% 237.8 ns IMPROVED
p95 213.0 ns 175.0 ns -17.8% 307.9 ns IMPROVED
p99 436.0 ns 245.0 ns -43.8% 901.5 ns IMPROVED
p99.9 3.00 µs 876.0 ns -70.8% 10.89 µs IMPROVED
p99.99 14.80 µs 7.03 µs -52.5% 48.36 µs IMPROVED
max 56.78 µs 47.41 µs -16.5% 325.23 µs INCONCLUSIVE
max_jitter 56.61 µs 47.15 µs -16.7% 324.45 µs INCONCLUSIVE
p99/median 2.24x 1.77x -21.0% 5.62x INCONCLUSIVE
p99.9/median 17.92x 6.62x -63.0% 55.32x IMPROVED
ns/msg 10.00 µs 10.00 µs +0.0% 12.00 µs INCONCLUSIVE
Topic_cross_process_oneway_pingpong@1024B — 1000000 samples
metric baseline current change band (<=) verdict
median 285.5 ns 185.0 ns -35.2% 347.8 ns IMPROVED
p95 366.0 ns 230.0 ns -37.2% 422.3 ns IMPROVED
p99 413.0 ns 255.0 ns -38.3% 516.8 ns IMPROVED
p99.9 1.44 µs 566.0 ns -60.6% 2.16 µs IMPROVED
p99.99 10.48 µs 5.61 µs -46.5% 15.72 µs INCONCLUSIVE
max 32.66 µs 21.75 µs -33.4% 130.22 µs INCONCLUSIVE
max_jitter 32.45 µs 21.63 µs -33.3% 129.87 µs INCONCLUSIVE
p99/median 1.44x 1.38x -4.5% 1.94x INCONCLUSIVE
p99.9/median 5.10x 3.06x -40.0% 8.28x IMPROVED
ns/msg 347.5 ns 243.5 ns -29.9% 417.0 ns IMPROVED
Topic_cross_process_oneway_pingpong@64B — 1000000 samples
metric baseline current change band (<=) verdict
median 155.0 ns 115.0 ns -25.8% 214.3 ns IMPROVED
p95 185.0 ns 140.0 ns -24.3% 244.3 ns IMPROVED
p99 207.5 ns 155.0 ns -25.3% 307.5 ns IMPROVED
p99.9 789.0 ns 486.0 ns -38.4% 1.31 µs IMPROVED
p99.99 7.78 µs 4.30 µs -44.7% 11.67 µs INCONCLUSIVE
max 23.95 µs 25.07 µs +4.7% 61.53 µs INCONCLUSIVE
max_jitter 23.82 µs 24.99 µs +4.9% 61.39 µs INCONCLUSIVE
p99/median 1.31x 1.30x -0.8% 1.81x INCONCLUSIVE
p99.9/median 4.58x 4.14x -9.6% 7.07x INCONCLUSIVE
ns/msg 204.0 ns 159.5 ns -21.8% 255.2 ns IMPROVED
Topic_cross_process_unpaced_queue_delay@64B — 200000 samples
metric baseline current change band (<=) verdict
median 159.5 ns 100.0 ns -37.3% 275.1 ns IMPROVED
p95 213.0 ns 156.0 ns -26.8% 319.7 ns IMPROVED
p99 248.0 ns 180.0 ns -27.4% 399.2 ns IMPROVED
p99.9 2.46 µs 581.0 ns -76.4% 10.46 µs IMPROVED
p99.99 9.21 µs 5.79 µs -37.1% 13.81 µs INCONCLUSIVE
max 40.09 µs 39.97 µs -0.3% 80.19 µs INCONCLUSIVE
max_jitter 39.99 µs 39.90 µs -0.2% 79.98 µs INCONCLUSIVE
p99/median 1.56x 1.80x +15.5% 2.12x INCONCLUSIVE
p99.9/median 15.74x 5.81x -63.1% 74.18x INCONCLUSIVE
ns/msg 115.4 ns 76.8 ns -33.5% 167.3 ns IMPROVED

Blocking metrics: median, p99/median. Report-only: p95, p99, p99.9, p99.99, max, max_jitter, p99.9/median, ns/msg. Tail metrics past p99.9 are advisory on a shared runner — one host preemption owns max.

Raw benchmark output
 495.7     4216.3     17145.5 |      983        99
  leg A->B (apparent)              1000000      76.3     116.3     166.4     777.3     5734.7     34156.9 |      974       100
  leg B->A (apparent)              1000000      74.0     124.0     164.1     244.2     5452.0     23308.5 |      908       100
  responder turnaround             1000000      20.0      30.0      30.0      40.1      110.2     14591.7 |      744        96
------------------------------------------------------------------------------------------------------------------------------------------
  samples=1000000  warmup=200000  transferred=2000000  lost=0  invalid(TSC-skew)=0  poll cost=7.7 ns  window=0.315 s
  measured window: minor_faults=0 major_faults=0 ctx_switches vol=62 invol=2
  core freq cpu3   (measured window only): start=n/a end=n/a min=0MHz max=0MHz mean=0MHz spread=0.00% (n=0)
  core freq cpu1   (measured window only): start=n/a end=n/a min=0MHz max=0MHz mean=0MHz spread=0.00% (n=0)
  cross-core TSC offset theta_hat = -3.9 ns (clock B - clock A). The `one-way, offset-free` row does not depend on it.

== cross-process ping-pong -- ONE-WAY LATENCY (2 processes, 2 topics, 1 role per handle) ==
   json metric key: `Topic_cross_process_oneway_pingpong` (identity used by the CI gate: `Topic_cross_process_oneway_pingpong@1024B`)
   payload Probe1k = 1024 B (POD=true) | ring capacity 256 slots | topology 1P/1C | cpu3 <-> cpu1 | pid 7341 <-> pid 7351
------------------------------------------------------------------------------------------------------------------------------------------
  metric (ns)                            n       min    median       p99     p99.9     p99.99         max |  n>p99.9  n>p99.99
------------------------------------------------------------------------------------------------------------------------------------------
  one-way, offset-free             1000000     120.2     185.3     255.4     560.8     5358.0     23525.0 |      974       100
  round trip (RTT)                 1000000     270.4     400.6     540.8    1161.7    14371.4     47240.2 |      974       100
  RTT halved (one-way)             1000000     135.2     200.3     270.4     580.9     7185.7     23620.1 |      974       100
  leg A->B (apparent)              1000000     106.3     186.4     266.6     877.5     8328.5     46815.8 |      961        99
  leg B->A (apparent)              1000000     114.0     184.1     284.3     344.4     5902.6     37669.9 |      993        99
  responder turnaround             1000000      20.0      30.0      40.1      40.1      120.2     18938.2 |      749       100
------------------------------------------------------------------------------------------------------------------------------------------
  samples=1000000  warmup=200000  transferred=2000000  lost=0  invalid(TSC-skew)=0  poll cost=29.3 ns  window=0.487 s
  measured window: minor_faults=0 major_faults=0 ctx_switches vol=96 invol=5
  core freq cpu3   (measured window only): start=n/a end=n/a min=0MHz max=0MHz mean=0MHz spread=0.00% (n=0)
  core freq cpu1   (measured window only): start=n/a end=n/a min=0MHz max=0MHz mean=0MHz spread=0.00% (n=0)
  cross-core TSC offset theta_hat = +1.2 ns (clock B - clock A). The `one-way, offset-free` row does not depend on it.

== cross-process paced stream -- ONE-WAY LATENCY (10000 ns publish period, publisher -> subscriber, direct) ==
   json metric key: `Topic_cross_process_oneway_paced_stream` (identity used by the CI gate: `Topic_cross_process_oneway_paced_stream@64B`)
   payload Probe64 = 64 B (POD=true) | ring capacity 256 slots | topology 1P/1C | cpu3 <-> cpu1 | pid 7341 <-> pid 7353
------------------------------------------------------------------------------------------------------------------------------------------
  metric (ns)                            n       min    median       p99     p99.9     p99.99         max |  n>p99.9  n>p99.99
------------------------------------------------------------------------------------------------------------------------------------------
  one-way (assumes TSC coherence)    200000      80.1     130.2     230.3     861.3     7030.5     63364.2 |      196        20
------------------------------------------------------------------------------------------------------------------------------------------
  samples=200000  warmup=20000  transferred=200000  lost=0  invalid(TSC-skew)=0  poll cost=5.4 ns  window=2.000 s
  measured backlog: 0.01 messages deep on the median sample (ring holds 256); delivered 100.0% of transferred
  measured window: minor_faults=0 major_faults=0 ctx_switches vol=395 invol=16
  core freq cpu3   (measured window only): start=n/a end=n/a min=0MHz max=0MHz mean=0MHz spread=0.00% (n=0)
  core freq cpu1   (measured window only): start=n/a end=n/a min=0MHz max=0MHz mean=0MHz spread=0.00% (n=0)
  this row assumes cross-core TSC coherence; theta_hat measured in the ping-pong at this size = -3.9 ns (add to correct).

== cross-process paced stream -- ONE-WAY LATENCY (10000 ns publish period, publisher -> subscriber, direct) ==
   json metric key: `Topic_cross_process_oneway_paced_stream` (identity used by the CI gate: `Topic_cross_process_oneway_paced_stream@1024B`)
   payload Probe1k = 1024 B (POD=true) | ring capacity 256 slots | topology 1P/1C | cpu3 <-> cpu1 | pid 7341 <-> pid 7355
------------------------------------------------------------------------------------------------------------------------------------------
  metric (ns)                            n       min    median       p99     p99.9     p99.99         max |  n>p99.9  n>p99.99
------------------------------------------------------------------------------------------------------------------------------------------
  one-way (assumes TSC coherence)    200000     109.8     179.9     730.7   50985.4   139858.3    201719.5 |      200        20
------------------------------------------------------------------------------------------------------------------------------------------
  samples=200000  warmup=20000  transferred=200000  lost=0  invalid(TSC-skew)=0  poll cost=10.4 ns  window=2.000 s
  measured backlog: 0.02 messages deep on the median sample (ring holds 256); delivered 100.0% of transferred
  measured window: minor_faults=0 major_faults=0 ctx_switches vol=394 invol=19
  core freq cpu3   (measured window only): start=n/a end=n/a min=0MHz max=0MHz mean=0MHz spread=0.00% (n=0)
  core freq cpu1   (measured window only): start=n/a end=n/a min=0MHz max=0MHz mean=0MHz spread=0.00% (n=0)
  this row assumes cross-core TSC coherence; theta_hat measured in the ping-pong at this size = +1.2 ns (add to correct).

== cross-process unpaced stream -- QUEUE DELAY / BACKPRESSURE, NOT LATENCY (publisher unthrottled; each sample includes its wait in the ring) ==
   !! Every number in this block is a PUBLISH-TO-RECEIVE DELAY that INCLUDES
   !! time spent waiting in the ring. It is a backpressure/throughput figure.
   !! Do not quote it as an IPC latency; that is the exact mistake the retired
   !! `Topic_cross_process` metric shipped for months. Read `measured backlog`
   !! below: ~1 message means the subscriber kept up and no queue formed.
   json metric key: `Topic_cross_process_unpaced_queue_delay` (identity used by the CI gate: `Topic_cross_process_unpaced_queue_delay@64B`)
   payload Probe64 = 64 B (POD=true) | ring capacity 256 slots | topology 1P/1C | cpu3 <-> cpu1 | pid 7341 <-> pid 7357
------------------------------------------------------------------------------------------------------------------------------------------
  metric (ns)                            n       min    median       p99     p99.9     p99.99         max |  n>p99.9  n>p99.99
------------------------------------------------------------------------------------------------------------------------------------------
  publish->receive incl. queue      200000      70.1     100.1     180.3     510.8     5788.6     40239.8 |      200        19
------------------------------------------------------------------------------------------------------------------------------------------
  samples=200000  warmup=20000  transferred=200000  lost=0  invalid(TSC-skew)=0  poll cost=7.3 ns  window=0.015 s
  measured backlog: 1.30 messages deep on the median sample (ring holds 256); delivered 100.0% of transferred
  measured window: minor_faults=0 major_faults=0 ctx_switches vol=3 invol=0
  core freq cpu3   (measured window only): start=n/a end=n/a min=0MHz max=0MHz mean=0MHz spread=0.00% (n=0)
  core freq cpu1   (measured window only): start=n/a end=n/a min=0MHz max=0MHz mean=0MHz spread=0.00% (n=0)
  this row assumes cross-core TSC coherence; theta_hat measured in the ping-pong at this size = -3.9 ns (add to correct).

SUMMARY — one row per published JSON metric (nanoseconds, wall clock)
------------------------------------------------------------------------------------------------------------------------------------------
  json metric key                                what it measures                  n    median       p99     p99.9     p99.99         max
------------------------------------------------------------------------------------------------------------------------------------------
  Topic_cross_process_oneway_pingpong@64B        ONE-WAY latency             1000000     115.2     155.2     475.7     4031.0     17130.5
  Topic_cross_process_oneway_pingpong@1024B      ONE-WAY latency             1000000     185.3     255.4     560.8     5358.0     23525.0
  Topic_cross_process_oneway_paced_stream@64B    ONE-WAY latency              200000     130.2     230.3     861.3     7030.5     63364.2
  Topic_cross_process_oneway_paced_stream@1024B  ONE-WAY latency              200000     179.9     730.7   50985.4   139858.3    201719.5
  Topic_cross_process_unpaced_queue_delay@64B    QUEUE DELAY (not latency)    200000     100.1     180.3     510.8     5788.6     40239.8
------------------------------------------------------------------------------------------------------------------------------------------
  Instrumentation is INCLUDED in every figure and never subtracted.
  `Topic_cross_process_oneway_pingpong` is the true one-way figure: it assumes
  nothing about cross-core TSC synchronisation and excludes the responder's turnaround.
  `RTT halved (one-way)` in the per-scenario tables is a ROUND TRIP DIVIDED BY TWO; it is
  the row to compare against any competitor publishing a halved ping-pong.
  `Topic_cross_process_unpaced_queue_delay` is a BACKPRESSURE figure: it includes
  however long the message waited in the ring, so it scales with the achieved backlog
  (printed in its scenario block) and is NOT a latency.
  A figure without its payload size, ring depth and topology is not comparable to anything.

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!! THE SAME WARNINGS AGAIN — they apply to every number above
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!!  CPU governor is ?/?, not `performance`. RDTSC is a wall clock while the work costs a fixed number of CORE cycles, so every nanosecond below scales with a frequency the governor is free to change. iceoryx2's maintainers measured ~300 ns of difference from the governor alone.
!!  No isolcpus= on the kernel command line: the benchmark cores are shared with everything else on this box. The involuntary context-switch count in each scenario block is the mechanism behind the far tail.
!!  paced_stream (64 B) `one-way (assumes TSC coherence)`: p99.99 rests on 20 observations. Below ~100 exceedances it is a max wearing a percentile's name; +/-5% needs ~400 (--iterations 4000000).
!!  paced_stream (1024 B) `one-way (assumes TSC coherence)`: p99.99 rests on 20 observations. Below ~100 exceedances it is a max wearing a percentile's name; +/-5% needs ~400 (--iterations 4000000).
!!  unpaced_queue (64 B): measured backlog is only 1.30 messages — the subscriber kept up, so the ring never filled and `Topic_cross_process_unpaced_queue_delay` is NOT a backpressure figure on this run. It is a publish-to-receive delay with no queueing in it, and it is still not a latency (see the paced and ping-pong rows for that).
!!  unpaced_queue (64 B) `publish->receive incl. queue`: p99.99 rests on 19 observations. Below ~100 exceedances it is a max wearing a percentile's name; +/-5% needs ~400 (--iterations 4000000).
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

Gate report written to /home/runner/work/horus/horus/benchmarks/gate-reports/cross_process-2.json
  keys: Topic_cross_process_oneway_pingpong@64B, Topic_cross_process_oneway_pingpong@1024B, Topic_cross_process_oneway_paced_stream@64B, Topic_cross_process_oneway_paced_stream@1024B, Topic_cross_process_unpaced_queue_delay@64B
  NOTE: `Topic_cross_process` is retired. A baseline window that still holds it
  will report it as missing and fail the gate until the window is refreshed —
  that is intended, and is why the key was not silently reused.
  Raw samples are omitted (pass --raw-in-json, or --raw-dir for a lossless cycle dump).
  6 warning(s) apply to these numbers and are NOT in this file; use --detail-json.
╔══════════════════════════════════════════════════════════════════╗
║        HORUS Robotics Message Types Benchmark                    ║
╠══════════════════════════════════════════════════════════════════╣
║  Testing real-world robotics message latency (REP 2014)          ║
╚══════════════════════════════════════════════════════════════════╝

Platform: AMD EPYC 9V45 96-Core Processor (4 cores)
Iterations: 50000
Warmup: 5000

╔═════════════════════════════════════════════════════════════════════════════╗
║ Message Type    │ Size (bytes) │ Typical Rate │ Use Case                    ║
╠═════════════════════════════════════════════════════════════════════════════╣
║ CmdVel          │           16 │ 1000+ Hz     │ Velocity control commands   ║
║ Imu             │          296 │ 500+ Hz      │ IMU sensor fusion           ║
║ LaserScan       │         1480 │ 10-40 Hz     │ 2D lidar navigation         ║
║ JointCommand    │         1032 │ 500+ Hz      │ Manipulator control         ║
╚═════════════════════════════════════════════════════════════════════════════╝


[Topic] Running benchmarks — timing tx.send() enqueue cost only
        (cross-thread consumer drains in the background; no sample
         waits for delivery, so these are NOT end-to-end latencies)
─────────────────────────────────────────────────
  CmdVel              16 bytes │ median:      40ns │ p99:      70ns │ CV: 0.5230
  Imu                304 bytes │ median:      40ns │ p99:      80ns │ CV: 1.5836
  LaserScan         1480 bytes │ median:      70ns │ p99:     110ns │ CV: 0.2525
  JointCommand       928 bytes │ median:      70ns │ p99:     130ns │ CV: 1.0264

╔══════════════════════════════════════════════════════════════════════════════════════════╗
║           SUMMARY BY MESSAGE TYPE — send() ENQUEUE COST, not end-to-end latency           ║
╠══════════════════════════════════════════════════════════════════════════════════════════╣
║ Message      │   Size │ Median (ns) │ p99 (ns) │    CV   │ Misses │ Sustained rate       ║
╠══════════════════════════════════════════════════════════════════════════════════════════╣
║ CmdVel       │     16 │          40 │       70 │  0.5230 │      0 │      7.20 M msg/s     ║
║ Imu          │    304 │          40 │       80 │  1.5836 │      0 │      4.54 M msg/s     ║
║ LaserScan    │   1480 │          70 │      110 │  0.2525 │      0 │      2.34 M msg/s     ║
║ JointCommand │    928 │          70 │      130 │  1.0264 │      0 │      3.69 M msg/s     ║
╚══════════════════════════════════════════════════════════════════════════════════════════╝
Sustained rate is wall-clock over the measured loop, so it includes the batch
drain-waits. It is a publish rate for this producer/consumer pair, not a bus limit.

╔══════════════════════════════════════════════════════════════════╗
║              REAL-TIME SUITABILITY ANALYSIS                      ║
╠══════════════════════════════════════════════════════════════════╣
║ These gates grade the p99 of the send() ENQUEUE COST against one ║
║ whole control period. They are not end-to-end control-loop       ║
║ guarantees: a real loop also pays the receive side, the node's   ║
║ compute, and the scheduler's jitter, none of which is measured   ║
║ here. The headroom is printed so a pass by four orders of        ║
║ magnitude cannot be read as a tight result.                      ║
╠══════════════════════════════════════════════════════════════════╣
║ CmdVel    (1kHz  control): ~ PASS by 14286x — gate not discriminating (p99=70ns vs 1000000ns)
║ CmdVel    (10kHz control): ~ PASS by 1429x — gate not discriminating (p99=70ns vs 100000ns)
║ Imu       (500Hz fusion) : ~ PASS by 25000x — gate not discriminating (p99=80ns vs 2000000ns)
║ LaserScan (40Hz  lidar)  : ~ PASS by 227273x — gate not discriminating (p99=110ns vs 25000000ns)
║ JointCmd  (500Hz control): ~ PASS by 15385x — gate not discriminating (p99=130ns vs 2000000ns)
╚══════════════════════════════════════════════════════════════════╝

Results written to: /home/runner/work/horus/horus/benchmarks/gate-reports/robotics-2.json
===== repetition 3/3 =====

HORUS cross-process benchmark — ONE-WAY LATENCY (headline) + unpaced queue delay
Two OS processes on two distinct physical cores over the real Topic SHM transport.
Every handle holds exactly one role, so the `role == Both` same-thread fast path that
produces the repo's headline send-cost figure is never taken. Round-trip rows are
labelled where they are a halved round trip. See the module header for the clock
decomposition and for why the `Topic_cross_process` JSON key was retired.

MACHINE — every nanosecond below is conditional on this block
------------------------------------------------------------------------------------------------------------------------------------------
  cpu model              : AMD EPYC 9V45 96-Core Processor (2 physical / 4 logical cores)
  kernel                 : 6.17.0-1022-azure (x86_64)
  benchmark cores        : cpu3 = process A (pinger / subscriber), cpu1 = process B (responder / publisher)  [DistinctPhysical]
  governor               : cpu3=n/a cpu1=n/a  driver=n/a
  scaling range          : n/a .. n/a  (hw max n/a)  turbo_disabled=unknown
  clocksource            : tsc   tsc flags: [tsc, rdtscp, constant_tsc, tsc_reliable, nonstop_tsc]
  TSC rate (calibrated)  : 2596.151 MHz (0.3852 ns/cycle) — a WALL clock; NOT the core clock
  clock instrumentation  : 26 cycles = 10.0 ns for serialize()+rdtsc()+rdtscp(), INCLUDED in every sample and never subtracted
  transparent_hugepage   : [always] madvise never
  sched_rt_runtime/period: 950000 / 1000000 us    perf_event_paranoid=4
  isolcpus=(none)  nohz_full=(none)  smt_active=1  mlockall(MCL_CURRENT)=true
------------------------------------------------------------------------------------------------------------------------------------------

running ping-pong (one-way latency), 64 B payload ...
running ping-pong (one-way latency), 1024 B payload ...
running paced stream (one-way latency), 64 B payload ...
running paced stream (one-way latency), 1024 B payload ...
running unpaced stream (QUEUE DELAY, not latency), 64 B payload ...

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!! READ THIS BEFORE QUOTING ANY NUMBER BELOW
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!!  CPU governor is ?/?, not `performance`. RDTSC is a wall clock while the work costs a fixed number of CORE cycles, so every nanosecond below scales with a frequency the governor is free to change. iceoryx2's maintainers measured ~300 ns of difference from the governor alone.
!!  No isolcpus= on the kernel command line: the benchmark cores are shared with everything else on this box. The involuntary context-switch count in each scenario block is the mechanism behind the far tail.
!!  paced_stream (64 B) `one-way (assumes TSC coherence)`: p99.99 rests on 20 observations. Below ~100 exceedances it is a max wearing a percentile's name; +/-5% needs ~400 (--iterations 4000000).
!!  paced_stream (1024 B) `one-way (assumes TSC coherence)`: p99.99 rests on 20 observations. Below ~100 exceedances it is a max wearing a percentile's name; +/-5% needs ~400 (--iterations 4000000).
!!  unpaced_queue (64 B): measured backlog is only 1.19 messages — the subscriber kept up, so the ring never filled and `Topic_cross_process_unpaced_queue_delay` is NOT a backpressure figure on this run. It is a publish-to-receive delay with no queueing in it, and it is still not a latency (see the paced and ping-pong rows for that).
!!  unpaced_queue (64 B) `publish->receive incl. queue`: p99.99 rests on 20 observations. Below ~100 exceedances it is a max wearing a percentile's name; +/-5% needs ~400 (--iterations 4000000).
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

== cross-process ping-pong -- ONE-WAY LATENCY (2 processes, 2 topics, 1 role per handle) ==
   json metric key: `Topic_cross_process_oneway_pingpong` (identity used by the CI gate: `Topic_cross_process_oneway_pingpong@64B`)
   payload Probe64 = 64 B (POD=true) | ring capacity 256 slots | topology 1P/1C | cpu3 <-> cpu1 | pid 7381 <-> pid 7389
------------------------------------------------------------------------------------------------------------------------------------------
  metric (ns)                            n       min    median       p99     p99.9     p99.99         max |  n>p99.9  n>p99.99
------------------------------------------------------------------------------------------------------------------------------------------
  one-way, offset-free             1000000      85.1     115.2     150.2     485.7     4301.4     75596.9 |      949       100
  round trip (RTT)                 1000000     200.3     260.4     330.5    1001.5     9053.4    151213.9 |      981       100
  RTT halved (one-way)             1000000     100.1     130.2     165.2     500.7     4526.7     75606.9 |      981       100
  leg A->B (apparent)              1000000      74.7     124.8     174.9     795.8     6674.5     82196.3 |      947       100
  leg B->A (apparent)              1000000      65.5     115.6     155.6     195.7     5673.8    151089.1 |      912        99
  responder turnaround             1000000      20.0      30.0      30.0      40.1      120.2     16103.8 |      425        95
------------------------------------------------------------------------------------------------------------------------------------------
  samples=1000000  warmup=200000  transferred=2000000  lost=0  invalid(TSC-skew)=0  poll cost=7.7 ns  window=0.319 s
  measured window: minor_faults=0 major_faults=0 ctx_switches vol=63 invol=1
  core freq cpu3   (measured window only): start=n/a end=n/a min=0MHz max=0MHz mean=0MHz spread=0.00% (n=0)
  core freq cpu1   (measured window only): start=n/a end=n/a min=0MHz max=0MHz mean=0MHz spread=0.00% (n=0)
  cross-core TSC offset theta_hat = +4.6 ns (clock B - clock A). The `one-way, offset-free` row does not depend on it.

== cross-process ping-pong -- ONE-WAY LATENCY (2 processes, 2 topics, 1 role per handle) ==
   json metric key: `Topic_cross_process_oneway_pingpong` (identity used by the CI gate: `Topic_cross_process_oneway_pingpong@1024B`)
   payload Probe1k = 1024 B (POD=true) | ring capacity 256 slots | topology 1P/1C | cpu3 <-> cpu1 | pid 7381 <-> pid 7391
------------------------------------------------------------------------------------------------------------------------------------------
  metric (ns)                            n       min    median       p99     p99.9     p99.99         max |  n>p99.9  n>p99.99
------------------------------------------------------------------------------------------------------------------------------------------
  one-way, offset-free             1000000     120.2     185.3     255.4     585.9     6319.4     21116.3 |      989       100
  round trip (RTT)                 1000000     270.4     400.6     540.8    1221.8    14281.1     83073.0 |      976       100
  RTT halved (one-way)             1000000     135.2     200.3     270.4     610.9     7140.6     41536.5 |      976       100
  leg A->B (apparent)              1000000     114.8     194.9     275.0     906.0     8367.0     42066.9 |      989       100
  leg B->A (apparent)              1000000     105.5     185.7     275.8     345.9     6234.6     35157.4 |      994        99
  responder turnaround             1000000      20.0      30.0      40.1      40.1      130.2     82642.3 |      634        87
------------------------------------------------------------------------------------------------------------------------------------------
  samples=1000000  warmup=200000  transferred=2000000  lost=0  invalid(TSC-skew)=0  poll cost=29.7 ns  window=0.488 s
  measured window: minor_faults=0 major_faults=0 ctx_switches vol=97 invol=6
  core freq cpu3   (measured window only): start=n/a end=n/a min=0MHz max=0MHz mean=0MHz spread=0.00% (n=0)
  core freq cpu1   (measured window only): start=n/a end=n/a min=0MHz max=0MHz mean=0MHz spread=0.00% (n=0)
  cross-core TSC offset theta_hat = +4.6 ns (clock B - clock A). The `one-way, offset-free` row does not depend on it.

== cross-process paced stream -- ONE-WAY LATENCY (10000 ns publish period, publisher -> subscriber, direct) ==
   json metric key: `Topic_cross_process_oneway_paced_stream` (identity used by the CI gate: `Topic_cross_process_oneway_paced_stream@64B`)
   payload Probe64 = 64 B (POD=true) | ring capacity 256 slots | topology 1P/1C | cpu3 <-> cpu1 | pid 7381 <-> pid 7393
------------------------------------------------------------------------------------------------------------------------------------------
  metric (ns)                            n       min    median       p99     p99.9     p99.99         max |  n>p99.9  n>p99.99
------------------------------------------------------------------------------------------------------------------------------------------
  one-way (assumes TSC coherence)    200000     256.1     516.5     726.8    6114.8    37211.2     47405.9 |      200        20
------------------------------------------------------------------------------------------------------------------------------------------
  samples=200000  warmup=20000  transferred=200000  lost=0  invalid(TSC-skew)=0  poll cost=5.4 ns  window=2.000 s
  measured backlog: 0.05 messages deep on the median sample (ring holds 256); delivered 100.0% of transferred
  measured window: minor_faults=0 major_faults=0 ctx_switches vol=395 invol=13
  core freq cpu3   (measured window only): start=n/a end=n/a min=0MHz max=0MHz mean=0MHz spread=0.00% (n=0)
  core freq cpu1   (measured window only): start=n/a end=n/a min=0MHz max=0MHz mean=0MHz spread=0.00% (n=0)
  this row assumes cross-core TSC coherence; theta_hat measured in the ping-pong at this size = +4.6 ns (add to correct).

== cross-process paced stream -- ONE-WAY LATENCY (10000 ns publish period, publisher -> subscriber, direct) ==
   json metric key: `Topic_cross_process_oneway_paced_stream` (identity used by the CI gate: `Topic_cross_process_oneway_paced_stream@1024B`)
   payload Probe1k = 1024 B (POD=true) | ring capacity 256 slots | topology 1P/1C | cpu3 <-> cpu1 | pid 7381 <-> pid 7395
------------------------------------------------------------------------------------------------------------------------------------------
  metric (ns)                            n       min    median       p99     p99.9     p99.99         max |  n>p99.9  n>p99.99
------------------------------------------------------------------------------------------------------------------------------------------
  one-way (assumes TSC coherence)    200000     106.3     166.4     526.9   12294.4   167954.8    181835.3 |      200        20
------------------------------------------------------------------------------------------------------------------------------------------
  samples=200000  warmup=20000  transferred=200000  lost=0  invalid(TSC-skew)=0  poll cost=11.6 ns  window=2.000 s
  measured backlog: 0.02 messages deep on the median sample (ring holds 256); delivered 100.0% of transferred
  measured window: minor_faults=0 major_faults=0 ctx_switches vol=395 invol=17
  core freq cpu3   (measured window only): start=n/a end=n/a min=0MHz max=0MHz mean=0MHz spread=0.00% (n=0)
  core freq cpu1   (measured window only): start=n/a end=n/a min=0MHz max=0MHz mean=0MHz spread=0.00% (n=0)
  this row assumes cross-core TSC coherence; theta_hat measured in the ping-pong at this size = +4.6 ns (add to correct).

== cross-process unpaced stream -- QUEUE DELAY / BACKPRESSURE, NOT LATENCY (publisher unthrottled; each sample includes its wait in the ring) ==
   !! Every number in this block is a PUBLISH-TO-RECEIVE DELAY that INCLUDES
   !! time spent waiting in the ring. It is a backpressure/throughput figure.
   !! Do not quote it as an IPC latency; that is the exact mistake the retired
   !! `Topic_cross_process` metric shipped for months. Read `measured backlog`
   !! below: ~1 message means the subscriber kept up and no queue formed.
   json metric key: `Topic_cross_process_unpaced_queue_delay` (identity used by the CI gate: `Topic_cross_process_unpaced_queue_delay@64B`)
   payload Probe64 = 64 B (POD=true) | ring capacity 256 slots | topology 1P/1C | cpu3 <-> cpu1 | pid 7381 <-> pid 7398
------------------------------------------------------------------------------------------------------------------------------------------
  metric (ns)                            n       min    median       p99     p99.9     p99.99         max |  n>p99.9  n>p99.99
------------------------------------------------------------------------------------------------------------------------------------------
  publish->receive incl. queue      200000      66.3      96.3     186.4     797.3     5975.0     17942.7 |      198        20
------------------------------------------------------------------------------------------------------------------------------------------
  samples=200000  warmup=20000  transferred=200000  lost=0  invalid(TSC-skew)=0  poll cost=7.7 ns  window=0.016 s
  measured backlog: 1.19 messages deep on the median sample (ring holds 256); delivered 100.0% of transferred
  measured window: minor_faults=0 major_faults=0 ctx_switches vol=3 invol=0
  core freq cpu3   (measured window only): start=n/a end=n/a min=0MHz max=0MHz mean=0MHz spread=0.00% (n=0)
  core freq cpu1   (measured window only): start=n/a end=n/a min=0MHz max=0MHz mean=0MHz spread=0.00% (n=0)
  this row assumes cross-core TSC coherence; theta_hat measured in the ping-pong at this size = +4.6 ns (add to correct).

SUMMARY — one row per published JSON metric (nanoseconds, wall clock)
------------------------------------------------------------------------------------------------------------------------------------------
  json metric key                                what it measures                  n    median       p99     p99.9     p99.99         max
------------------------------------------------------------------------------------------------------------------------------------------
  Topic_cross_process_oneway_pingpong@64B        ONE-WAY latency             1000000     115.2     150.2     485.7     4301.4     75596.9
  Topic_cross_process_oneway_pingpong@1024B      ONE-WAY latency             1000000     185.3     255.4     585.9     6319.4     21116.3
  Topic_cross_process_oneway_paced_stream@64B    ONE-WAY latency              200000     516.5     726.8    6114.8    37211.2     47405.9
  Topic_cross_process_oneway_paced_stream@1024B  ONE-WAY latency              200000     166.4     526.9   12294.4   167954.8    181835.3
  Topic_cross_process_unpaced_queue_delay@64B    QUEUE DELAY (not latency)    200000      96.3     186.4     797.3     5975.0     17942.7
------------------------------------------------------------------------------------------------------------------------------------------
  Instrumentation is INCLUDED in every figure and never subtracted.
  `Topic_cross_process_oneway_pingpong` is the true one-way figure: it assumes
  nothing about cross-core TSC synchronisation and excludes the responder's turnaround.
  `RTT halved (one-way)` in the per-scenario tables is a ROUND TRIP DIVIDED BY TWO; it is
  the row to compare against any competitor publishing a halved ping-pong.
  `Topic_cross_process_unpaced_queue_delay` is a BACKPRESSURE figure: it includes
  however long the message waited in the ring, so it scales with the achieved backlog
  (printed in its scenario block) and is NOT a latency.
  A figure without its payload size, ring depth and topology is not comparable to anything.

!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!! THE SAME WARNINGS AGAIN — they apply to every number above
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
!!  CPU governor is ?/?, not `performance`. RDTSC is a wall clock while the work costs a fixed number of CORE cycles, so every nanosecond below scales with a frequency the governor is free to change. iceoryx2's maintainers measured ~300 ns of difference from the governor alone.
!!  No isolcpus= on the kernel command line: the benchmark cores are shared with everything else on this box. The involuntary context-switch count in each scenario block is the mechanism behind the far tail.
!!  paced_stream (64 B) `one-way (assumes TSC coherence)`: p99.99 rests on 20 observations. Below ~100 exceedances it is a max wearing a percentile's name; +/-5% needs ~400 (--iterations 4000000).
!!  paced_stream (1024 B) `one-way (assumes TSC coherence)`: p99.99 rests on 20 observations. Below ~100 exceedances it is a max wearing a percentile's name; +/-5% needs ~400 (--iterations 4000000).
!!  unpaced_queue (64 B): measured backlog is only 1.19 messages — the subscriber kept up, so the ring never filled and `Topic_cross_process_unpaced_queue_delay` is NOT a backpressure figure on this run. It is a publish-to-receive delay with no queueing in it, and it is still not a latency (see the paced and ping-pong rows for that).
!!  unpaced_queue (64 B) `publish->receive incl. queue`: p99.99 rests on 20 observations. Below ~100 exceedances it is a max wearing a percentile's name; +/-5% needs ~400 (--iterations 4000000).
!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!

Gate report written to /home/runner/work/horus/horus/benchmarks/gate-reports/cross_process-3.json
  keys: Topic_cross_process_oneway_pingpong@64B, Topic_cross_process_oneway_pingpong@1024B, Topic_cross_process_oneway_paced_stream@64B, Topic_cross_process_oneway_paced_stream@1024B, Topic_cross_process_unpaced_queue_delay@64B
  NOTE: `Topic_cross_process` is retired. A baseline window that still holds it
  will report it as missing and fail the gate until the window is refreshed —
  that is intended, and is why the key was not silently reused.
  Raw samples are omitted (pass --raw-in-json, or --raw-dir for a lossless cycle dump).
  6 warning(s) apply to these numbers and are NOT in this file; use --detail-json.
╔══════════════════════════════════════════════════════════════════╗
║        HORUS Robotics Message Types Benchmark                    ║
╠══════════════════════════════════════════════════════════════════╣
║  Testing real-world robotics message latency (REP 2014)          ║
╚══════════════════════════════════════════════════════════════════╝

Platform: AMD EPYC 9V45 96-Core Processor (4 cores)
Iterations: 50000
Warmup: 5000

╔═════════════════════════════════════════════════════════════════════════════╗
║ Message Type    │ Size (bytes) │ Typical Rate │ Use Case                    ║
╠═════════════════════════════════════════════════════════════════════════════╣
║ CmdVel          │           16 │ 1000+ Hz     │ Velocity control commands   ║
║ Imu             │          296 │ 500+ Hz      │ IMU sensor fusion           ║
║ LaserScan       │         1480 │ 10-40 Hz     │ 2D lidar navigation         ║
║ JointCommand    │         1032 │ 500+ Hz      │ Manipulator control         ║
╚═════════════════════════════════════════════════════════════════════════════╝


[Topic] Running benchmarks — timing tx.send() enqueue cost only
        (cross-thread consumer drains in the background; no sample
         waits for delivery, so these are NOT end-to-end latencies)
─────────────────────────────────────────────────
  CmdVel              16 bytes │ median:      40ns │ p99:      70ns │ CV: 1.2826
  Imu                304 bytes │ median:      50ns │ p99:      80ns │ CV: 1.6053
  LaserScan         1480 bytes │ median:      60ns │ p99:     100ns │ CV: 1.2939
  JointCommand       928 bytes │ median:      70ns │ p99:     120ns │ CV: 1.1344

╔══════════════════════════════════════════════════════════════════════════════════════════╗
║           SUMMARY BY MESSAGE TYPE — send() ENQUEUE COST, not end-to-end latency           ║
╠══════════════════════════════════════════════════════════════════════════════════════════╣
║ Message      │   Size │ Median (ns) │ p99 (ns) │    CV   │ Misses │ Sustained rate       ║
╠══════════════════════════════════════════════════════════════════════════════════════════╣
║ CmdVel       │     16 │          40 │       70 │  1.2826 │      0 │      7.20 M msg/s     ║
║ Imu          │    304 │          50 │       80 │  1.6053 │      0 │      4.50 M msg/s     ║
║ LaserScan    │   1480 │          60 │      100 │  1.2939 │      0 │      2.40 M msg/s     ║
║ JointCommand │    928 │          70 │      120 │  1.1344 │      0 │      3.61 M msg/s     ║
╚══════════════════════════════════════════════════════════════════════════════════════════╝
Sustained rate is wall-clock over the measured loop, so it includes the batch
drain-waits. It is a publish rate for this producer/consumer pair, not a bus limit.

╔══════════════════════════════════════════════════════════════════╗
║              REAL-TIME SUITABILITY ANALYSIS                      ║
╠══════════════════════════════════════════════════════════════════╣
║ These gates grade the p99 of the send() ENQUEUE COST against one ║
║ whole control period. They are not end-to-end control-loop       ║
║ guarantees: a real loop also pays the receive side, the node's   ║
║ compute, and the scheduler's jitter, none of which is measured   ║
║ here. The headroom is printed so a pass by four orders of        ║
║ magnitude cannot be read as a tight result.                      ║
╠══════════════════════════════════════════════════════════════════╣
║ CmdVel    (1kHz  control): ~ PASS by 14286x — gate not discriminating (p99=70ns vs 1000000ns)
║ CmdVel    (10kHz control): ~ PASS by 1429x — gate not discriminating (p99=70ns vs 100000ns)
║ Imu       (500Hz fusion) : ~ PASS by 25000x — gate not discriminating (p99=80ns vs 2000000ns)
║ LaserScan (40Hz  lidar)  : ~ PASS by 250000x — gate not discriminating (p99=100ns vs 25000000ns)
║ JointCmd  (500Hz control): ~ PASS by 16667x — gate not discriminating (p99=120ns vs 2000000ns)
╚══════════════════════════════════════════════════════════════════╝

Results written to: /home/runner/work/horus/horus/benchmarks/gate-reports/robotics-3.json


Gated against a rolling window of recent main runs. Each metric's band is max(relative, center + k·sigma_observed, absolute floor), so it is derived from the noise the baseline itself measured — where a band is printed wider than the change you care about, this run did not detect a regression and could not have. Thresholds and their reasoning: RegressionPolicy in benchmarks/src/output.rs.
Commit: 974c3d1

@github-actions
github-actions Bot merged commit 79373e7 into main Sep 14, 2026
74 checks passed
@dependabot
dependabot Bot deleted the dependabot/cargo/zstd-0.14.0 branch September 14, 2026 08:16
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

dependencies Pull requests that update a dependency file rust Pull requests that update rust code

Projects

None yet

Development

Successfully merging this pull request may close these issues.

0 participants