fix(benches/transfer): remove throughput #2007
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
mxinden
requested review from
KershawChang,
martinthomson and
larseggert
as code owners
The `neqo-transport/benches/transfer.rs` benchmarks use the `test-fixture/src/sim` simulator. The simulator can travel in time, i.e. it simulates time. The _wall-clock time_ of a single benchmark run is not the amount of time it took to transfer `TRANSFER_AMOUNT`. The _simulated time_ is the amount of time it took to transfer `TRANSFER_AMOUNT`. `criterion` will use the _wall-clock time_, not the _simulated time_ to calculate the throughput based on `TRANSFER_AMOUNT`. The resulting throughput number is not meaningful. This commit removes the call to `group.throughput`, thus removing the misleading `criterion` throughput reporting.
mxinden
force-pushed
the
transfer-no-throughput
branch
from
c886e58 to
e09e7a5
Compare
Codecov ReportAll modified and coverable lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## main #2007 +/- ##
=======================================
Coverage 95.00% 95.00%
=======================================
Files 112 112
Lines 36364 36364
=======================================
+ Hits 34546 34549 +3
+ Misses 1818 1815 -3 ☔ View full report in Codecov by Sentry. |
Failed Interop TestsQUIC Interop Runner, client vs. server neqo-latest as client
neqo-latest as server
All resultsSucceeded Interop TestsQUIC Interop Runner, client vs. server neqo-latest as client
neqo-latest as server
Unsupported Interop TestsQUIC Interop Runner, client vs. server neqo-latest as client
neqo-latest as server
|
Benchmark resultsPerformance differences relative to 7a39675. coalesce_acked_from_zero 1+1 entries: 💚 Performance has improved. time: [189.97 ns 190.47 ns 191.00 ns]
change: [-2.9299% -2.4592% -2.0057%] (p = 0.00 < 0.05)
Performance has
Found 2 outliers among 100 measurements (2.00%)
1 (1.00%) high mild
1 (1.00%) high severecoalesce_acked_from_zero 3+1 entries: 💚 Performance has improved. time: [230.52 ns 230.97 ns 231.47 ns]
change: [-2.9535% -2.4520% -1.7883%] (p = 0.00 < 0.05)
Performance has
Found 11 outliers among 100 measurements (11.00%)
1 (1.00%) low mild
1 (1.00%) high mild
9 (9.00%) high severecoalesce_acked_from_zero 10+1 entries: Change within noise threshold. time: [230.20 ns 230.75 ns 231.47 ns]
change: [-2.7996% -1.8925% -0.5057%] (p = 0.00 < 0.05)
Change within
Found 5 outliers among 100 measurements (5.00%)
5 (5.00%) high severecoalesce_acked_from_zero 1000+1 entries: Change within noise threshold. time: [217.60 ns 217.82 ns 218.07 ns]
change: [-1.3794% -0.6760% +0.0194%] (p = 0.05 < 0.05)
Change within
Found 13 outliers among 100 measurements (13.00%)
7 (7.00%) high mild
6 (6.00%) high severeRxStreamOrderer::inbound_frame(): No change in performance detected. time: [118.65 ms 118.83 ms 119.09 ms]
change: [-0.0166% +0.1665% +0.4268%] (p = 0.12 > 0.05)
No change
Found 1 outliers among 100 measurements (1.00%)
1 (1.00%) high severetransfer/pacing-false/varying-seeds: No change in performance detected. time: [39.139 ms 41.438 ms 43.731 ms]
change: [-7.4449% -1.0747% +5.4487%] (p = 0.76 > 0.05)
No change
Found 1 outliers among 100 measurements (1.00%)
1 (1.00%) high mildtransfer/pacing-true/varying-seeds: No change in performance detected. time: [50.893 ms 53.748 ms 56.634 ms]
change: [-13.440% -6.3842% +1.1913%] (p = 0.10 > 0.05)
No change
Found 4 outliers among 100 measurements (4.00%)
2 (2.00%) low mild
2 (2.00%) high mildtransfer/pacing-false/same-seed: No change in performance detected. time: [47.983 ms 49.488 ms 50.937 ms]
change: [-3.8273% +0.2587% +4.7279%] (p = 0.90 > 0.05)
No change
Found 4 outliers among 100 measurements (4.00%)
4 (4.00%) low mildtransfer/pacing-true/same-seed: No change in performance detected. time: [68.613 ms 75.000 ms 81.355 ms]
change: [-6.3302% +5.4390% +19.507%] (p = 0.40 > 0.05)
No change1-conn/1-100mb-resp (aka. Download)/client: No change in performance detected. time: [290.42 ms 298.80 ms 306.74 ms]
thrpt: [326.01 MiB/s 334.67 MiB/s 344.33 MiB/s]
change:
time: [-3.5019% +0.3637% +4.4492%] (p = 0.87 > 0.05)
thrpt: [-4.2597% -0.3624% +3.6290%]
No change1-conn/10_000-parallel-1b-resp (aka. RPS)/client: 💚 Performance has improved. time: [399.13 ms 402.39 ms 405.71 ms]
thrpt: [24.648 Kelem/s 24.852 Kelem/s 25.055 Kelem/s]
change:
time: [-4.2455% -3.0686% -1.9164%] (p = 0.00 < 0.05)
thrpt: [+1.9539% +3.1657% +4.4338%]
Performance has1-conn/1-1b-resp (aka. HPS)/client: No change in performance detected. time: [68.372 ms 68.771 ms 69.221 ms]
thrpt: [14.447 elem/s 14.541 elem/s 14.626 elem/s]
change:
time: [-0.1329% +0.6064% +1.4190%] (p = 0.14 > 0.05)
thrpt: [-1.3991% -0.6028% +0.1331%]
No change
Found 11 outliers among 100 measurements (11.00%)
1 (1.00%) low mild
10 (10.00%) high severeClient/server transfer resultsTransfer of 33554432 bytes over loopback.
|
larseggert
approved these changes
Jul 26, 2024
larseggert
approved these changes
Jul 30, 2024
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
The
neqo-transport/benches/transfer.rsbenchmarks use thetest-fixture/src/simsimulator. The simulator can travel in time, i.e. it simulates time.The wall-clock time of a single benchmark run is not the amount of time it took to transfer
TRANSFER_AMOUNT. The simulated time is the amount of time it took to transferTRANSFER_AMOUNT.criterionwill use the wall-clock time, not the simulated time to calculate the throughput based onTRANSFER_AMOUNT. The resulting throughput number is not meaningful.This commit removes the call to
group.throughput, thus removing the misleadingcriterionthroughput reporting.Related to #1998.