Skip to content

Develop test result script#97

Merged
huitema merged 5 commits into
mainfrom
develop-test-result-script
Jun 18, 2026
Merged

Develop test result script#97
huitema merged 5 commits into
mainfrom
develop-test-result-script

Conversation

@huitema

@huitema huitema commented Jun 18, 2026

Copy link
Copy Markdown
Contributor

Develop a simple script that presents the results of many tests.

The tests are grouped in categories, as coded in the script. This will probably evolve over time, as we add more tests.

In each category, we present a set of tables tracking either the execution time for "batch" tests, or the measure latency for "media" tests. In fact, we will present two tables:

  • for batch tests, a table displaying the average time and another displaying the 90th percentile time.
  • for media test, a table displaying the average frame latency and another displaying the 90th percentile of the max latency as found in 100 tests.

Each line will present three columns: the time or latency measured for C4, then for the same test unsing BBR and Cubic.

This presentation answers a critic that we heard during IETF 125:

  • we should present the results of the 100 tests
  • we should compare the C4 results against the results of BBR and Cubic.

@huitema huitema requested a review from suhasHere June 18, 2026 02:19
@huitema

huitema commented Jun 18, 2026

Copy link
Copy Markdown
Contributor Author

Here are the first results, as measured with the current version of C4:

Statistics

Here is a collection of statistics on all test cases.

simple

Here the statistics for the simple test cases.

average time for simple tests

average time for simple tests c4 bbr cubic
alone 4647943 4688512 4476998
low_and_up 7926859 7506316 8049855
drop_and_back 7892651 7627307 7628463
blackhole 5628255 5811376 5695208
satellite 7660773 7472446 6704244

top 90% time for simple tests

top 90% time for simple tests c4 bbr cubic
alone 4806735 4700741 4521935
low_and_up 7929953 7511559 8071886
drop_and_back 7896874 7632814 7632383
blackhole 5629067 5815444 5699325
satellite 7661570 7472352 6704246

wifi

Here the statistics for the wifi test cases.

average time for wifi tests

average time for wifi tests c4 bbr cubic
wifi_bad 4095173 5316761 4060841
wifi_fade 5154116 5406372 5367521
wifi_suspension 4562814 4615882 4600515

top 90% time for wifi tests

top 90% time for wifi tests c4 bbr cubic
wifi_bad 4663533 7321342 4319893
wifi_fade 5479860 5572588 5581560
wifi_suspension 4563522 4616912 4600827

media

Here the statistics for the media test cases.

average av_latency for media tests

average av_latency for media tests c4 bbr cubic
media 33511 33427 33512
media10 45214 45008 47760
media_short_long 101035 133963 100771
media_wb 79290 85020 78309
media_wf 83584 85481 81348
media_ws 22844 21645 22492

top 90% max_latency for media tests

top 90% max_latency for media tests c4 bbr cubic
media 43453 43453 43453
media10 71128 71128 92163
media_short_long 117983 334491 110858
media_wb 270360 315884 265915
media_wf 289548 365087 304576
media_ws 197821 195521 197821

compete

Here the statistics for the compete test cases.

average time for compete tests

average time for compete tests c4 bbr cubic
vs_bbr 2945713 4502793 2852396
vs_c4 4446590 6314435 6198056
vs_cubic 3648071 3637191 5266135
wifi_bad_bbr 7487152 7909440 8354502
wifi_bad_c4 9903181 10548742 9243776
wifi_bad_cubic 8550470 8828130 9734756

top 90% time for compete tests

top 90% time for compete tests c4 bbr cubic
vs_bbr 2963475 4569995 2873067
vs_c4 4955209 6748970 6606557
vs_cubic 3971537 3966392 5578272
wifi_bad_bbr 11906585 12292138 13517663
wifi_bad_c4 12316529 12599665 12291643
wifi_bad_cubic 11211242 12016981 13410701

others

Here the statistics for the other test cases.

average time c4 bbr cubic
after_c4 5963997
alone_200 1174457
before_c4 3100256
ecn 4490759
ecn_bbr 13280026
ecn_c4 11591693
ecn_cubic 8615196
media_600fr 20057739
short_long 17442217
vs_bbr_lg 17203354
vs_bbr_lg2 17484790
vs_c4_lg 21166353
vs_c4_lg2 21164105
vs_cubic_lg 17638810
vs_cubic_lg2 17381870

@huitema huitema merged commit b408de6 into main Jun 18, 2026
1 check passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants