test: Make criterion calculate transfer benchmark throughputs

[larseggert]

Like:

transfer/Run multiple transfers with varying seeds
                        time:   [213.61 ms 224.95 ms 235.29 ms]
                        thrpt:  [17.000 MiB/s 17.782 MiB/s 18.726 MiB/s]
                 change:
                        time:   [-17.204% -6.9690% +1.8324%] (p = 0.21 > 0.05)
                        thrpt:  [-1.7994% +7.4910% +20.778%]
                        No change in performance detected.
Found 5 outliers among 100 measurements (5.00%)
  5 (5.00%) low mild

transfer/Run multiple transfers with the same seed
                        time:   [175.41 ms 188.40 ms 201.32 ms]
                        thrpt:  [19.869 MiB/s 21.231 MiB/s 22.803 MiB/s]
                 change:
                        time:   [-5.9393% +5.5417% +18.399%] (p = 0.35 > 0.05)
                        thrpt:  [-15.540% -5.2507% +6.3144%]
                        No change in performance detected.
Found 14 outliers among 100 measurements (14.00%)
  8 (8.00%) low mild
  5 (5.00%) high mild
  1 (1.00%) high severe

[github-actions]

QUIC Interop Runner

Saving logs to logs.
Run took 0:03:32.727096
Server: neqo-latest. Client: neqo-latest. Running test case: handshake
Server: quic-go. Client: neqo-latest. Running test case: handshake
Server: ngtcp2. Client: neqo-latest. Running test case: handshake
Server: neqo. Client: neqo-latest. Running test case: handshake
Server: msquic. Client: neqo-latest. Running test case: handshake
Server: neqo-latest. Client: quic-go. Running test case: handshake
Server: neqo-latest. Client: ngtcp2. Running test case: handshake
Server: neqo-latest. Client: neqo. Running test case: handshake
Server: neqo-latest. Client: msquic. Running test case: handshake
+-------------+-------------+---------+--------+------+--------+
|             | neqo-latest | quic-go | ngtcp2 | neqo | msquic |
+-------------+-------------+---------+--------+------+--------+
| neqo-latest |     ✓(H)    |   ✓(H)  |  ✓(H)  | ✓(H) |  ✓(H)  |
|             |     ?()     |   ?()   |  ?()   | ?()  |  ?()   |
|             |     ✕()     |   ✕()   |  ✕()   | ✕()  |  ✕()   |
+-------------+-------------+---------+--------+------+--------+
|   quic-go   |     ✓(H)    |         |        |      |        |
|             |     ?()     |         |        |      |        |
|             |     ✕()     |         |        |      |        |
+-------------+-------------+---------+--------+------+--------+
|    ngtcp2   |     ✓(H)    |         |        |      |        |
|             |     ?()     |         |        |      |        |
|             |     ✕()     |         |        |      |        |
+-------------+-------------+---------+--------+------+--------+
|     neqo    |     ✓(H)    |         |        |      |        |
|             |     ?()     |         |        |      |        |
|             |     ✕()     |         |        |      |        |
+-------------+-------------+---------+--------+------+--------+
|    msquic   |     ✓(H)    |         |        |      |        |
|             |     ?()     |         |        |      |        |
|             |     ✕()     |         |        |      |        |
+-------------+-------------+---------+--------+------+--------+

⬇️ Download logs

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 89.48%. Comparing base (f408321) to head (9f6d5e5).

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #1738   +/-   ##
=======================================
  Coverage   89.48%   89.48%           
=======================================
  Files         126      126           
  Lines       38866    38866           
=======================================
  Hits        34780    34780           
  Misses       4086     4086           

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[martinthomson]

LGTM.

I would observe that this isn't going to provide any indication of throughput in the congestion control sense, since these tests do not operate in real time. We're talking about raw packet production and consumption power here. That might be worth noting in the code.

(Approved in any case.)

[github-actions]

Benchmark results

Performance differences relative to 2750423.

• coalesce_acked_from_zero 1+1 entries
  time: [255.39 ns 255.87 ns 256.38 ns]
  change: [-0.2847% +0.0787% +0.4584%] (p = 0.68 > 0.05)
  No change in performance detected.

• coalesce_acked_from_zero 3+1 entries
  time: [291.71 ns 292.45 ns 293.23 ns]
  change: [-0.2100% +0.0943% +0.4136%] (p = 0.55 > 0.05)
  No change in performance detected.

• coalesce_acked_from_zero 10+1 entries
  time: [290.93 ns 292.00 ns 293.24 ns]
  change: [-0.7008% +0.1914% +1.2216%] (p = 0.76 > 0.05)
  No change in performance detected.

• coalesce_acked_from_zero 1000+1 entries
  time: [268.01 ns 268.19 ns 268.39 ns]
  change: [-0.5810% +0.0053% +0.6360%] (p = 0.99 > 0.05)
  No change in performance detected.

• RxStreamOrderer::inbound_frame()
  time: [126.66 ms 126.83 ms 127.07 ms]
  change: [-0.5367% -0.3090% -0.0598%] (p = 0.00 < 0.05)
  Change within noise threshold.

• transfer/Run multiple transfers with varying seeds
  time: [189.58 ms 190.17 ms 190.74 ms]
  thrpt: [20.971 MiB/s 21.034 MiB/s 21.099 MiB/s]

• transfer/Run multiple transfers with the same seed
  time: [190.84 ms 191.24 ms 191.67 ms]
  thrpt: [20.869 MiB/s 20.916 MiB/s 20.960 MiB/s]

⬇️ Download logs