----------------------------------------------------------- This is an automatically generated e-mail. To reply, visit: https://reviews.apache.org/r/69885/ -----------------------------------------------------------
(Updated Feb. 5, 2019, 9:27 a.m.) Review request for mesos, Benjamin Mahler and Klaus Ma. Bugs: MESOS-8835 https://issues.apache.org/jira/browse/MESOS-8835 Repository: mesos Description ------- This patch moves computation of some resource benchmark test parameters from test instantiation time to test execution time. This prevents us from having to perform the expensive calculation of test parameters even when not executing the benchmark. As a result the startup time of the Mesos tests binary is improved, while the total wall time required to run these particular benchmarks is degraded accordingly. Diffs (updated) ----- src/tests/resources_tests.cpp f762d17376cc5c29e8556ef5aa2b981e8fe19985 Diff: https://reviews.apache.org/r/69885/diff/3/ Changes: https://reviews.apache.org/r/69885/diff/2-3/ Testing ------- Benchmarked `./src/mesos-tests --gtest_list_tests` with clang-9.0.0, lld-2.27. Overall execution time is improved, especially for not optimized builds. ``` Benchmark #1: Before patch, debug Time (mean ± ?): 2.706 s ± 0.018 s [User: 2.472 s, System: 0.168 s] Range (min … max): 2.690 s … 2.732 s 10 runs Benchmark #2: After patch, debug Time (mean ± ?): 683.7 ms ± 18.1 ms [User: 474.2 ms, System: 152.9 ms] Range (min … max): 673.4 ms … 734.2 ms 10 runs ``` ``` Benchmark #3: Before patch, optimized Time (mean ± ?): 783.0 ms ± 15.0 ms [User: 537.4 ms, System: 144.9 ms] Range (min … max): 772.2 ms … 815.5 ms 10 runs Benchmark #4: After patch, optimized Time (mean ± ?): 572.5 ms ± 6.7 ms [User: 343.3 ms, System: 138.4 ms] Range (min … max): 562.2 ms … 588.7 ms 10 runs ``` Remaining time is due to the long list of filters `mesos-tests` uses. Thanks, Benjamin Bannier