86c1c0b9354bc7ae4ca8bb3557459b1b9540e771 |
|
14-Oct-2016 |
Evelina Dumitrescu <evelinad@google.com> |
user_activity: Fixed a regression in ComputeScoreForBenchmarkSet. The score fraction for a group was computed wrong. The fraction should be computed as the score of the group over the score of that group in the ideal case (each CWP function matching that group should have the 1.0 value). BUG=None TEST=None Change-Id: I294b42f0c8112ab9bd6f5d983876b6d779aad5e4 Reviewed-on: https://chrome-internal-review.googlesource.com/297675 Tested-by: Evelina Dumitrescu <evelinad@google.com> Reviewed-by: Evelina Dumitrescu <evelinad@google.com> Reviewed-by: Luis Lozano <llozano@chromium.org> Reviewed-on: https://chromium-review.googlesource.com/435921 Commit-Ready: Luis Lozano <llozano@chromium.org> Tested-by: Luis Lozano <llozano@chromium.org>
/external/toolchain-utils/user_activity_benchmarks/benchmark_metrics.py
|
edc6eb87218afbde3c3df6e0faa99958f709c825 |
|
28-Sep-2016 |
Evelina Dumitrescu <evelinad@google.com> |
user_activity: Added script to select the optimal benchmark set. It generates all possible combinations of benchmark sets of a given size and it computes for every set a metric. Three different metrics have been used: function count, distance variation and score. TEST=None BUG=None Change-Id: I4e4af9f46a3056cfbcdc9d89b2c1333650b41ccc Reviewed-on: https://chrome-internal-review.googlesource.com/292435 Reviewed-by: George Burgess <gbiv@google.com> Reviewed-by: Ting-Yuan Huang <laszio@google.com> Reviewed-by: Evelina Dumitrescu <evelinad@google.com> Tested-by: Evelina Dumitrescu <evelinad@google.com> Reviewed-on: https://chromium-review.googlesource.com/435914 Commit-Ready: Luis Lozano <llozano@chromium.org> Tested-by: Luis Lozano <llozano@chromium.org> Reviewed-by: Luis Lozano <llozano@chromium.org>
/external/toolchain-utils/user_activity_benchmarks/benchmark_metrics.py
|
77abf01f2cff8bb3a5564f554d9218c085b83d65 |
|
13-Sep-2016 |
Evelina Dumitrescu <evelinad@google.com> |
user_activity: Added implementation for the metrics. For every function, we compute a distance based on the sum of inclusive value fraction differences of pairs of common child functions from a sample, and reference function sets. For every function, we assign a score based on the distance, inclusive value fraction from the sample, and reference function sets. For every Chrome OS component, we compute a set of metrics consisting in the number of functions, the average, and cumulative distance and the total score of the functions matching the group. BUG=None TEST=benchmark_metrics_unittest passes Change-Id: I7810ec86bc4fa488b49ee18de3b3bda2ef95ba09 Reviewed-on: https://chrome-internal-review.googlesource.com/286161 Reviewed-by: Luis Lozano <llozano@chromium.org> Tested-by: Evelina Dumitrescu <evelinad@google.com> Reviewed-on: https://chromium-review.googlesource.com/435910 Commit-Ready: Luis Lozano <llozano@chromium.org> Tested-by: Luis Lozano <llozano@chromium.org>
/external/toolchain-utils/user_activity_benchmarks/benchmark_metrics.py
|