Improve the sharding algorithm & reshard linux_perf, mac_1013_high_end, and mac_1012_low_end
This CL improve the sharding algorithm by adjusting the policy of picking a story at the boundary a benchmark shard. Before this, we will keep adding story to a benchmark shard until the total time assigned is greater than expected total time (assuming assignment is perfect). This CL changes it to keep adding story until the absolute value of the difference between total time assigned and expeted total time is minimal. The effectiveness of this change can be shown by looking at the predict stats ("extra_info" field) in the 3 shards: linux_perf_shard_map, mac_1012_low_end_26_shard_map, mac_1013_high_end_26_shard_map. Change-Id: I3ace373fb1393658cc21aca1a61a66d865a8c059 NOTRY=true # tests covered by PRESUBMIT Change-Id: I3ace373fb1393658cc21aca1a61a66d865a8c059 Reviewed-on: https://chromium-review.googlesource.com/1108374 Commit-Queue: Ned Nguyen <nednguyen@google.com> Reviewed-by:Emily Hanley <eyaich@chromium.org> Reviewed-by:
Ashley Enstad <ashleymarie@chromium.org> Cr-Commit-Position: refs/heads/master@{#569241}
Showing
Please register or sign in to comment