Small improvements to DBGet microbenchmark (#11498)
Summary: Follow a couple best practices: - Allowed Google benchmark to decide number of iterations. Previously we hardcoded a value, which circumvented benchmark's heuristic for iterating until the result is stable. - Made each iteration do similar work. Previously, an iteration could do different work depending if the key was found in the first, second, third, or no L0 file. Pull Request resolved: https://github.com/facebook/rocksdb/pull/11498 Test Plan: none as I am unable to prove it is better Reviewed By: hx235 Differential Revision: D46339050 Pulled By: ajkr fbshipit-source-id: fcfc6da4111c5b3ae86d79d908afc5f61f96675b
Showing
想要评论请 注册 或 登录