m[ft] No need to use atomic ops for cached_x_scale We have added a mutex since, so no need for atomicity.
2021-02-22T12:50:41