新的调度器解决上述问题的方法是,基于每个 CPU 来分布时间片,并且取消了全局同步和重算循环。
The new scheduler addresses these issues by distributing timeslices on a per-CPU basis and eliminating the global synchronization and recalculation loop.
这种类型的无限循环有时是由于多线程访问共享的不同步的数据结构造成的。
This type of infinite loops can sometimes occur due to multithreaded access into a Shared unsynchronized data structure.
根据我们的观察,我们似乎需要对一些锁的访问进行优化,比如线程执行的同步块代码在一个循环体中。
In light of our observation it seems as if we should be able to optimize access in the case where a thread is looping over a synchronized block of code.
应用推荐