Traditional multi-purpose operating systems are designed to support a wide range of usage models and requirements.
Unpredictable operating system overhead is one significant reason a processor might take longer to reach the synchronization point than the others.
Custom lightweight kernel operating systems, used on some of the fastest computers in the world, help alleviate this problem.
[4] Although it is surprisingly difficult to exactly define what a lightweight kernel is,[5] there are some common design goals: LWK implementations vary, but all strive to provide applications with predictable and maximum access to the central processing unit (CPU) and other system resources.
By restricting services to only those that are absolutely necessary and by streamlining those that are provided, the overhead (sometimes called noise) of the lightweight operating system is minimized.
Since the application can make consistent progress on each processor, they will reach their synchronization points faster, ideally at the same time.