Skip to content

RwLock result in a twofold degradation of performance #64

@AshinGau

Description

@AshinGau

Description

Even in situations where the conflict domain is small, RwLock result in a twofold degradation of performance.

As illustrated in the code below, the number of times a write lock is acquired and the instances where conditions are not met are counted. In the continuous block test, the occurrences of unmet conditions are virtually negligible, indicating a high probability of achieving correct results without employing the write lock. Over ten thousand write locks are applied in total; however, the critical section governed by these locks merely compares the magnitude of two integers, which leads to a halving of performance. This implies that by excising this segment of safety code, the parallel execution time could be reduced to approximately 240 milliseconds.
branch: lock_free_dag
Image
Image

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions