RH Lock: A Scalable Hierarchical Spin Lock
نویسندگان
چکیده
Scalable architectures with non-uniform memory access time (NUMAs) have gained increased popularity in recent years. The increased scalability have increased the demand for scalable lock implementations, such as the queue-based locks of Mellor-Crummey and Scott (MCS lock), and of Craig, Landin and Hagersten (CLH lock). This paper demonstrates that the first-come first-served nature of queue-based locks make them less suitable for nonuniform communication architectures (NUCAs), for example NUMAs built from a few large nodes. In contrast, the simpler test-and-set locks gives a unfair advantage to neighboring processors when a lock is released, which will create a fast lock handover time as well as create more locality for the data accessed in the critical region. We also propose the new RH lock that explores the NUCA architectures by creating a controlled unfairness in combination with a much reduced traffic compared with the testand-set locks. A critical section guarded by the RH lock is shown to take less than half the time to execute compared with the same critical section guarded by any other lock. We also investigate the effectiveness of our new lock on a set of real SPLASH-2 applications. For example, execution time for Raytrace with 30 processors can be improved between 1.83 and 5.70 times by using the RH locks instead of any other tested locks.
منابع مشابه
Hydraulic anti-lock, anti-skid braking system using fuzzy controller
To maintain the stability trajectory of vehicles under critical driving conditions, anti lock-anti skid controllers, consisting of four anti-lock sub-controllers for each wheel and two anti-skid sub-controllers for left and right pair wheels have been separately designed. Wheel and body systems have been simulated with seven degrees of freedom to evaluate the proper functioning of controllers. ...
متن کاملLock cohorting: A general technique for designing NUMA locks Citation
Multicore machines are quickly shifting to NUMA and CC-NUMA architectures, making scalable NUMA-aware locking algorithms, ones that take into account the machines’ non-uniform memory and caching hierarchy, ever more important. This paper presents lock cohorting, a general new technique for designing NUMA-aware locks that is as simple as it is powerful. Lock cohorting allows one to transform any...
متن کاملOptimizing a FIFO, scalable spin lock using consistent memory
This paper presents a FIFO queue-based spin lock that (1) uses only one atomic swap operation; (2) is scalable as it requires a constant amount of communication; (3) runs without a coherent cache support; and (4) provides a timing guarantee required for real-time applications. The algorithm is optimal in terms of the number of atomic operations required to solve a scalable mutual exclusion prob...
متن کاملHierarchical Backoff Locks for Nonuniform Communication Architectures
This paper identifies node affinity as an important property for scalable general-purpose locks. Nonuniform communication architectures (NUCAs), for example CCNUMAs built from a few large nodes or from chip multiprocessors (CMPs), have a lower penalty for reading data from a neighbor’s cache than from a remote cache. Lock implementations that encourages handing over locks to neighbors will impr...
متن کامل