RCU-HTM: Combining RCU with HTM to Implement Highly Efficient Concurrent Binary Search Trees

Siakavaras, Dimitrios; Nikas, Konstantinos; Goumas, Georgios; Koziris, Nectarios

doi:10.1109/pact.2017.17

Cited by 10 publications

(5 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We use an internal AVL tree as an example, but the same procedure can be used for any type of ST. While the application of RCU‐HTM on binary STs has been presented in Reference 23, in this work we showcase the universality of the technique and its efficacy and applicability on any type of search tree. We implement RCU‐HTM versions of 12 types of STs, showcasing its applicability. We implement three unbalanced binary STs (internal, partially external, and external), six balanced binary STs (internal, partially external, and external versions of both AVL and Red‐Black trees), 1 relaxed‐balanced partially external AVL ST, a B+‐tree (i.e., the external version of a B‐tree) and an (a‐b)‐tree (i.e., a relaxed‐balanced variant of a B+‐tree). We evaluate RCU‐HTM STs and compare their performance with several state‐of‐the‐art implementations that use four different synchronization methods, namely, locks, atomic operations, RCU and HTM.…”

Section: Introductionmentioning

confidence: 67%

RCU‐HTM: A generic synchronization technique for highly efficient concurrent search trees

Siakavaras

Nikas

Goumas

et al. 2021

Concurrency and Computation

Self Cite

View full text Add to dashboard Cite

Concurrent search trees (STs) are among the most widely used data structures to store and retrieve data in contemporary multithreaded applications. Despite the high amount of prior work, it still remains challenging to implement highly efficient concurrent STs. This is mainly due to the fact that both traditional synchronization methods (i.e., locks and atomic operations) and more novel ones (i.e., read‐copy‐update and transactional memory) fail to provide solutions that are generic and at the same time able to attain high performance under diverse workloads and contention levels. In this work, we present RCU‐HTM, a technique that combines read‐copy‐update (RCU) and hardware transactional memory (HTM), and: (a) supports the implementation of a concurrent version of any type of search tree, and (b) achieves high performance across all execution scenarios. We also incorporate a low‐overhead, epoch‐based memory reclamation scheme to make our RCU‐HTM trees practical for large‐scale long‐running applications. To showcase the capabilities of our technique, we implement and evaluate multiple RCU‐HTM trees and compare their performance with several state‐of‐the‐art competitors. More specifically, we apply RCU‐HTM to 12 different types of binary, B+ and (a,b)‐trees and compare against 18 state‐of‐the‐art implementations that use four different synchronization mechanisms, namely, locks, atomic operations, RCU, and HTM. We evaluate the trees under different levels of contention by varying the size of the tree, the operations mix, and the number of threads, for a total of 210 execution scenarios for each implementation. Our evaluation shows that in the majority of executions, RCU‐HTM trees outperform their state‐of‐the‐art alternatives, and even in the cases where they do not, their performance is very close to that of the best implementation.

show abstract

Section: Introductionmentioning

confidence: 67%

RCU‐HTM: A generic synchronization technique for highly efficient concurrent search trees

Siakavaras

Nikas

Goumas

et al. 2021

Concurrency and Computation

Self Cite

View full text Add to dashboard Cite

show abstract

“…First, we evaluate pointer chasing workloads, i.e., lock-based concurrent data structures from the ASCYLIB library [31], used as key-value sets. In ASCYLIB's Binary Search Tree (BST) [37], the lock memory requests are only 0.1% of the total memory requests, so we also evaluate an external fine-grained locking BST from [130]. Data structures are initialized with a fixed size and statically partitioned across NDP units, except for BSTs, which are distributed randomly.…”

Section: Ndp Coresmentioning

confidence: 99%

“…Stack [31] 100K -100% push Queue [31,104] 100K -100% pop Array Map [31,56] 10 -100% lookup Priority Queue [11,31,118] 20K -100% deleteMin Skip List [31,118] 5K -100% deletion Hash Table [31,63] 1K -100% lookup Linked List [31,63] 20K -100% lookup Binary Search Tree Fine-Grained (BST_FG) [130] 20K -100% lookup Binary Search Tree Drachsler (BST_Drachsler) [31,37] 10K -100% deletion…”

Section: Data Structure Configurationmentioning

confidence: 99%

SynCron: Efficient Synchronization Support for Near-Data-Processing Architectures

Giannoula¹,

Vijaykumar²,

Παπαδοπούλου³

et al. 2021

Preprint

Self Cite

View full text Add to dashboard Cite

Near-Data-Processing (NDP) architectures present a promising way to alleviate data movement costs and can provide significant performance and energy benefits to parallel applications. Typically, NDP architectures support several NDP units, each including multiple simple cores placed close to memory. To fully leverage the benefits of NDP and achieve high performance for parallel workloads, efficient synchronization among the NDP cores of a system is necessary. However, supporting synchronization in many NDP systems is challenging because they lack shared caches and hardware cache coherence support, which are commonly used for synchronization in multicore systems, and communication across different NDP units can be expensive.This paper comprehensively examines the synchronization problem in NDP systems, and proposes SynCron, an endto-end synchronization solution for NDP systems. SynCron adds low-cost hardware support near memory for synchronization acceleration, and avoids the need for hardware cache coherence support. SynCron has three components: 1) a specialized cache memory structure to avoid memory accesses for synchronization and minimize latency overheads, 2) a hierarchical message-passing communication protocol to minimize expensive communication across NDP units of the system, and 3) a hardware-only overflow management scheme to avoid performance degradation when hardware resources for synchronization tracking are exceeded.We evaluate SynCron using a variety of parallel workloads, covering various contention scenarios. SynCron improves performance by 1.27× on average (up to 1.78×) under highcontention scenarios, and by 1.35× on average (up to 2.29×) under low-contention real applications, compared to state-ofthe-art approaches. SynCron reduces system energy consumption by 2.08× on average (up to 4.25×).

show abstract

“…Given we did not implement opt-s-tree in HTM due to it requiring lightweight reads, one should be able to combine the approaches of COP and s-tree for improved performance. Siakavaras et al 50 applied Read-Copy-Update (RCU) techniques to context of AVL trees in HTM by having update operations create a copy of the section of the tree they modify, and then linking in the modified tree while keeping the original nodes in the read set to prevent conflicts. This allows read operations to be performed entirely without synchronization.…”

Section: Related Workmentioning

confidence: 99%

“…Different to s-tree, modifications due to rotations may still result in large transactions resulting in a trade-off between fast reads and a possible increase in conflicts. Note that both the works of Avni and Kuszmaul 49 and Siakavaras et al 50 relied on the strong memory model semantics of Intel's HTM between transactional and non-transactional accesses, while the non-optimized version of s-tree does not.…”

Section: Related Workmentioning

confidence: 99%

A speculation‐friendly binary search tree

Crain

Gramoli

Raynal

2018

Concurrency and Computation

View full text Add to dashboard Cite

Abstract:We introduce the first binary search tree algorithm designed for speculative executions. Prior to this work, tree structures were mainly designed for their pessimistic (non-speculative) accesses to have a bounded complexity. Researchers tried to evaluate transactional memory using such tree structures whose prominent example is the red-black tree library developed by Oracle Labs that is part of multiple benchmark distributions. Although well-engineered, such structures remain badly suited for speculative accesses, whose step complexity might raise dramatically with contention.We show that our speculation-friendly tree outperforms the existing transaction-based version of the AVL and the red-black trees. Its key novelty stems from the decoupling of update operations: they are split into one transaction that modifies the abstraction state and multiple ones that restructure its tree implementation in the background. In particular, the speculation-friendly tree is shown correct, reusable and it speeds up a transaction-based travel reservation application by up to 3.5×. Key-words: Transactional Memory, concurrent data structures, balanced binary treesUn arbre binaire de recherche dédié aux accès transactionnels Résumé : Les transactions, qui utilisent une technique de synchronisation optimiste, simplifient la programmation des architectures multi-coeurs. En effet, un programmeur a seulement besoin d'insérer des opérations dans des transactions afin d'obtenir un programme concurrent correct. Les programmeurs ont donc tout naturellement utilisé cette encapsulation sur plusieurs structures de données originellement dédiéesà la programmation pessimiste (non optimistes) pourévaluer les transactions. L'exemple le plus probantétant l'arbre transactionnel rouge-noir développé par Oracle Labs et présent dans plusieurs distributions.Malheureusement, ces structures de données ne sont pas adaptéesà des exécutions optimistes car elles reposent sur des invariants contraignants, comme la profondeur logarithmique d'un arbre, pour borner le coût des accès pessimistes. Une telle complexité ne s'applique pas aux accés optimistes et garantir de tels invariants n'est pas nécessaire et induit de la contention en augmentant la probabilité pour une transaction d'avorter et de recommencer.Dans ce papier, nous présentons un arbre binaire de recherche qui viole de façon transitoire ces invariants pour gagner en efficacité. Nous montrons que cet arbre est plus efficace que les arbres AVL et rouge-noir. Sa nouveauté majeure réside dans la séparation de ses opérations de modifications : elles sont coupées en une transaction qui modifie l'abstraction et plusieurs autres qui restructurent son implémentation. La bibliothèque qui en résulte limite la quantité d'avortements en augmentant légèrement le nombre d'accès, en particulier, elle améliore une application de réservation de voyage basée sur les transactions par un facteur multiplicatif de 3,5.

show abstract

RCU-HTM: Combining RCU with HTM to Implement Highly Efficient Concurrent Binary Search Trees

Cited by 10 publications

References 22 publications

RCU‐HTM: A generic synchronization technique for highly efficient concurrent search trees

RCU‐HTM: A generic synchronization technique for highly efficient concurrent search trees

SynCron: Efficient Synchronization Support for Near-Data-Processing Architectures

A speculation‐friendly binary search tree

Contact Info

Product

Resources

About