CSMV: A highly scalable multi-versioned software transactional memory for GPUs

被引：0

作者：

Nunes, Diogo ^{[1
,2
]}

Castro, Daniel ^{[1
,2
]}

Romano, Paolo ^{[1
,2
]}

机构：

[1] Univ Lisbon, INESC ID, Rua Alves Redol 9, P-1000029 Lisbon, Portugal

[2] Univ Lisbon, Inst Super Tecn, Rua Alves Redol 9, P-1000029 Lisbon, Portugal

来源：

JOURNAL OF PARALLEL AND DISTRIBUTED COMPUTING | 2023年 / 180卷

关键词：

Multi-version concurrency control; Synchronization; GPU; Transactional memory; Concurrent programming; MULTIVERSION CONCURRENCY-CONTROL;

D O I：

10.1016/j.jpdc.2023.04.002

中图分类号：

TP301 [理论、方法];

学科分类号：

081202 ;

摘要：

This paper introduces CSMV (Client Server Multiversioned), a multi-versioned Software TM (STM) for GPUs that adopts an innovative client-server design. By decoupling the execution of transactions from their commit process, CSMV provides two main benefits: (i) it enables the use of fast on chip memory to access the global metadata used to synchronize transaction (ii) it allows for implementing highly efficient collaborative commit procedures, tailored to take full advantage of the architectural characteristics of GPUs. Via an extensive experimental study, we show that CSMV achieves up to 3 orders of magnitude speed-ups with respect to state of the art STMs for GPUs and that it can accelerate by up to 20x irregular applications running on state of the art STMs for CPUs. & COPY; 2023 Elsevier Inc. All rights reserved.

引用

页数：17

共 6 条

[1] CSMV: A Highly Scalable Multi-Versioned Software Transactional Memory for GPUs
Nunes, Diogo
Castro, Daniel
Romano, Paolo
2022 IEEE 36TH INTERNATIONAL PARALLEL AND DISTRIBUTED PROCESSING SYMPOSIUM (IPDPS 2022), 2022, : 526 - 536
[2] Lock-free and Scalable Multi-Version Software Transactional Memory
Fernandes, Sergio Miguel
Cachopo, Joao
ACM SIGPLAN NOTICES, 2011, 46 (08) : 179 - 188
[3] The Power of DCAS: Highly-Concurrent Software Transactional Memory
Attiya, Hagit
Hillel, Eshcar
PODC'07: PROCEEDINGS OF THE 26TH ANNUAL ACM SYMPOSIUM ON PRINCIPLES OF DISTRIBUTED COMPUTING, 2007, : 342 - 343
[4] CMS: Checkpoint-Based Multi-versioning System for Software Transactional Memory
Ghosh, Ammlan
Chaki, Rituparna
Chaki, Nabendu
PROGRESS IN INTELLIGENT COMPUTING TECHNIQUES: THEORY, PRACTICE, AND APPLICATIONS, VOL 1, 2018, 518 : 471 - 482
[5] An autonomic-computing approach on mapping threads to multi-cores for software transactional memory
Zhou, Naweiluo
Delaval, Gwenael
Robu, Bogdan
Rutten, Eric
Mehaut, Jean-Francois
CONCURRENCY AND COMPUTATION-PRACTICE & EXPERIENCE, 2018, 30 (18)
[6] Accelerating multi-dimensional population balance model simulations via a highly scalable framework using GPUs
Sampat, Chaitanya
Baranwal, Yukteshwar
Ramachandran, Rohit
COMPUTERS & CHEMICAL ENGINEERING, 2020, 140 (140)

← 1 →