Filters








15,970 Hits in 3.9 sec

On Partial Wait-Freedom in Transactional Memory [article]

Petr Kuznetsov, Srivatsan Ravi
2014 arXiv   pre-print
In this paper, we explore the costs of providing wait-freedom to only a subset of transactions.  ...  It is known that dynamic transactional memory cannot provide wait-free progress in the sense that every transaction commits in a finite number of its own steps.  ...  Perhaps, the most attractive progress property a TM may satisfy is wait-freedom: every transaction commits regardless of the behavior of concurrent processes [12] .  ... 
arXiv:1407.6876v2 fatcat:utmhoxk3xfb3fhor6xnpa7eq2m

Distributed computing column 45

Idit Keidar
2012 ACM SIGACT News  
, such as obstruction-freedom and wait-freedom.  ...  , and it should avoid double synchronization: one in the TM and the other in the underlying memory.  ... 
doi:10.1145/2160649.2160671 fatcat:jlgskapmtngbpavipzyck6x3km

4 Concurrent programming [chapter]

1987 Lecture Notes in Computer Science  
Software transactional memory (STM) simplifies development of concurrent code by allowing the programmer to mark sections of code to be executed concurrently and atomically in an optimistic manner.  ...  In contrast to locks, STMs are easy to compose and do not suffer from deadlocks. We have designed and implemented two STMs for graphics processors, one blocking and one non-blocking.  ...  Acknowledgements This work was partially supported by the EU as part of FP7 Project PEPPHER (www.peppher.eu) under grant 248481 and the Swedish Research Council under grant number 37252706.  ... 
doi:10.1007/3-540-18008-7_4 fatcat:3o2qdhb67zaadox7iwcqnkxlwi

Distributed Computing Column 58

Jennifer L. Welch
2015 ACM SIGACT News  
References Concluding Thoughts While the discussion above spans much of the history of transactional memory, and mentions many open questions, the coverage has of necessity been spotty, and the choice  ...  Most of all, my thanks and admiration to Maurice Herlihy for his seminal contributions, not only to transactional memory, but to nonblocking algorithms, topological analysis, and so many other aspects  ...  3: Weaker liveness properties Linearisable wait-free implementations tend to be complex, and one culprit seems to be wait freedom.  ... 
doi:10.1145/2789149.2789163 fatcat:ueub5knoofar3jee6plc763yni

Concurrent Data Structures [chapter]

Mark Moir, Nir Shavit
2004 Handbook of Data Structures and Applications  
Clearly, wait-freedom is a stronger condition than lock-freedom, and lock-freedom in turn is stronger than obstruction-freedom.  ...  To formalize this idea, various nonblocking progress conditions-such as wait-freedom [49, 83] , lock-freedom [49] , and obstruction-freedom [54] -have been considered in the literature.  ...  , 1-4 obstruction-freedom, 1-5 optimistic concurrency control, 1-12 optimistic synchronization, 1-6 parallel random access machines, 1-7 pools, 1-17 preemption-safe locks, 1-10  ... 
doi:10.1201/9781420035179.ch47 fatcat:b6onln3r3fb2ff3sitg64nilhi

Turbocharging boosted transactions or

Chinmay Eishan Kulkarni, Osman Unsal, Adrián Cristal, Eduard Ayguadé, Mateo Valero
2008 Proceedings of the 14th ACM SIGPLAN symposium on Principles and practice of parallel programming - PPoPP '09  
Our proposals, as implemented in DSTM2, increase the transactional throughput of the system, often by more than 30%.  ...  We describe a linear-time algorithm to detect transactions that cannot make progress, which transactions need to be aborted, and when.  ...  wait on no more than one lock.  ... 
doi:10.1145/1504176.1504228 dblp:conf/ppopp/KulkarniUCAV09 fatcat:2h7buodaofb2bcbcsf7rsu6oj4

Turbocharging boosted transactions or

Chinmay Eishan Kulkarni, Osman Unsal, Adrián Cristal, Eduard Ayguadé, Mateo Valero
2009 SIGPLAN notices  
Our proposals, as implemented in DSTM2, increase the transactional throughput of the system, often by more than 30%.  ...  We describe a linear-time algorithm to detect transactions that cannot make progress, which transactions need to be aborted, and when.  ...  wait on no more than one lock.  ... 
doi:10.1145/1594835.1504228 fatcat:txcbuuht3jdjfg7togbgfngrru

Brief announcement

Konrad Siek, Paweł T. Wojciechowski
2013 Proceedings of the 25th ACM symposium on Parallelism in algorithms and architectures - SPAA '13  
However, in a distributed setting, where partial transaction failures may happen, the inability of pessimistic transactional memories to roll back is a major shortcoming.  ...  Transactional memory, an approach aiming to replace cumbersome locking mechanisms in concurrent systems, has become a popular research topic.  ...  However, distributed transactional memory, pessimistic or optimistic, must still support rollback, because it is possible for partial failures to occur in the system.  ... 
doi:10.1145/2486159.2486166 dblp:conf/spaa/SiekW13 fatcat:tatriwvcabaslpssf55izjbv3m

Preconditioning techniques for the conjugate gradient solver on a parallel distributed memory computer

C. Vollaire, L. Nicolas
1998 IEEE transactions on magnetics  
This paper describes the parallelization of the conjugate gradient algorithm fitted with three types of preconditioning in order to compute large finite element complex sparse system of equations on a  ...  distributed memory parallel computer.  ...  The concatenation of these partial results is performed by messages passing in SPMD mode.  ... 
doi:10.1109/20.717787 fatcat:vjnroa63n5alvazikcqgwoq4xy

Braids and fibers: Language constructs with architectural support for adaptive responses to memory latencies

D. F. Bacon, X. Shen
2006 IBM Journal of Research and Development  
The programming constructs facilitate the creation of programs that are partially ordered, in which the partial orders can be used to support adaptive responses to memory access latencies.  ...  As processor speeds continue to increase at a much higher rate than memory speeds, memory latencies may soon approach a thousand processor cycles.  ...  Acknowledgments This material is partially based upon work supported by the Defense Advanced Research Projects Agency (DARPA) under Contract No. NBCH3039004.  ... 
doi:10.1147/rd.502.0209 fatcat:2oqawgplqvhsfm3loonbrjvhdq

Mastering Concurrent Computing Through Sequential Thinking: A Half-century Evolution [article]

Sergio Rajsbaum, Michel Raynal
2018 arXiv   pre-print
Right from the start in the 1960s, the main way of dealing with concurrency has been by reduction to sequential reasoning.  ...  It is very difficult to master, yet it arises all over modern computing systems, both when the communication medium is shared memory and when it is by message passing.  ...  Lock-based implementations Mutex-free implementations Obstruction-freedom [31] Deadlock-freedom [16] Non-blocking [36] Starvation-freedom [37] Wait-freedom [32] READ/WRITE REGISTERS ON TOP  ... 
arXiv:1812.06011v1 fatcat:56rajh2lgngu3e5rlm42gdtjhe

Inherent limitations on disjoint-access parallel implementations of transactional memory

Hagit Attiya, Eshcar Hillel, Alessia Milani
2009 Proceedings of the twenty-first annual symposium on Parallelism in algorithms and architectures - SPAA '09  
In fact, a lower bound of Ω(t) is proved on the number of writes needed in order to implement a read-only transaction of t items, which successfully terminates in a disjoint-access parallel TM implementation  ...  Transactional memory (TM) is a popular approach for alleviating the difficulty of programming concurrent applications; TM guarantees that a transaction, consisting of a sequence of operations, appear to  ...  Section 3.2 strengthens this result and shows that a read-only transaction on t items (in a disjoint-access parallel transactional memory implementation with wait-free read-only transactions) must apply  ... 
doi:10.1145/1583991.1584015 dblp:conf/spaa/AttiyaHM09 fatcat:q4mcpbcphfae7o34t5l3oeweyi

Scaling HDFS with a Strongly Consistent Relational Model for Metadata [chapter]

Kamal Hakimzadeh, Hooman Peiro Sajjad, Jim Dowling
2014 Lecture Notes in Computer Science  
In this paper, we address the problem of how to migrate the HDFS' metadata to a relational model, so that we can support larger amounts of storage on a shared-nothing, in-memory, distributed database.  ...  both explicit locks and implicit locks on subtrees in the hierarchy.  ...  Transactions follow this partial ordering when taking locks, ensuring that the circular wait condition for deadlock never holds.  ... 
doi:10.1007/978-3-662-43352-2_4 fatcat:6wzxstuyyfc4fph5pjbh6neme4

Queue-Based and Adaptive Lock Algorithms for Scalable Resource Allocation on Shared-Memory Multiprocessors

Deli Zhang, Brendan Lynch, Damian Dechev
2014 International journal of parallel programming  
Many existing approaches solve this problem in a distributed system, but the explicit message passing paradigm they adopt is not optimal for shared-memory.  ...  Our methodology is based on a non-blocking queue where competing threads spin on previous conflicting resource requests.  ...  The authors would also like to thank Dimitry Vyukov for providing insightful implementation tips on the non-blocking queue.  ... 
doi:10.1007/s10766-014-0317-6 fatcat:5waj5ideqfharlijwbbvpjid2q

Single global lock semantics in a weakly atomic STM

Vijay Menon, Steven Balensiefer, Tatiana Shpeisman, Ali-Reza Adl-Tabatabai, Richard L. Hudson, Bratin Saha, Adam Welc
2008 SIGPLAN notices  
In contrast to database transactions, transaction memory (TM) semantics are complicated by the fact that programs may access the same memory locations both inside and outside transactions.  ...  As memory transactions have been proposed as a language-level replacement for locks, there is growing need for well-defined semantics.  ...  In Figure 1 , Thread 1's transaction will wait on Thread 2's transaction before it linearizes, which, in turn, prevents Thread 2 from reading an updated value from ready (which would still be in Thread  ... 
doi:10.1145/1402227.1402235 fatcat:hqnarklutvhgvb3gkpajtqvm2i
« Previous Showing results 1 — 15 out of 15,970 results