Filters








37,270 Hits in 6.0 sec

On the Cost of Concurrency in Transactional Memory [article]

Petr Kuznetsov, Srivatsan Ravi
2013 arXiv   pre-print
We evaluate the cost of concurrency by measuring the amount of expensive synchronization that must be employed in an STM implementation that ensures positive concurrency, i.e., allows for concurrent transaction  ...  The crux of software transactional memory (STM) is to combine an easy-to-use programming interface with an efficient utilization of the concurrent-computing abilities provided by modern machines.  ...  The authors are grateful to Michel Raynal and Rachid Guerraoui for inspiring discussions on the properties and costs of STM and Damien Imbs for valuable comments on the previous drafts.  ... 
arXiv:1103.1302v9 fatcat:kndzjcwvmbeo3bkq3g5lptv4fi

On the Cost of Concurrency in Transactional Memory [article]

Srivatsan Ravi
2015 arXiv   pre-print
But what are the inherent costs associated with providing high degrees of concurrency in TMs? This is the central question of the thesis.  ...  TM allows the programmer to speculatively execute sequences of shared-memory operations as atomic transactions with all-or-nothing semantics: the transaction can either commit, in which case it appears  ...  We say that a concurrent implementation of a given sequential data type is locally serializable if it ensures that the local execution of each operation is equivalent to some execution of its sequential  ... 
arXiv:1511.01779v1 fatcat:ahlaq4z7dffklp7fiytspfmbti

On the Cost of Concurrency in Transactional Memory [chapter]

Petr Kuznetsov, Srivatsan Ravi
2011 Lecture Notes in Computer Science  
Schlussendlich beweisen wir, dass optimistische, auf spekulativen Ausführungen basierende, Synchronisierungstechniken, in einem präzisen Sinne, besser geeignet sind um Nebenläugkeit auszunutzen als pessimistische  ...  this observation, our HyTM implementation described in Algorithm 7.2 overcomes the linear per-read instrumentation cost by allowing hardware readers to abort due to a concurrent software writer, but maintains  ...  progressiveness for software transactions, unlike [34, 96, 99] .  ... 
doi:10.1007/978-3-642-25873-2_9 fatcat:7ehtxssf45abzc2bnigccm6evq

Enhanced Disk-Based Databases Towards Improved Hybrid In-Memory Systems

Samuel Kaspi, Sitalakshmi Venkatraman
2015 Zenodo  
This paper proposes a novel EMA technique and presents a comparative study between disk-based EMA systems and in-memory systems running on hardware configurations of equivalent power in terms of the number  ...  In-memory database systems are becoming popular due to the availability and affordability of sufficiently large RAM and processors in modern high-end servers with the capacity to manage large in-memory  ...  not change the value of the Fig. 7 7 Performance of disk-based EMA (no cost) & in-memory systems (20 Processors, 1 GIPS) Fig. 8 Performance of disk-based EMA (no cost) & in-memory systems (10 Processors  ... 
doi:10.5281/zenodo.1098924 fatcat:nnsuisu2xjcldhen5jmikit5t4

Inherent Limitations of Hybrid Transactional Memory [chapter]

Dan Alistarh, Justin Kopinsky, Petr Kuznetsov, Srivatsan Ravi, Nir Shavit
2015 Lecture Notes in Computer Science  
On the cost of instrumentation and concurrency The following theorems can be proven using the model presented in the previous section.  ...  In particular, what are the inherent instrumentation costs of building a HyTM, and what are the trade-offs between these costs and the provided concurrency, i.e., the ability of the HyTM system to run  ... 
doi:10.1007/978-3-662-48653-5_13 fatcat:mc2wdxbumfh3homcjepsjlpghm

Inherent limitations of hybrid transactional memory

Dan Alistarh, Justin Kopinsky, Petr Kuznetsov, Srivatsan Ravi, Nir Shavit
2017 Distributed computing  
On the cost of instrumentation and concurrency The following theorems can be proven using the model presented in the previous section.  ...  In particular, what are the inherent instrumentation costs of building a HyTM, and what are the trade-offs between these costs and the provided concurrency, i.e., the ability of the HyTM system to run  ... 
doi:10.1007/s00446-017-0305-3 fatcat:odg2m3ikzzg3hazw4svhy2o3aa

Performance Characteristics of Hardware Transactional Memory for Molecular Dynamics Application on BlueGene/Q: Toward Efficient Multithreading Strategies for Large-Scale Scientific Applications

Manaschai Kunaseth, Rajiv K. Kalia, Aiichiro Nakano, Priya Vashishta, David F. Richards, James N. Glosli
2013 2013 IEEE International Symposium on Parallel & Distributed Processing, Workshops and Phd Forum  
We have investigated the performance characteristics of hardware transactional memory (HTM) on the BlueGene/Q computer in comparison with conventional concurrency control mechanisms, using a molecular  ...  We found that the bookkeeping cost of HTM is high but that the rollback cost is low.  ...  Hardware Transactional Memory on BlueGene/Q TM is an opportunistic concurrency control mechanism.  ... 
doi:10.1109/ipdpsw.2013.29 dblp:conf/ipps/KunasethKNVRG13 fatcat:otsjmjsn7jaaldkvoywxxdge5a

Concurrent Shadow Paging in the Flask Architecture [chapter]

David Munro, Richard Connor, Ron Morrison, Stephan Scheuerl, David Stemple
2000 Fully Integrated Data Environments  
Flask is a architecture which is designed to achieve this goal by defining a layering whereby the memory management is not tied to any one concurrency control scheme operating above.  ...  One solution is to provide a flexible system in which concurrency control schemes may be specified according to the particular task in hand, allowing the same data to be used in conjunction with different  ...  The work was supported in part by SERC grant GR/G 17578 and ESPRIT III Basic Research Action 6309 -FIDE 2 .  ... 
doi:10.1007/978-3-642-59623-0_34 fatcat:mdemvufpf5hmbj73b2ejvhbjue

Concurrent Shadow Paging in the Flask Architecture [chapter]

D. S. Munro, R. C. H. Connor, R. Morrison, S. Scheuerl, D. W. Stemple
1995 Persistent Object Systems  
Flask is a architecture which is designed to achieve this goal by defining a layering whereby the memory management is not tied to any one concurrency control scheme operating above.  ...  One solution is to provide a flexible system in which concurrency control schemes may be specified according to the particular task in hand, allowing the same data to be used in conjunction with different  ...  The work was supported in part by SERC grant GR/G 17578 and ESPRIT III Basic Research Action 6309 -FIDE 2 .  ... 
doi:10.1007/978-1-4471-2122-0_3 dblp:conf/pos/MunroCMSS94 fatcat:2icbcex7avektb3sr3x7f7gbve

Performance Analysis of Concurrency Control Mechanisms for OLTP Databases

Samuel Kaspi, Sitalakshmi Venkatraman
2014 International Journal of Information and Education Technology  
However, with the recent advances of cost-effective main memory or in-memory storage that can support much higher transaction rates than disk-based systems, there is sufficient motivation to re-investigate  ...  Index Terms-Concurrency control, online transaction processing (OLTP), in-memory databases, performance.  ...  The cost of wasted work is largely dependant on the speed at which the work can be done. An important factor in determining speed is the speed of the processor.  ... 
doi:10.7763/ijiet.2014.v4.420 fatcat:xt7otnchnjebtf56iggjqyic6u

Needed

Hagit Attiya
2008 ACM SIGACT News  
This experience indicates what should be incorporated in a theory for transactional memory-and in a broader perspective, concurrent data structures. The first ingredient are specifications.  ...  It's all very well in practice, but it will never work in theory" (French management saying) Transactional memory is a leading paradigm for designing concurrent applications for tomorrow's multi-core architectures  ...  (1) Is the transaction stated as a list of atomic actions (like in classical database transactions) or more semantically as capturing a high-level operation (as in distributed shared memory)?  ... 
doi:10.1145/1360443.1360457 fatcat:kdadokwocnaefoh5oszegrzr6i

Integrated concurrency control and recovery mechanisms: design and performance evaluation

Rakesh Agrawal, David J. Dewitt
1985 ACM Transactions on Database Systems  
In spite of the wide variety of concurrency control and recovery mechanisms proposed during the past decade, the behavior and the performance of various concurrency control and recovery mechanisms remain  ...  In this paper, we take a unified view of the problems associated with concurrency control and recovery for transaction-oriented multiuser centralized database management systems, and we present several  ...  The anonymous referees provided many insightful comments and suggestions on an earlier version of this paper.  ... 
doi:10.1145/4879.4958 fatcat:xwzympk4ofhurespdlui53pkga

Clustered Checkpointing and Partial Rollbacks for Reducing Conflict Costs in STMs

Monika Gupta, Rudrapatna K Shyamasundar, Shivali Agarwal
2010 International Journal of Computer Applications  
With the advent of massive multi-cores, more transactions can be initiated concurrently, however resulting in an increase in the percentage of conflicting transactions.  ...  The algorithm is based on intelligent checkpointing of transactions as they proceed, and, in case of conflict, rolling them back to a safe, consistent, intermediate checkpoint, thus reducing conflict costs  ...  each other. • Further, it shows that CCPR can deliver a cost reduction of 16% to 18% in terms of reduction in the total number of shared memory read operations.  ... 
doi:10.5120/438-668 fatcat:hzbyblpf45ckhnc5abr7bpfp6i

Why nothing matters

Xi Yang, Stephen M. Blackburn, Daniel Frampton, Jennifer B. Sartor, Kathryn S. McKinley
2011 SIGPLAN notices  
The direct cost of zero initialization is surprisingly high: up to 12.7%, with average costs ranging from 2.7 to 4.5% on a high performance virtual machine on IA32 architectures.  ...  Existing virtual machines either: a) minimize direct costs by zeroing in large blocks, or b) minimize indirect costs by zeroing in the allocation sequence, which reduces cache displacement and bandwidth  ...  Concurrent Zeroing Bulk zeroing attacks the direct cost of zeroing.  ... 
doi:10.1145/2076021.2048092 fatcat:tai7u7rcczfyxa5yqleee2w7py

Why nothing matters

Xi Yang, Stephen M. Blackburn, Daniel Frampton, Jennifer B. Sartor, Kathryn S. McKinley
2011 Proceedings of the 2011 ACM international conference on Object oriented programming systems languages and applications - OOPSLA '11  
The direct cost of zero initialization is surprisingly high: up to 12.7%, with average costs ranging from 2.7 to 4.5% on a high performance virtual machine on IA32 architectures.  ...  Existing virtual machines either: a) minimize direct costs by zeroing in large blocks, or b) minimize indirect costs by zeroing in the allocation sequence, which reduces cache displacement and bandwidth  ...  Concurrent Zeroing Bulk zeroing attacks the direct cost of zeroing.  ... 
doi:10.1145/2048066.2048092 dblp:conf/oopsla/YangBFSM11 fatcat:akoq4aoylzhjtnzlvldhwh25ye
« Previous Showing results 1 — 15 out of 37,270 results