In this paper, we investigate the roles of replication vs. repair to achieve durability in large-scale distributed storage systems. Specifically, we address the fundamental questions: How does the lifetime of an objec...
详细信息
the proceedings contain 28 papers. the topics discussed include: optimizing the reliability of pipelined applications under throughput constraints;decomposition based algorithm for state prediction in large scale dist...
ISBN:
(纸本)9780769541204
the proceedings contain 28 papers. the topics discussed include: optimizing the reliability of pipelined applications under throughput constraints;decomposition based algorithm for state prediction in large scale distributed systems;operational semantics of the Marte repetitive structure modeling concepts for data-parallel applications design;butterfly automorphisms and edge faults;cost performance analysis in multi-level tree networks;pretty good accuracy in matrix multiplication with GPUs;exploiting the power of GPUs for multi-gigabit wireless baseband processing;parallel cycle based logic simulation using graphics processing units;practical uniform peer sampling under churn;improving grid fault tolerance by means of global behavior modeling;toward a reliable distributed data management system;energy minimization on thread-level speculation in multicore systems;and resource-aware compiler prefetching for many-cores.
An approach to carrying out asynchronous distributed simulation of multiprocessor message passing architectures is presented. Aiming at achieving better performance on Conservative DEVS-based simulations, we introduce...
详细信息
ISBN:
(纸本)9780769542515
An approach to carrying out asynchronous distributed simulation of multiprocessor message passing architectures is presented. Aiming at achieving better performance on Conservative DEVS-based simulations, we introduce the GLM protocol which borrows the idea of safe processing intervals from the conservative time window algorithm and maintains global synchronization in a fashion similar to the distributed snapshot technique. Under the GLM scheme, a central lookahead manager (LM) exists which is in charge of receiving every LP's lookahead, identifying the global minimum lookahead of the system, and broadcasting it via null messages to all LPs. the simulation is divided into cycles of two phases: parallel phase and Broadcast phase. the GLM protocol is asynchronous and the central lookahead manager is not expected to be a bottleneck since the only message transmissions involving it take place when all LPs are blocked waiting for permission to advance their LVTs. the results presented in this paper show that the GLM protocol not only significantly reduces the total number of null messages, but it improves the performance and higher speedups are achieved.
Transactional memory is an alternative programming model for managing contention in accessing shared in-memory data objects. distributed transactional memory (TM) promises to alleviate difficulties with lock-based (di...
详细信息
Software transactional memory (STM) algorithms often rely on centralized components to achieve atomicity, isolation and consistency. In a distributed setting, centralized components are undesirable as they impair scal...
详细信息
Loop tiling is an important compiler transformation used for enhancing data locality and exploiting coarsegrained parallelism. Tiled codes in which tile sizes are runtime parameters - called parametrically-tiled codes...
详细信息
Memoryless online routing (MOR) algorithms are important for the applications with only local information available to make routing decisions. this paper gives two new MOR algorithms for a class of geometric graphs ca...
详细信息
New architectures have recently been proposed and deployed to support end-to-end advance reservation of network resources. these architectures rely on the use a centralized scheduler, which may be unpractical in large...
详细信息
this paper investigates scheduling loosely coupled task-bundles in highly heterogeneous distributed systems. Two allocation quality metrics are used in pay-per-service distributed applications: efficiency in terms of ...
详细信息
Sorting is a commonly used process with a wide breadth of applications in the high performance computing field. Early research in parallelprocessing has provided us with comprehensive analysis and theory for parallel...
详细信息
暂无评论