Artículos relacionados a Detecting Nondeterminism in Shared Memory Parallel...

Detecting Nondeterminism in Shared Memory Parallel Programs: Technical Report 526 (Classic Reprint) - Tapa blanda

Dinning, Anne

 
9781332088386: Detecting Nondeterminism in Shared Memory Parallel Programs: Technical Report 526 (Classic Reprint)

Sinopsis

Excerpt from Detecting Nondeterminism in Shared Memory Parallel Programs: Technical Report 526

Build Parent Vector Using Modification Set Check Concurrent Using Iterate Vectors Task Assignment and Free Task Management Free Task Path Compression.

About the Publisher

Forgotten Books publishes hundreds of thousands of rare and classic books. Find more at www.forgottenbooks.com

This book is a reproduction of an important historical work. Forgotten Books uses state-of-the-art technology to digitally reconstruct the work, preserving the original format whilst repairing imperfections present in the aged copy. In rare cases, an imperfection in the original, such as a blemish or missing page, may be replicated in our edition. We do, however, repair the vast majority of imperfections successfully; any imperfections that remain are intentionally left to preserve the state of such historical works.

"Sinopsis" puede pertenecer a otra edición de este libro.

Reseña del editor

Excerpt from Detecting Nondeterminism in Shared Memory Parallel Programs: Technical Report 526

Build Parent Vector Using Modification Set Check Concurrent Using Iterate Vectors Task Assignment and Free Task Management Free Task Path Compression.

About the Publisher

Forgotten Books publishes hundreds of thousands of rare and classic books. Find more at www.forgottenbooks.com

This book is a reproduction of an important historical work. Forgotten Books uses state-of-the-art technology to digitally reconstruct the work, preserving the original format whilst repairing imperfections present in the aged copy. In rare cases, an imperfection in the original, such as a blemish or missing page, may be replicated in our edition. We do, however, repair the vast majority of imperfections successfully; any imperfections that remain are intentionally left to preserve the state of such historical works.

Reseña del editor

Excerpt from Detecting Nondeterminism in Shared Memory Parallel Programs: Technical Report 526

This thesis addresses the problem of detecting of a specific type of
nondeterminism in shared memory parallel programs known as access anomalies. An access anomaly occurs when an update to a shared variable X is concurrent with either a read of X or another update of X.

The first part of the work considers dynamic detection of access anomalies. We introduce a new technique called task recycling that detects access anomalies "on the fly" by monitoring the program execution. This technique is designed with two goals in mind. The first goal is minimal monitoring overhead. Costs are incurred only at thread create, terminate, and coordinate operations and every time a monitored variable is accessed. Because variable accesses are generally the most frequent operation, the task recycling technique reduces the overhead per variable access to a small constant. The second goal is generality. The task recycling technique is applicable to a wide variety of parallel constructs and all common synchronous and asynchronous coordination primitives. Combined with a protocol for specifying ordering constraints, the method of representing
concurrency relationships in task recycling cam be extended to detect general race conditions in parallel programs.

The second part of the thesis involves static detection of several types of nondeterminism that makes dynamic anomaly detection inefficient. In particular, the notion of nondeterminism arising from critical section coordination is refined by distinguishing between three types of nondeterminism - parallel, sequential, and reference nondeterminism. The presence of these types of nondeterminism in a program impacts access anomaly detection in two significant ways: (i) how critical section coordination is modeled during anomaly detection, and (ii) the confidence level and complexity of guaranteeing that a program has no access anomalies. In particular, it is shown that access anomalies can be detected efficiently only if a program is parallel, sequential and reference deterministic. Heuristics are presented that make access anomaly detection tractable in the presence of other nondeterminism through a better classification and semantic understanding of a coordination protocol.

About the Publisher

Forgotten Books publishes hundreds of thousands of rare and classic books. Find more at www.forgottenbooks.com

This book is a reproduction of an important historical work. Forgotten Books uses state-of-the-art technology to digitally reconstruct the work, preserving the original format whilst repairing imperfections present in the aged copy. In rare cases, an imperfection in the original, such as a blemish or missing page, may be replicated in our edition. We do, however, repair the vast majority of imperfections successfully; any imperfections that remain are intentionally left to preserve the state of such historical works.

"Sobre este título" puede pertenecer a otra edición de este libro.