J. Mellor-Crummey. Compile-time support for efficient data race detection in shared-memory parallel programs. In Proceedings of the 3rd ACM Workshop on Parallel and Distributed Debugging, pages 129--139, May 1993.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....in race detection is made out as the run time overhead in determining whether an access is to shared memory or not. They describe that the application gets slow down by an average factor of approximately 2. There is a study on reducing the monitored behavior of program execution. Mellor Crummey [11] proposed an instrumentation tool for run time race detection which applies compile time analysis to prune variable references that need not be monitored at run time. Using dependence analysis and interprocedural analysis of scalar side effects, the tool was able to reduce the dynamic counts of ....

J. Mellor-Crummey. Compile-time support for efficient data race detection in shared-memory parallel programs. In Proceedings of the 3rd ACM Workshop on Parallel and Distributed Debugging, pages 129--139, May 1993.

....in the iteration. 4 Related Work Many approaches for efficiently detecting races on the fly for parallel programs have been reported. In this section, we briefly mention some important work to improve the scalability of on the fly race detection. This work falls into two groups: compiler support [7] to reduce the number of monitored accesses, and underlying system support [8, 12, 13] using scalable distributed shared memory systems. Our technique is novel in that the scalability is provided with simple but powerful instrumented code which can be applied to most existing techniques. ....

....the number of monitored accesses, and underlying system support [8, 12, 13] using scalable distributed shared memory systems. Our technique is novel in that the scalability is provided with simple but powerful instrumented code which can be applied to most existing techniques. Mellor Crummey [7] describes an instrumentation tool for on the fly race detection which applies compile time analysis to identify variable references that need not be monitored at run time. Using dependence analysis and interprocedural analysis of scalar side effects, the tool was able to reduce the dynamic counts ....

Mellor-Crummey, J., "Compile-time Support for Efficient Data Race Detection in Shared-Memory Parallel Programs," 3rd Workshop on Parallel and Distributed Debugging, pp. 129-139, ACM, May 1993.

.... number of methods have been developed in the context of on the fly race detection that could be used as polynomial time algorithms for determining event orders in branch free fork join programs[MC91, DS90, NR88] Some recent efforts have focused on reducing the number of events that must be traced[MC93] or recorded[Net93] As these fork join analysis algorithms read the trace only once and have limited storage requirements they can often be executed on the fly, concurrently with the parallel program they are analyzing. Critical Sections with Lock Unlock In programs that contain only ....

John Mellor-Crummey. Compile-time support for efficient data race detection in shared-memory parallel programs. In Proc. ACM/ONR Workshop on Parallel and Distributed Debugging, pages 129--139, 1993.

....of parallelism are supported, this memory overhead seems quite manageable, making the technique more attractive to race detection. The execution time overhead is still very high, because every reference monitored has to be logged and checked against the access history in a critical section. In [26] an order of magnitude increase in execution time of instrumented codes is reported for experiments on a sequential machine. Even after reducing the shadowed references through compile time analysis, the time expansion factor remains around 5. The 20 need for critical sections for the parallel ....

John Mellor-Crummey. Compile-time support for efficient data race detection in shared-memory parallel programs. In Proc. if the ACM/ONR Workshop on Parallel and Distributed Debugging, pages 129--139, San Diego, CA, May 1993.

....of parallelism are supported, this memory overhead seems quite manageable, making the technique more attractive to race detection. The execution time overhead is still very high, because every reference monitored has to be logged and checked against the access history in a critical section. In [30] an order of magnitude increase in execution time of instrumented codes is reported for experiments on a sequential machine. Even after reducing the shadowed references through compile time analysis, the time expansion factor remains around 5. The need for critical sections for the parallel ....

John Mellor-Crummey. Compile-time support for efficient data race detection in shared-memory parallel programs. In Proc. if the ACM/ONR Workshop on Parallel and Distributed Debugging, pages 129--139, San Diego, CA, May 1993.

....first at a high level and then at a level low enough to reveal the main performance critical implementation techniques. Finally, we describe the experience we have had using Eraser with a number of multithreaded programs. Eraser bears no relationship to the tool by the same name constructed by Mellor Crummey [1993] for detecting data races in shared memory parallel Fortran programs as part of the ParaScope Programming Environment. 1.1 Definitions A lock is a simple synchronization object used for mutual exclusion; it is either available,orowned by a thread. The operations on a lock mu are lock(mu) and ....

MELLOR-CRUMMEY, J. 1993. Compile-time support for efficient data race detection in shared-memory parallel programs. In Proceedings of the ACM/ONR Workshop on Parallel and Distributed Debugging (San Diego, Calif., May). ACM, New York, 129--139.

.... the memory system is providing sequential consistency or a weaker correctness condition [11] work on detecting violations of sequential consistency within the memory system itself [14, 15] work on testing the serializability of database transactions [31] work on detecting data races (e.g. [2, 23, 28, 29]) work on proving that weak memory systems provide sequential consistency for programs that are free of data races (e.g. 1, 20, 21] work on testing uniprocessor memories [9] work on algorithms for testing data structures on uniprocessors (e.g. 10] work on verifying specific properties of ....

J. Mellor-Crummey, Compile-time support for efficient data race detection in shared-memory parallel programs, in Proc. 3rd ACM/ONR Workshop on Parallel and Distributed Debugging, May 1993, pp. 129--139.

....This paper presents the design and evaluation of an online data race detection technique for explicitly parallel shared memory applications. This technique is applicable for shared memory programs written for the lazy releaseconsistent (LRC) 9] memory model. Our work differs from previous work [3, 4, 5, 7, 16, 15] in that data race detection is performed both on the fly and without compiler support. In common with other dynamic systems, we address only the problem of detecting data races that occur in a given execution, not the more general problem of detecting all races allowed by program semantics [17] ....

John Mellor-Crummey. Compile-time support for efficient data race detection in shared-memory parallel programs. Technical Report CRPC-TR92232, Rice University, September 1992.

....detection work that needs to be done at run time. They also propose a new scheme for more efficient tagging and concurrency detection of blocks in programs with no nested parallelism. Mellor Crummey also presents an extension to the work with ParaScope that makes use of interprocedural analysis [MC93] Without the benefit of this analysis, it must be assumed that a procedure call can potentially modify any shared variable passed in as an actual parameter, as well as any global variable accessible to it. The analysis shows which parameters and variables are modified within procedures, as well ....

John Mellor-Crummey. Compile-time support for efficient data race detection in shared-memory parallel programs. In Proceedings of the ACM/ONR Workshop on Parallel and Distributed Debugging, pages 129--139, 1993.

....the procedure is always invoked inside an outer parallel loop. To behave correctly in the presence of different calling contexts, analysis can employ procedure cloning to replicate the procedure body and tailor it to the set of instrumented variables and OuterPar for a particular calling context [5, 15]. Alternatively, the procedure body can examine flags passed to it as parameters at run time to decide whether instrumentation is required for the current calling context. For expediency, our implementation currently uses the latter solution in this case. 5 Instrumentation Transformation ....

J. Mellor-Crummey. Compile-time support for efficient data race detection in shared-memory parallel programs. In Proceedings of ACM/ONR Workshop on Parallel and Distributed Debugging, pages 129--139, May 1993.

....programs, distinguishing determinacy races from atomicity races. They reference several algorithms for atomicity race detection, but we do not discuss this type of race detection here. Static analysis of parallel programs to uncover nondeterminacy has been studied extensively, for example, in [9, 13]. Various systems have been developed for determinacy race detection that do not allow nested parallelism, as for example [2] We now review related work on determinacy race detection for programs with nested parallelism. Nudler and Rudolph [16] give an English Hebrew labeling algorithm that ....

John Mellor-Crummey. Compile-time support for efficient data race detection in shared-memory parallel programs. In Proceedings of the ACM/ONR Workshop on Parallel and Distributed Debugging, pages 129--139, San Diego, California, May 1993. ACM Press.

....programs, distinguishing determinacy races from atomicity races. They reference several algorithms for atomicity race detection, but we do not discuss this type of race detection here. Static analysis of parallel programs to uncover nondeterminacy has been studied extensively, for example, in [9, 13]. Various systems have been developed for determinacyrace detection that do not allow nested parallelism, as for example [2] We now review related work on determinacy race detection for programs with nested parallelism. Nudler and Rudolph [16] give an English Hebrew labeling algorithm that ....

John Mellor-Crummey. Compile-time support for efficient data race detection in shared-memory parallel programs. In Proceedings of the ACM/ONR Workshop on Parallel and Distributed Debugging, pages 129--139, San Diego, California, May 1993. ACM Press.

....that is recorded. However, the work by Netzer [24] is for shared memory programs, while the work by Netzer and Miller [26] applies only to message passing programs with blocking sends and receives. Finally, many researchers have studied race conditions in parallel programs that use shared memory [1, 2, 9, 11, 14, 17, 22, 24, 25]. In this paper we describe an algorithm for detecting race conditions in parallel programs. It is a sequential algorithm that reads a trace of the communication events in a message passing parallel program and determines if the execution contains a race condition. The test is exact, so if the ....

J. Mellor-Crummey, "Compile-Time Support for Efficient Data Race Detection in Shared-Memory Parallel Programs", in Proc. ACM/ONR Workshop on Parallel and Distributed Debugging, pp. 129--139, 1993.

....the design and evaluation of an on the fly race detection technique for explicitly parallel shared memory applications. This technique is applicable to shared memory programs written for the lazy release consistent (LRC) 11] see Section 3.1) memory model. Our work differs from previous work [3, 4, 7, 9, 18, 17] in that data race detection is performed both on the fly and without compiler support. In common with other dynamic systems, we address only the problem of detecting data races that occur in a given execution, not the more general problem of detecting all races allowed by program semantics [19, ....

....long running applications. However, slowdowns would be even smaller if the first iteration were counted. Over the five applications, non optimized execution time slows only by an average factor of 3.8. This number compares quite favorably even with systems that exploit extensive compiler analysis [17, 7]. The last three columns are discussed in Section 5.2. Figure 5 breaks down the application slowdown into five categories (again, without the optimizations described in Section 4.5) CVM Mods is the overhead added by the modifications to CVM, primarily setting up the data structures necessary ....

John Mellor-Crummey. Compile-time support for efficient data race detection in shared-memory parallel programs. Technical Report CRPC-TR92232, Rice University, September 1992.

....Web automatically, allowing distributed systems to be restarted from their saved states, see these computations unfold, and follow the links to related computations. The idea of archiving states and replaying events has been employed previously in such contexts as data backup, compiler analysis (Mellor Crummey, 1992), and application debugging. Our contribution is that of exploring methods for, and The Caltech Infospheres Project is sponsored by the Air Force Office of Scientific Research under grant AFOSR F49620 94 1 0244, by the CISE directorate of the National Science Foundation under Problem Solving ....

Mellor-Crummey, J. (1992). Compile-time support for efficient data race detection in sharedmemory parallel programs. Technical report -- Center for Research on Parallel Computation CRPC-TR92232, Rice University.

....strategy can cause program execution time to increase by more than a factor of fifteen. The eraser data race instrumentation system was developed to identify accesses that could never participate in data races and therefore need not be instrumented and subject to run time monitoring [43]. Built on top of the ParaScope infrastructure, eraser uses ParaScope s data dependence analysis to identify pairs of references that might be involved in data races. In the absence of interprocedural information, however, each procedure must assume that it can be called from within a parallel ....

J. M. Mellor-Crummey. Compile-time support for efficient data race detection in shared-memory parallel programs. In Proc. ACM/ONR Workshop on Parallel and Distributed Debugging, San Diego, CA, May 1993.

No context found.

J. Mellor-Crummey, "Compile-time Support for Efficient Data Race Detection in Shared-Memory Parallel Programs," SIGPLAN Notices (ACM/ONR Workshop on Parallel and Distributed Debugging):129-39, 1993.

No context found.

Mellor-Crummey, J. Compile-time Support for Efficient Data Race Detection in Shared-Memory Parallel Programs. In Proceedings of the ACM/ONR Workshop on Parallel and Distributed Debugging, pages 129--139, May 1993.

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC