Efficient system-enforced deterministic parallelism

Aviram, Amittai; Weng, Shu-Chun; Hu, Sen; Ford, Bryan

doi:10.1145/2160718.2160742

Cited by 86 publications

(121 citation statements)

References 51 publications

(121 reference statements)

Supporting

Mentioning

121

Contrasting

Order By: Relevance

“…Deterministic operating systems Determinator [21] and dOS [22] enforce determinism at user and OS level by using page protection hardware. Instead of significant changes of an OS, DeTrans-lib is implemented at user level, can be used with any OS, ensures libc-level determinism and guarantees deterministic order of invoking system calls in an application.…”

Section: Related Workmentioning

confidence: 99%

Determinism at Standard-Library Level in TM-Based Applications

Smiljković

Ünsal

Cristal

et al. 2015

Int J Parallel Prog

View full text Add to dashboard Cite

Abstract. Deterministic execution of a multi-threaded application guarantees that threads access shared memory in the same order and the application gives the same output when it runs with the same input parameters. Determinism provides repeatability, which helps a programmer in testing and debugging. Additionally, Transactional Memory (TM) simplifies development of applications that use transactions (instead of locks) to synchronize accesses to shared memory. However, transactions that call standard library functions have to be serialized, and the serialization causes a deadlock when applications run with the deterministic systems proposed so far. In this paper, we present DeTrans-lib, the first standard C library that provides deterministic execution of TM-based applications at user and standard-library level. DeTrans-lib avoids deadlocks by performing transaction serialization in deterministic order. We evaluate DeTrans-lib using benchmarks that perform I/O operations.

show abstract

Section: Related Workmentioning

confidence: 99%

Determinism at Standard-Library Level in TM-Based Applications

Smiljković

Ünsal

Cristal

et al. 2015

Int J Parallel Prog

View full text Add to dashboard Cite

show abstract

“…Determinator [7] advocates a new, radical programming model that converts all races, including races on memory and other shared resources, into exceptions, to achieve pervasive determinism. This programming model is not designed to be backward-compatible.…”

Section: Related Workmentioning

confidence: 99%

Efficient deterministic multithreading through schedule relaxation

Cui

Gallagher

et al. 2011

Proceedings of the Twenty-Third ACM Symposium on Operating Systems Principles

108

100

View full text Add to dashboard Cite

Deterministic multithreading (DMT) eliminates many pernicious software problems caused by nondeterminism. It works by constraining a program to repeat the same thread interleavings, or schedules, when given same input. Despite much recent research, it remains an open challenge to build both deterministic and efficient DMT systems for general programs on commodity hardware. To deterministically resolve a data race, a DMT system must enforce a deterministic schedule of shared memory accesses, or memschedule, which can incur prohibitive overhead. By using schedules consisting only of synchronization operations, or sync-schedule, this overhead can be avoided. However, a sync-schedule is deterministic only for race-free programs, but most programs have races.Our key insight is that races tend to occur only within minor portions of an execution, and a dominant majority of the execution is still race-free. Thus, we can resort to a mem-schedule only for the "racy" portions and enforce a sync-schedule otherwise, combining the efficiency of sync-schedules and the determinism of memschedules. We call these combined schedules hybrid schedules.Based on this insight, we have built PEREGRINE, an efficient deterministic multithreading system. When a program first runs on an input, PEREGRINE records an execution trace. It then relaxes this trace into a hybrid schedule and reuses the schedule on future compatible inputs efficiently and deterministically. PEREGRINE further improves efficiency with two new techniques: determinismpreserving slicing to generalize a schedule to more inputs while preserving determinism, and schedule-guided simplification to precisely analyze a program according to a specific schedule. Our evaluation on a diverse set of programs shows that PEREGRINE is deterministic and efficient, and can frequently reuse schedules for half of the evaluated programs.

show abstract

“…Burckhardt et al defined isolation and revision types in C# to buffer and merge concurrent data changes in r-fork/r-join threads [6]. Determinator was developed as a new operating system that buffers processes and threads in private workspaces and terminates an execution if concurrent data writes are detected [3]. CoreDet ensured determinism in threaded execution using versioned memory and a deterministic commit protocol [4].…”

Section: Related Workmentioning

confidence: 99%

Access Annotation for Safe Program Parallelization

Ding

Liu

2013

Lecture Notes in Computer Science

View full text Add to dashboard Cite

Abstract. The safety of speculative parallelization depends on monitoring all program access to shared data. The problem is especially difficult in softwarebased solutions. Till now, automatic techniques use either program instrumentation, which can be costly, or virtual memory protection, which incurs false sharing. In addition, not all access requires monitoring. It is worth considering a manual approach in which programmers insert access annotations to reduce the cost and increase the precision of program monitoring.This paper presents an interface for access annotation and two techniques to check the correctness of user annotation, i.e. whether all parallel executions are properly monitored and guaranteed to produce the sequential result. It gives a quadratic-time algorithm to check the exponential number of parallel interleavings. The paper then uses the annotation interface to parallelize several programs with uncertain parallelism. It demonstrates the efficiency of program monitoring by a performance comparison with OpenMP, which does not monitor data access or guarantee safety.

show abstract

Efficient system-enforced deterministic parallelism

Cited by 86 publications

References 51 publications

Determinism at Standard-Library Level in TM-Based Applications

Determinism at Standard-Library Level in TM-Based Applications

Efficient deterministic multithreading through schedule relaxation

Access Annotation for Safe Program Parallelization

Contact Info

Product

Resources

About