
TL;DR
This paper establishes precise bounds for deadlock and serializability in parallel PV-model threads, showing that analyzing a finite subset of resources suffices to determine these properties for any number of parallel threads.
Contribution
It introduces sharp cut-off theorems for deadlock and serializability in PV-model parallel programs, providing practical criteria for their analysis.
Findings
Deadlock freedom in T^n for all n iff T^M is deadlock free, with M as sum of resource capacities.
Serializability in T^n for all n iff T^2 is serializable, for capacity 1.
Existence of local obstructions to serializability characterized by a finite subset of subprograms.
Abstract
We prove cut-off results for deadlocks and serializability of a -thread run in parallel with itself: For a thread which accesses a set of resources, each with a maximal capacity , the PV-program , where copies of are run in parallel, is deadlock free for all if and only if is deadlock free where . This is a sharp bound: For all and finite there is a thread using these resources such that has a deadlock, but does not for . Moreover, we prove a more general theorem: There are no deadlocks in if and only if there are no deadlocks in for any subset . For , is serializable for all if and…
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
