MPI Application Development Using the Analysis Tool MARMOT

Abstract

Abstract. The Message Passing Interface (MPI) is widely used to write parallel programs using message passing. Due to the complexity of parallel programming there is a need for tools supporting the development process. There are many situations where incorrect usage of MPI by the application programmer can automatically be detected. Examples are the introduction of irreproducibility, deadlocks and incorrect management of resources like communicators, groups, datatypes and operators. We also describe the tool MARMOT that implements some of these tests. Finally we describe our experiences with three applications of the CrossGrid project regarding the usability and performance of this tool.

MPI Runtime Error Detection with MUST: Advanced Error Reports

“…This work directly relates to other runtime error detection approaches for MPI applications, which include Marmot [2], Umpire [3], ISP [11], MPI-Check [12], and Intel's approach [13]. While MUST is the successor of both Marmot and Umpire, the MPI-Check tool and Intel's approach use a timeout-based deadlock detection.…”

“…Runtime error detection, i.e., detecting errors during an application run, is one tool class that provides this support. We develop the Marmot Umpire Scalable Tool (MUST), named after its predecessor tools Marmot [2] and Umpire [3], for this purpose. Recent advances in runtime deadlock detection [4] and datatype correctness checks [5] allow MUST to efficiently detect complex errors.…”

MPI runtime error detection with MUST: Advances in deadlock detection

“…Our work is closely related to MPI runtime error detection tools such as ISP [3], MPI-Check [4], MPICH extension [5], Marmot [6], and Umpire [7]. Fig.…”

“…Thus, we restrict our approach to only consider deadlocks that manifest themselves in a given run. As a result, we use the AND⊕OR model [2] that enables a graph-based deadlock detection, to implement a runtime deadlock detection approach in the Marmot Umpire Scalable Tool (MUST), which extends and scales the functionality of its predecessors: Marmot [6] and Umpire [7].…”

Open Issues in MPI Implementation

“…A similar tool is described in [31]. Other tools also exist for checking program correctness, such as MARMOT [16], Umpire [32], and Intel Trace Analyzer and Collector [12], but more work is needed in this area. Parallel programs are also prone to suffer from deadlocks and race conditions that may remain undetected for a long time because they are timing dependent [17].…”