Runtime Optimization of Application Level Communication Patterns

Abstract

Abstract-This paper introduces the Abstract Data and Communication Library (ADCL). ADCL is an application level communication library aiming at providing the highest possible performance for application level communication operations on a given execution environment. The library provides for each communication pattern a large number of implementations and incorporates a runtime selection logic in order to choose the implementation leading to the highest performance of the application on the current platform. Two different runtime selection algorithms are currently available within ADCL: the library can either apply a brute force search strategy which tests all available implementations of a given communication pattern; alternatively, a heuristic relying on attributes characterizing an implementation has been developed in order to speed up the runtime decision procedure. The paper also evaluates the performance of a finite difference code using ADCL on an AMD Opteron cluster using InfiniBand and Gigabit Ethernet interconnects.

Outlier detection in performance data of parallel applications

“…The different implementations of the neighborhood communication provided by ADCL differ in the number of simultaneous communication partners for each process (e.g., pair-wise or all-to-all), the data transfer primitives used (e.g., synchronous, asynchronous, one-sided communication) and the handling of non-contiguous data (e.g., derived data types, pack/unpack). As of now, ADCL provides twelve different implementations for the neighborhood communication which are detailed in [4]. The execution time of each individual instance of the neighborhood communication is stored in an output file on a per process basis.…”

“…The library provides for each communication pattern a large number of implementations and incorporates a runtime selection logic in order to choose the implementation leading to the highest performance. Two different runtime selection algorithms are currently available within ADCL: the library can either apply a brute force search strategy which tests all available implementations of a given communication pattern or, alternatively, a heuristic relying on attributes characterizing an implementation which has been developed in order to speed up the runtime decision procedure [4]. The ADCL API offers high level interfaces of application level collective operations.…”

“…This procedure based on a heuristic was first presented in a simpler form in [4]. There it was shown that with the right choice of two parameters, namely a bound b, which defines a measurement to be an outlier, and the maximum number of accepted outliers nmax o , satisfactory results can be obtained.…”

“…message sizes depending on input data). To overcome the limitations of statically tuned software, we have recently introduced the Abstract Data and Communication Library (ADCL) [4]. The main goals of ADCL are threefold: firstly to define higher level abstractions for often occurring application level communication patterns, secondly to provide a large number of implementations for each communication pattern and thirdly to choose at runtime the implementation giving the best performance.…”

Towards performance portability through runtime adaptation for high‐performance computing applications

“…While this approach guarantees to find the best performing implementation on a given platform, it has the drawback that testing all implementations can take a significant amount of time. In order to speed up the selection logic, an alternative runtime heuristic based on attributes characterizing an implementation has been developed [7]. The heuristic is based on the assumption that the fastest implementation for a given problem size on a given execution environment is also the implementation having 'optimal' values for the attributes.…”