Language and compiler support for auto-tuning variable-accuracy algorithms

Abstract

Abstract-Approximating ideal program outputs is a common technique for solving computationally difficult problems, for adhering to processing or timing constraints, and for performance optimization in situations where perfect precision is not necessary. To this end, programmers often use approximation algorithms, iterative methods, data resampling, and other heuristics. However, programming such variable accuracy algorithms presents difficult challenges since the optimal algorithms and parameters may change with different accuracy requirements and usage environments. This problem is further compounded when multiple variable accuracy algorithms are nested together due to the complex way that accuracy requirements can propagate across algorithms and because of the size of the set of allowable compositions. As a result, programmers often deal with this issue in an ad-hoc manner that can sometimes violate sound programming practices such as maintaining library abstractions.In this paper, we propose language extensions that expose trade-offs between time and accuracy to the compiler. The compiler performs fully automatic compile-time and installtime autotuning and analyses in order to construct optimized algorithms to achieve any given target accuracy. We present novel compiler techniques and a structured genetic tuning algorithm to search the space of candidate algorithms and accuracies in the presence of recursion and sub-calls to other variable accuracy code. These techniques benefit both the library writer, by providing an easy way to describe and search the parameter and algorithmic choice space, and the library user, by allowing high level specification of accuracy requirements which are then met automatically without the need for the user to understand any algorithm-specific parameters. Additionally, we present a new suite of benchmarks, written in our language, to examine the efficacy of our techniques. Our experimental results show that by relaxing accuracy requirements, we can easily obtain performance improvements ranging from 1.1x to orders of magnitude of speedup.

Portable performance on heterogeneous architectures

“…The autotuner must produce an algorithm which meets a given accuracy target. These variable accuracy features are described in more detail in [4].…”

Autotuning algorithmic choice for input sensitivity

“…The autotuner must then consider a two dimensional objective space, where its first objective is to meet the accuracy target (with a given level of confidence) and the second objective is to maximize performance. A detailed description of the variable accuracy features of PetaBricks is given in [5]. Figure 3 describes the usage of our system for input sensitive algorithm design.…”

CoCo: Sound and Adaptive Replacement of Java Collections

“…Function recalculateIndex is shown in lines 35-46. Next, we add this incoming object x into the active container Γ i (lines [4][5][6][7][8][9][10][11][12][13][14][15]. If the element at newIndex is an abstract element (line 6), this abstraction is split into two separate abstractions (line 9) and then x is inserted between them (line 10).…”

“…After inserting a concrete element o into the list (line 5), method add$CoCo adds an abstraction of o into all inactive lists (lines [6][7][8]. Once an abstraction is found in a retrieval (lines [12][13][14][15][16][17], it is concretized to get an array of concrete elements it represents (line 14), which are then inserted into the list to replace this abstraction.…”