This paper concerns the use of program slicing to perform a certain kind of program-specialization operation. The specialization operation that slicing performs is different from the specialization operations performed by algorithms for partial evaluation, supercompilation, bifurcation, and deforestation. In particular, we present an example in which the specialized program that we create via slicing could not be created as the result of applying partial evaluation, supercompilation, bifurcation, or deforestation to the original unspecialized program. Specialization via slicing also possesses an interesting property that partial evaluation, supercompilation, and bifurcation do not possess: The latter operations are somewhat limited in the sense that they support tailoring of existing software only according to the ways in which parameters of functions and procedures are used in a program. Because parameters to functions and procedures represent the range of usage patterns that the designer of a piece of software has anticipated, partial evaluation, supercompilation, and bifurcation support specialization only in ways that have already been “foreseen ” by the software’s author. In contrast, the specialization operation that slicing supports permits programs to be specialized in ways

This paper presents a principled approach for optimizing iterative (or recursive) programs. The approach formulates a loop body as a function f and a change operation \Phi, incrementalizes f with respect to \Phi, and adopts an incrementalized loop body to form a new loop that is more efficient. Three general optimizations are performed as part of the adoption; they systematically handle initializations, termination conditions, and final return values on exits of loops. These optimizations are either omitted, or done in implicit, limited, or ad hoc ways in previous methods. The new approach generalizes classical loop optimization techniques, notably strength reduction, in optimizing compilers, and it unifies and systematizes various optimization strategies in transformational programming. Such principled strength reduction performs drastic program efficiency improvement via incrementalization and appreciably reduces code size via associated optimizations. We give examples where this app...

This paper presents program analyses and transformations for strengthening invariants for the purpose of efficient computation. Finding the stronger invariants corresponds to discovering a general class of auxiliary information for any incremental computation problem. Combining the techniques with previous techniques for caching intermediate results, we obtain a systematic approach that transforms non-incremental programs into ecient incremental programs that use and maintain useful auxiliary information as well as useful intermediate results. The use of auxiliary information allows us to achieve a greater degree of incrementality than otherwise possible. Applications of the approach include strength reduction in optimizing compilers and finite differencing in transformational programming.

This paper describes the precise specification, design, analysis, implementation, and measurements of an efficient algorithm for solving regular tree grammar based constraints. The particular constraints are for dead-code elimination on recursive data, but the method used for the algorithm design and complexity analysis is general and applies to other program analysis problems as well. The method is centered around Paige's finite differencing, i.e., computing expensive set expressions incrementally, and allows the algorithm to be derived and analyzed formally and implemented easily. We study higherlevel transformations that make the derived algorithm concise and allow its complexity to be analyzed accurately. Although a rough analysis shows that the worst-case time complexity is cubic in program size, an accurate analysis shows that it is linear in the number of live program points and in other parameters, including mainly the arity of data constructors and the number of selector applications into whose arguments the value constructed at a program point might flow. These parameters explain the performance of the analysis in practice. Our implementation also runs two to ten times as fast as a previous implementation of an informally designed algorithm.

. Tupling transformation strategy can be used to merge loops together by combining recursive calls and also to eliminate redundant calls for a class of programs. In the latter case, this transformation can produce super-linear speedup. Existing works in deriving a safe and automatic tupling only apply to a very limited class of programs. In this paper, we present a novel parameter analysis, called synchronisation analysis, to solve the termination problem for tupling. With it, we can perform tupling on functions with multiple recursion and accumulative arguments without the risk of non-termination. This significantly widens the scope for tupling, and potentially enhances its usefulness. The analysis is shown to be of polynomial complexity; this makes tupling suitable as a compiler optimisation. 1 Introduction Source-to-source transformation can achieve global optimisation through specialisation for recursive functions. Two well-known techniques are partial evaluation [9] a...

Given a program f and an input change \Phi, we wish to obtain an incremental program that computes f(x \Phi y) efficiently by making use of the value of f(x), the intermediate results computed in computing f(x), and auxiliary information about x that can be inexpensively maintained. Obtaining such incremental programs is an essential part of the transformational-programming approach to software development and enhancement. This paper presents a systematic approach that discovers a general class of useful auxiliary information, combines it with useful intermediate results, and obtains an efficient incremental program that uses and maintains these intermediate results and auxiliary information. We give a number of examples from list processing, VLSI circuit design, image processing, etc. 1 Introduction Software engineering is the systematic approach to the development, operation, maintenance, and retirement of software [1]. The transformational-programming approach to software engine...

Uncertain data processing is critical in a wide range of applications such as scientific computation handling data with inevitableerrorsandfinancialdecisionmakingrelyingonhuman provided parameters. While increasingly studied in the area of databases, uncertain data processing is often carried out by software, and thus software based solutions are attractive. In particular, Monte Carlo (MC) methods execute software with many samples from the uncertain inputs and observe the statistical behavior of the output. In this paper, we propose a technique to improve the cost-effectiveness of MC methods. Assuming only part of the input is uncertain, the certain part of the input always leads to the same execution across multiple sample runs. We remove such redundancy by coalescing multiple sample runs in a single run. In the coalesced run, the program operates on a vector of values if uncertainty is present and a single value otherwise. We handle cases where control flow and pointers are uncertain. Our results show that we can speed up the execution time of 30 sample runs by an average factor of 2.3 without precision lost or by up to 3.4 with negligible precision lost.

We investigate factors affecting the performance of caching to speed up discrete event simulation. Walsh and Sirer have shown that a variant of function caching (staged simulation) can improve the performance of simulation in a networking application. However, the effectiveness of caching depends significantly on cache size, the cost of consulting the cache, the hit rate, and the cost of completing the computation in case of a cache miss. We hypothesize that adaptive techniques can be used to optimize caching parameters and demonstrate an adaptive scheme that decides whether to utilize caching depending on observed cache performance and event processing times. We focus on evaluating quantitative relationships, using our own caching implementation with the P-Hold synthetic workload application running on the GTW simulation kernel. Experiments show that as the cache size is increased, performance improves to a point, then degrades, and also that the adaptive technique can substantially improve speedup. 1

Memoization is a well-known optimization technique used to eliminate redundant calls for pure functions. If a call to a function f with argument v yields result r, a subsequent call to f with v can be immediately reduced to r without the need to re-evaluate f ’s body. Understanding memoization in the presence of concurrency and communication is significantly more challenging. For example, if f communicates with other threads, it is not sufficient to simply record its input/output behavior; we must also track inter-thread dependencies induced by these communication actions. Subsequent calls to f can be elided only if we can identify an interleaving of actions from these call-sites that lead to states in which these dependencies are satisfied. Similar issues arise if f spawns additional threads. In this paper, we consider the memoization problem for a higher-order concurrent language whose threads may communicate through synchronous message-based communication. To avoid the need to perform unbounded state space search that may be necessary to determine if all communication dependencies manifest in an earlier call can be satisfied in a later one, we introduce a weaker notion of memoization called partial memoization that gives implementations the freedom to avoid performing some part, if not all, of a previously memoized call. To validate the effectiveness of our ideas, we consider the benefits of memoization for reducing the overhead of recomputation for streaming and transactional applications executed on a multi-core machine. We show that on some workloads, memoization can lead to substantial performance improvements without incurring high memory costs. 1.

Memoization is a well-known optimization technique used to eliminate redundant calls for pure functions. If a call to a function f with argument v yields result r, a subsequent call to f with v can be immediately reduced to r without the need to re-evaluate f ’s body. Understanding memoization in the presence of concurrency and communication is significantly more challenging. For example, if f communicates with other threads, it is not sufficient to simply record its input/output behavior; we must also track inter-thread dependencies induced by these communication actions. Subsequent calls to f can be elided only if we can identify an interleaving of actions from these call-sites that lead to states in which these dependencies are satisfied. Similar issues arise if f spawns additional threads. In this paper, we consider the memoization problem for a higher-order concurrent language whose threads may communicate through synchronous message-based communication. To avoid the need to perform unbounded state space search that may be necessary to determine if all communication dependencies manifest in an earlier call can be satisfied in a later one, we introduce a weaker notion of memoization called partial memoization that gives implementations the freedom to avoid performing some part, if not all, of a previously memoized call. To validate the effectiveness of our ideas, we consider the benefits of memoization for reducing the overhead of recomputation for streaming, server-based, and transactional applications executed on a multi-core machine. We show that on a variety of workloads, memoization can lead to substantial performance improvements without incurring high memory costs.