This paper compares hash-based approaches derived from the classic local algorithm

This paper presents a system that provides code generation at load-time and continuous program optimization at run-time. First, the architecture of the system is presented. Then, two optimization techniques are discussed that were developed specifically in the context of continuous optimization. The first of these optimizations continually adjusts the storage layouts of dynamic data structures to maximize data cache locality, while the second performs profile-driven instruction re-scheduling to increase instruction-level parallelism. These two optimizations have very di#erent cost/benefit ratios, presented in a series of benchmarks. The paper concludes with an outlook to future research directions and an enumeration of some remaining research problems. The empirical results presented in this paper make a case in favor of continuous optimization, but indicate that it needs to be applied judiciously. In many situations, the costs of dynamic optimizations outweigh their benefit, so that no break-even point is ever reached. In favorable circumstances, on the other hand, speed-ups of over 120% have been observed. It appears as if the main beneficiaries of continuous optimization are shared libraries, which at di#erent times can be optimized in the context of the currently dominant client application.

Dataflow analyses can have mutually beneficial interactions. Previous e#orts to exploit these interactions have either (1) iteratively performed each individual analysis until no further improvements are discovered or (2) developed "superanalyses " that manually combine conceptually separate analyses. We have devised a new approach that allows analyses to be defined independently while still enabling them to be combined automatically and profitably. Our approach avoids the loss of precision associated with iterating individual analyses and the implementation di#culties of manually writing a super-analysis. The key to our approach is a novel method of implicit communication between the individual components of a super-analysis based on graph transformations. In this paper, we precisely define our approach; we demonstrate that it is sound and it terminates; finally we give experimental results showing that in practice (1) our framework produces results at least as precise as iterating the individual analyses while compiling at least 5 times faster, and (2) our framework achieves the same precision as a manually written super-analysis while incurring a compiletime overhead of less than 20%.

A pointer-analysis algorithm can be either flow-sensitive or flow-insensitive. While flow-sensitive analysis usually provides more precise information, it is also usually considerably more costly in terms of time and space. The main contribution of this paper is the presentation of another option in the form of an algorithm that can be `tuned' to provide a range of results that fall between the results of flow-insensitive and flow-sensitive analysis. The algorithm combines a flow-insensitive pointer analysis with static single assignment (SSA) form and uses an iterative process to obtain progressively better results. 1 Introduction Having information about what pointer variables may point to is very useful (and often necessary) when performing many kinds of program analyses. Obviously, the better (or more precise) the information, the more useful the information is. A points-to analysis that takes into account the order in which statements may be executed (i.e., a flow-sensitive anal...

The value dependence graph (VDG) is a sparse dataflow-like representation that simplifies program analysis and transformation. It is a functional representation that represents control flow as data flow and makes explicit all machine quantities, such as stores and I/O channels. We are developing a compiler that builds a VDG representing a program, analyzes and transforms the VDG, then produces a control flow graph (CFG) [ASU86] from the optimized VDG. This framework simplifies transformations and improves upon several published results. For example, it enables more powerful code motion than [CLZ86, FOW87], eliminates as many redundancies as [AWZ88, RWZ88] (except for redundant loops), and provides important information to the code scheduler [BR91]. We exhibit a one-pass method for elimination of partial redundancies that never performs redundant code motion [KRS92, DS93] and is simpler than the classical [MR79, Dha91] or SSA [RWZ88] methods. These results accrue from eliminating the CFG from the analysis/transformation phases and using demand dependences in preference to control dependences.

When applying optimizations, a number of decisions are made using fixed strategies, such as always applying an optimization if it is applicable, applying optimizations in a fixed order and assuming a fixed configuration for optimizations such as tile size and loop unrolling factor. While it is widely recognized that these fixed strategies may not be the most appropriate for producing high quality code, especially for embedded systems, there are no general and automatic strategies that do otherwise. In this paper, we present a framework that enables these decisions to be made based on predicting the impact of an optimization, taking into account resources and code context. The framework consists of optimization models, code models and resource models, which are integrated for predicting the impact of applying optimizations. Because data cache performance is important to embedded codes, we focus on cache performance and present an instance of the framework for cache performance in this paper. Since most opportunities for cache improvement come from loop optimizations, we describe code, optimization and cache models tailored to predict the impact of applying loop optimizations for data locality. Experimentally we demonstrate the need to selectively apply optimizations and show the performance benefit of our framework in predicting when to apply an optimization. We also show that our framework can be used to choose the most beneficial optimization when a number of optimizations can be applied to a loop nest. And lastly, we show that we can use the framework to combine optimizations on a loop nest.

This paper presents a new al gS ithm for operator strengM reduction, called OSR. OSR improves upon an earlier alg orithm due to Allen, Cocke, and Kennedy [Allen et al. 1981]. OSR operates on the static sing e assig4 ent (SSA) form of a procedure [Cytron et al. 1991]. By taking advantag of the properties of SSA form, we have derived an alg--- ithm that is simple to understand, quick to implement, and, in practice, fast to run. Its asymptotic complexity is, in the worst case, the same as the Allen, Cocke, and Kennedy al gS ithm (ACK). OSR achieves optimization results that are equivalent to those obtained with the ACK alg orithm. OSR has been implemented in several research and production compilers

Abstract not found

Abstract. We show how declarative diagnosis techniques can be extended to cope with verification of operational properties, such as computed answers, and of abstract properties, such as types and groundness dependencies. The extension is achieved by using a simple semantic framework, based on abstract interpretation. The resulting technique (abstract diagnosis) leads to elegant bottom-up and top-down verification methods, which do not require to determine the symptoms in advance, and which are effective in the case of abstract properties described by finite domains.

Software designers face many challenges when developing applications for embedded systems. A major challenge is meeting the conflicting constraints of speed, code density, and power consumption. Traditional optimizing compiler technology is usually of little help in addressing this challenge. To meet speed, power, and size constraints, application developers typically resort to hand-coded assembly language. The results are software systems that are not portable, less robust, and more costly to develop and maintain. This paper describes a new code improvement paradigm implemented in a system called vista that can help achieve the cost/ performance trade-offs that embedded applications demand. Unlike traditional compilation systems where the smallest unit of compilation is typically a function and the programmer has no control over the code improvement process other than what types of code improvements to perform, the vista system opens the code improvement process and gives the application programmer, when necessary, the ability to finely control it. In particular, vista allows the application developer to (1) direct the order and scope in which the code improvement phases are applied, (2) manually specify code transformations, (3) undo previously applied transformations, and (4) view the low-level program representation graphically. vista can be used by embedded systems developers to produce applications, by compiler writers to prototype and debug new lowlevel code transformations, and by instructors to illustrate code transformations (e.g., in a compilers course).