This paper presents the results of our investigation of code positioning techniques using execution profile data as input into the compilation process. The primary objective of the positioning is to reduce the overhead of the instruction memory hierarchy.

In recent years, many architectural design efforts have focused on maximizing performance for frequently executed, simple instructions. The authors describe how a small set of primitive instructions combined with what is considered careful frequency analysis and clever programming allows the Hewlett-Packard (HP) Precision Architecture integer multiplication and division implementation to provide adequate performance at little or no hardware cost.>

In recent years, many architectural design efforts have focused on maximizing performance for frequently executed, simple instructions. Although these efforts have resulted in machines with better average price/performance ratios, certain complex instructions and, thus, certain classes of programs which heavily depend on these instructions may suffer by comparison. Integer multiplication and division are one such set of complex instructions. This paper describes how a small set of primitive instructions combined with careful frequency analysis and clever programming allows the Hewlett-Packard Precision Architecture integer multiplication and division implementation to provide adequate performance at little or no hardware cost.

The intrinsic dimensionality of a set of patterns is important in determining an appropriate number of features for representing the data and whether a reasonable two- or three-dimensional representation of the data exists. We propose an intuitively appealing, noniterative estimator for intrinsic dimensionality which is based on nearneighbor information. We give plausible arguments supporting the consistency of this estimator. The method works well in identifying the true dimensionality for a variety of artificial data sets and is fairly insensitive to the number of samples and to the algorithmic parameters. Comparisons between this new method and the global eigenvalue approach demonstrate the utility of our estimator.