Clocked References in X10

1. INTRODUCTION 2. BACKGROUND 2.1 The X10 Programming Language X10 is a programming language that is currently under development at IBM Research. It is a strongly typed, concurrent , imperative, and object-oriented programming language. Of those descriptors, concurrent is the most important , as X10 was designed with multi-core and clustered systems in mind [4, 3]. The goal of X10 is to allow programmers to easily produce code that can be distributed over multiple cores and/or machines, with good scalability [2]. This is becoming more and more relevant as manufacturers turn towards improving processors by increasing the number of cores, rather than raw speed. In addition, computers in general are becoming cheaper and available in larger numbers , and we begin to see that being able to take advantage of multi-core processors and clusters is a more and more attractive idea [7]. X10 is designed specifically to do this, and to make it easy, it contains several language constructs that allow programmers to readily, and easily, write concurrent code [3]. Places, which can be thought of as analogous to processes, provide a shared memory environment in which concurrent code can be executed. This memory is not shared between Places, which allows Places to be parallelised, and distributed across multiple machines. Within Places, concurrent execution is achieved by the use of Activities, which are analogous to Threads. (The specifics of these constructs will be dealt with in the next section). X10 is built primarily as an extension to Java, using Poly-glot to handle the translation from X10 source code to Java source code. X10 can also be compiled to C++ source code. The X10 Runtime is written primarily in X10 itself, giving it the ability to be compiled into one of several back-ends. Currently , there is a Java-based runtime environment (using the X10 Runtime as libraries for the JVM), a C++ based run-time environment, and a CUDA (Compute Unified Device Architecture—a parallel computing architecture developed by Nvidia, that can be executed on GPUs) runtime environment. This project deals solely with the Java back-end for X10, but could be easily applied to other back-ends, as the modifications were not implementation specific. X10 source code bears a similarity to Java source code; Figure 1 shows a short source file that performs a Monte Carlo simulation to estimate a value for Pi. Since X10 is an class-based object-oriented language, it has the concept …