A Static Analysis for Synthesizing Parametric Specifications of Dynamic Memory Consumption.

Abstract

We present a static analysis for computing a parametric upper-bound of the amount of memory dynamically allocated by (Java-like) imperative object-oriented programs. We propose a general procedure for synthesizing non-linear formulas which conservatively estimate the quantity of memory explicitly allocated by a method as a function of its parameters. We have implemented the procedure and evaluated it on several benchmarks. Experimental results produced exact estimations for most test cases, and quite precise approximations for many of the others. We also apply our technique to compute usage in the context of scoped memory and discuss some open issues.

Symbolic and Analytic Techniques for Resource Analysis of Java Bytecode

“…The methods of Clauss seem to have remained largely within the highperformance/parallel computing community (see [24,32] for example) until 2006, when Braberman et al [13] (and see also [12]) showed how to adapt these techniques to predict the memory usage of (iterative) Java programs; at present this appears to be the only application of polytope methods within the programming language community.…”

The Weight-Watcher Service and its Lightweight Implementation

“…In this paper, we consider resources as being the number of bytes that the code has dynamically allocated (Memory), the number of CPU cycles or bytecodes executed (CPU), the energy, in micro Joules, consumed during the execution (Energy) and the time spent executing the action 2 Meaning that resources consumed by native code are not accounted.…”

“…In [2], the estimation of the memory consumption of a method is given by a parametric function, where the parameters of the function are the actual method parameters. Let us formalize this function as f Memory ({x 1 , .…”

“…In order to provide resource consumption measurements following the execution of an action, we implemented a dynamic Profiler (by modifying the KVM [17] Java virtual machine) which gives perfect measurements, at the VM level, 2 for CPU, Memory, Network and Time, resulting in a slowdown as low as 6.46% -9.86% (see Section VI-A). In comparison, the slowdown induced by Java bytecode instrumentation techniques [1], lies on average around 20% [1] to 40% [10].…”

A Type System for Usage of Software Components

“…For the imperative paradigm, which is closer to the system described here, e.g. [18,19,20] treat memory usage. The problem of component usage in a parallel setting is related to prevention of deadlocks and race conditions.…”