In the previous post, we looked at different types and design objectives of high-level IRs used by ML compilers. Today, we are going to look at different optimizations that are performed using high-level IRs.These optimizations can be performed either offline or online. In online mode, the optimizations are done before performing the inference, while in offline mode, the runtime saves the optimized graph to disk. In this post, I have attempted to further extend Li et al.’s [1] classification of high-level IR optimizations into five categories to the best of my knowledge. Figure 1: Overview of Optimizations done on High-level…
Category: static analysis
A survey of concurrency bug patterns
Humans tend to make the same mistake again and again. We, software developers, are no different. Hello everyone, in this blog post, I’m going to talk about common concurrency bug patterns. Bug patterns give good insight into developer’s psychology, specifically, the assumptions we make about the programming languages, compilers, 3rd party code, OS scheduler, and hardware architecture.
Understanding Dependency Graphs for Program Analysis
Hi, In this blog post, I’ll briefly cover several types of dependency graphs used for program analysis, transformation, and representation. Following is the table of content: Prerequisites Reaching definition analysis Pointer analysis Dominator tree Call graphs Control flow graphs Control dependency graphs Data dependency graphs Program dependency graphs System dependency graphs Prerequisites Reaching definition analysis (RDA) RDA aims to determine the set of variable definitions (assignments) that may (or must) reach a given program point. Every program statement has an Entry set and Exit set associated with it, where the entry set is a set of all variable definitions reaching…
Static Program Analysis: Motivation and Techniques
Static analysis is a method to extract specific information about the program without actually running it. In this blog post, we will first have a look at different types of information that we can obtain using static analysis. Later, we will have a comprehensive overview of different techniques for static program analysis. Motivation There are broadly two categories of information, based on their usage, that we can extract: Proscriptive information and Descriptive information[1]. Proscriptive information is obtained when we statically analyze the program behavior against a set of prescribed rules. For instance, identification, localization, and explanation of all software anomalies…
Embedded Software Testing, Validation, and Verification
Embedded software, unlike general software, is highly dependent on the hardware they are developed for. Thus it became essential to create a hardware-based or hardware-like environment simultaneously with the software for effective testing. Where simulators and emulators can, up to some extent, provide a hardware-like environment for testing embedded software, they can not always be relied upon while testing for safety-critical applications like flight control system. It is this, high dependency on hardware, which makes testing, verification, and validation of embedded Softwares far more demanding than of general software. Moreover, several safety-critical embedded systems are also strictly real-time, or at…