From 621fc44fa70dbb89da98b82156257cf9b0b1162e Mon Sep 17 00:00:00 2001 From: mbty <mbty@protonmail.com> Date: Tue, 18 Jun 2024 23:45:09 +0200 Subject: [PATCH] remove unreferenced chapter --- irr/irr.tex | 36 ------------------------------------ 1 file changed, 36 deletions(-) delete mode 100644 irr/irr.tex diff --git a/irr/irr.tex b/irr/irr.tex deleted file mode 100644 index c27a7d9..0000000 --- a/irr/irr.tex +++ /dev/null @@ -1,36 +0,0 @@ -\clearemptydoublepage -\chapter{IRR: an Intermediate Representation for Reasoning} -\label{ch:irr} - -This chapter motivates and discusses the methodology used for verifying hardware -throughout this thesis. - -In a few words, it is about turning descriptions of hardware written in a -high-level HDL to a lower-level, explicit form. All formal reasoning, be it -manual or automatic, is done on terms of this generic form, which we call an IRR -(Intermediate Representation for Reasoning). - -There are two main reasons why we do this. First, remember that we are working -from within the Coq proof assistant. Large workloads such as descriptions of -processors really put a strain on current generation proof assistants. As we -shall see, there are some ways of getting around this issue, but none of them -are convenient. It turns out that keeping everything explicit and staying clear -of fancy features really does help with performance. Note that this is more of -an engineering problem more than a fundamental issue with formal methods. The -second reason is that such low-level, explicit forms are very close to what -powerful proof search automation tools such as SMT solvers take as input. One of -the main issues with formal methods is the cost that they incur. Bringing -automation tools in is a good way to bridge the complexity gap between verified -and unverified developments. - -% This chapter starts with a discussion of the performance profile of proof -% assistants in section~\ref{sec:perf}. Section~\ref{sec:hw_verif_perf} follows -% with a discussion of how this is relevant to the problem of processor -% verification and compares approaches for dealing with the unique challenges of -% this field. The approach followed throughout this thesis is presented in greater -% detail in section~\ref{sec:irr_detail}. Finally, -% section~\ref{sec:irr_extensions} closes this chapter by mentioning some -% possible extensions of this approach. - -\section{Generalizations of IRR} -\label{sec:irr_extensions} -- GitLab