Abstract GPU programs are widely used in industry. To obtain the best performance, a typical development process involves the manual or semi-automatic application of optimizations prior to compiling the code. Such optimizations can introduce errors. To avoid the introduction of errors, we can augment GPU programs with (pre- and postcondition-style) annotations to capture functional properties. However, keeping these annotations correct when optimizing GPU programs is labor-intensive and error-prone. This paper presents an approach to automatically apply optimizations to GPU programs while preserving provability by defining annotation-aware transformations . It applies frequently-used GPU optimizations, but besides transforming code, it also transforms the annotations. The approach has been implemented in the Alpinist tool and we evaluate Alpinist in combination with the VerCors program verifier, to automatically apply optimizations to a collection of verified programs and reverify them.

Abstract This paper gives an overview of the most recent developments on the VerCors verifier. VerCors is a deductive verifier for concurrent software, written in multiple programming languages, where the specifications are written in terms of pre-/postcondition contracts using permission-based separation logic. In essence, VerCors is a program transformation tool: it translates an annotated program into input for the Viper framework, which is then used as verification back-end. The paper discusses the different programming languages and features for which VerCors provides verification support. It also discusses how the tool internally has been reorganised to become easily extendible, and to improve the connection and interaction with Viper. In addition, we also introduce two tools built on top of VerCors, which support correctness-preserving transformations of verified programs. Finally, we discuss how the VerCors verifier has been used on a range of realistic case studies.

Abstract Over the last years, deductive program verifiers have substantially improved, and their applicability on non-trivial applications has been demonstrated. However, a major bottleneck is that for every new programming language, a new deductive verifier has to be built. This paper describes the first steps in a project that aims to address this problem, by language-agnostic support for deductive verification: Rather than building a deductive program verifier for every programming language, we develop deductive program verification technology for a widely-used intermediate representation language (LLVM IR), such that we eventually get verification support for any language that can be compiled into the LLVM IR format. Concretely, this paper describes the design of VCLLVM, a prototype tool that adds LLVM IR as a supported language to the VerCors verifier. We discuss the challenges that have to be addressed to develop verification support for such a low-level language. Moreover, we also sketch how we envisage to build verification support for any specified source program that can be compiled into LLVM IR on top of VCLLVM.

Abstract GPU programs are widely used in industry. To obtain the best performance, a typical development process involves the manual or semi-automatic application of optimizations prior to compiling the code. To avoid the introduction of errors, we can augment GPU programs with (pre- and postcondition-style) annotations to capture functional properties. However, keeping these annotations correct when optimizing GPU programs is labor-intensive and error-prone. This paper introduces Alpinist, an annotation-aware GPU program optimizer. It applies frequently-used GPU optimizations, but besides transforming code, it also transforms the annotations. We evaluate Alpinist, in combination with the VerCors program verifier, to automatically optimize a collection of verified programs and reverify them.