We introduce Kuiper, a language for safe and verified efficient CPU/GPU programming embedded as an extensible library within the F* dependently typed language. We rely on F*’s support for dependent types and its associated Pulse concurrent separation logic to develop a program logic in which to prove CPU/GPU programs safe, data-race free, and functionally correct. Our model of the GPU includes several intricacies, including the memory hierarchy, kernel launches, and synchronization within a single comprehensive framework. To do so, we extend the Pulse program logic with a novel notion of located resources and a new connective to structure reasoning about massively parallel programs, and present new proof rules to lift the per-thread view of GPU kernels to an end-to-end correctness specification. We have used Kuiper to program and prove correct a variety of GPU kernels, including full functional correctness proofs of an optimized matrix multiplication using two levels of block tiling and tensor cores. In doing so, we have developed a range of libraries to enable programs and proofs at a high level of abstraction but without imposing any runtime overhead. These allow Kuiper programs to be polymorphic (over types, operations, memory layout, and more) and compile to efficient, specialized CUDA code, while enabling a novel form of verified auto-tuning. Our experimental evaluation confirms that Kuiper programs match the performance of their handwritten CUDA counterparts and are competitive with closed source, state-of-the-art kernels in cuBLAS.

Equations are ubiquitous in mathematical reasoning. Often, however, they only hold under certain conditions. As these conditions are usually clear from context, mathematicians regularly omit them when performing equational reasoning on paper. In contrast, interactive theorem provers pedantically insist on every detail to be convinced that a theorem holds, hindering equational reasoning at the more abstract level of pen-andpaper mathematics. In this paper, we address this issue by raising the level of equational reasoning to enable pen-and-paper style in interactive theorem provers. We achieve this by interpreting theorems as conditional rewrite rules, and use equality saturation to automatically derive equational proofs. Conditions that cannot be automatically proven may be surfaced as proof obligations. Concretely, we present how to interpret theorems as conditional rewrite rules for a significant class of theorems. Handling these theorems goes beyond simple syntactic rewriting, and deals with aspects like propositional conditions and type classes. We evaluate our approach by implementing it as a tactic in Lean, using the egg library for equality saturation with e-graphs. We show four use cases demonstrating the efficacy of this higher level of abstraction for equational reasoning.

Traditionally, compiler researchers either conduct experiments within an existing production compiler or develop their own prototype compiler; both options come with trade-offs. On one hand, prototyping in a production compiler can be cumbersome, as they are often optimized for program compilation speed at the expense of software simplicity and development speed. On the other hand, the transition from a prototype compiler to production requires significant engineering work. To bridge this gap, we introduce the concept of sidekick compiler frameworks, an approach that uses multiple frameworks that interoperate with each other by leveraging textual interchange formats and declarative descriptions of abstractions. Each such compiler framework is specialized for specific use cases, such as performance or prototyping. Abstractions are by design shared across frameworks, simplifying the transition from prototyping to production. We demonstrate this idea with xDSL, a sidekick for MLIR focused on prototyping and teaching. xDSL interoperates with MLIR through a shared textual IR and the exchange of IRs through an IR Definition Language. The benefits of sidekick compiler frameworks are evaluated by showing on three use cases how xDSL impacts their development: teaching, DSL compilation, and rewrite system prototyping. We also investigate the trade-offs that xDSL offers, and demonstrate how we simplify the transition between frameworks using the IRDL dialect. With sidekick compilation, we envision a future in which engineers minimize the cost of development by choosing a framework built for their immediate needs, and later transitioning to production with minimal overhead.

To take full advantage of a specific hardware target, performance engineers need to gain control on compilers in order to leverage their domain knowledge about the program and hardware. Yet, modern compilers are poorly controlled, usually by configuring a sequence of coarse-grained monolithic black-box passes, or by means of predefined compiler annotations/pragmas. These can be effective, but often do not let users precisely optimize their varying compute loads. As a consequence, performance engineers have to resort to implementing custom passes for a specific optimization heuristic, requiring compiler engineering expert knowledge. In this paper, we present a technique that provides fine-grained control of general-purpose compilers by introducing the Transform dialect, a controllable IR-based transformation system implemented in MLIR. The Transform dialect empowers performance engineers to optimize their various compute loads by composing and reusing existing---but currently hidden---compiler features without the need to implement new passes or even rebuilding the compiler. We demonstrate in five case studies that the Transform dialect enables precise, safe composition of compiler transformations and allows for straightforward integration with state-of-the-art search methods.

While enabling use cases such as backtracking search and probabilistic programming, multiple resumptions have the reputation of being incompatible with efficient implementation techniques, such as stack switching. This paper sets out to resolve this conflict and thus bridge the gap between expressiveness and performance. To this end, we present a compilation strategy and runtime system for lexical effect handlers with support for multiple resumptions and stack-allocated mutable state. By building on garbage-free reference counting and associating stacks with stable prompts, our approach enables constant-time continuation capture and resumption when resumed exactly once, as well as constant-time state access. Nevertheless, we also support multiple resumptions by copying stacks when necessary. We practically evaluate our approach by implementing an LLVM backend for the Effekt language. A performance comparison with state-of-the-art systems, including dynamic and lexical effect handler implementations, suggests that our approach achieves competitive performance and the increased expressiveness only comes with limited overhead.

Equality saturation has gained significant interest as a powerful optimization and reasoning technique. At its heart is the e-graph data structure, that space-efficiently represents equal sub-terms uniquely. An important open problem in this context is extending this efficient representation to languages featuring (bound) variables. Independent of how we represent variables in e-graphs, either as names or nameless (using de Bruijn indices), sharing is broken as sub-terms that differ only in the names of their variables are represented separately. This results in aggressive e-graph growth, bad performance, as well as reduced expressiveness. In this paper, we present a novel approach to representing bound variables in e-graphs by making them a first-class built-in feature of the data structure. Our slotted e-graph represents terms that differ only by (bound or free) variable names uniquely. To do so, e-classes that represent equivalent terms via e-nodes are parameterized by slots , abstracting over free variables of the represented terms. Referring to an e-class from an e-node now requires relating the variables from its context to the slots of the e-class. Our evaluation of slotted e-graph uses two case studies from compiler optimization and theorem proving to show that performing equality saturation for languages with bound variables is greatly simplified and that we can solve practically relevant problems that cannot be solved with e-graphs using de Bruijn indices.

Domain Specific Languages (DSLs) increase programmer productivity and provide high performance. Their targeted abstractions allow scientists to express problems at a high level, providing rich details that optimizing compilers can exploit to target current- and next-generation supercomputers. The convenience and performance of DSLs come with significant development and maintenance costs. The siloed design of DSL compilers and the resulting inability to benefit from shared infrastructure cause uncertainties around longevity and the adoption of DSLs at scale. By tailoring the broadly-adopted MLIR compiler framework to HPC, we bring the same synergies that the machine learning community already exploits across their DSLs (e.g. Tensorflow, PyTorch) to the finite-difference stencil HPC community. We introduce new HPC-specific abstractions for message passing targeting distributed stencil computations. We demonstrate the sharing of common components across three distinct HPC stencil-DSL compilers: Devito, PSyclone, and the Open Earth Compiler, showing that our framework generates high-performance executables based upon a shared compiler ecosystem.

On behalf of the Program Committee of the International Symposium on Code Generation and Optimization (CGO), we are delighted to present the papers featured in the 2024 edition of the conference. This year marks a significant milestone in the history of CGO, as the symposium embraced a dual-round submission process for the first time. The initial round had a deadline set on May 19th, 2024, while the second round followed the traditional CGO deadline on September 1st, 2024. Although the program committee remained consistent throughout both rounds, each round had distinct program co-chairs overseeing coordination. The firstround coordinators were Jingling Xue and Michel Steuwer, while the second-round coordinators were Fernando Pereira and Guilherme Ottoni.

Modern programming languages provide programmers with rich abstractions for data collections as part of their standard libraries, e.g., Containers in the C++ STL, the Java Collections Framework, or the Scala Collections API. Typically, these collections frameworks are organised as hierarchies that provide programmers with common abstract data types (ADTs) like lists, queues, and stacks. While convenient, this approach introduces problems which ultimately affect application performance due to users over-specifying collection data types limiting implementation flexibility. In this article, we develop Collection Skeletons which provide a novel, declarative approach to data collections. Using our framework, programmers explicitly select properties for their collections, thereby truly decoupling specification from implementation. By making collection properties explicit, immediate benefits materialise in forms of reduced risk of over-specification and increased implementation flexibility. We have prototyped our declarative abstractions for collections as a C++ library, and demonstrate that benchmark applications rewritten to use Collection Skeletons incur little or no overhead. We also show how Collection Skeletons help shielding the application developer from parallel implementation details, either by encapsulating implicit parallelism or through explicit properties that capture the requirements of parallel algorithmic skeletons. We observe performance improvements across most of the 17 benchmarks resulting from the use of Collection Skeletons before trying to parallelise those benchmarks, while also enhancing performance portability across three different hardware platforms.

Rewriting is a principled term transformation technique with uses across theorem proving and compilation. In theorem proving, each rewrite is a proof step; in compilation, rewrites optimize a program term. While developing rewrite sequences manually is possible, this process does not scale to larger rewrite sequences. Automated rewriting techniques, like greedy simplification or equality saturation, work well without requiring human input. Yet, they do not scale to large search spaces, limiting the complexity of tasks where automated rewriting is effective, and meaning that just a small increase in term size or rewrite length may result in failure. This paper proposes a semi-automatic rewriting technique as a means to scale rewriting by allowing human insight at key decision points. Specifically, we propose guided equality saturation that embraces human guidance when fully automated equality saturation does not scale. The rewriting is split into two simpler automatic equality saturation steps: from the original term to a human-provided intermediate guide , and from the guide to the target. Complex rewriting tasks may require multiple guides, resulting in a sequence of equality saturation steps. A guide can be a complete term, or a sketch containing undefined elements that are instantiated by the equality saturation search. Such sketches may be far more concise than complete terms. We demonstrate the generality and effectiveness of guided equality saturation using two case studies. First, we integrate guided equality saturation in the Lean 4 proof assistant. Proofs are written in the style of textbook proof sketches, as a series of calculations omitting details and skipping steps. These proofs conclude in less than a second instead of minutes when compared to unguided equality saturation, and can find complex proofs that previously had to be done manually. Second, in the compiler of the RISE array language, where unguided equality saturation fails to perform optimizations within an hour and using 60 GB of memory, guided equality saturation performs the same optimizations with at most 3 guides, within seconds using less than 1 GB memory.

Graphics Processing Units (GPU) offer tremendous computational power by following a throughput oriented paradigm where many thousand computational units operate in parallel. Programming such massively parallel hardware is challenging. Programmers must correctly and efficiently coordinate thousands of threads and their accesses to various shared memory spaces. Existing mainstream GPU programming languages, such as CUDA and OpenCL, are based on C/C++ inheriting their fundamentally unsafe ways to access memory via raw pointers. This facilitates easy to make, but hard to detect bugs, such as data races and deadlocks . In this paper, we present Descend : a safe GPU programming language. In contrast to prior safe high-level GPU programming approaches, Descend is an imperative GPU systems programming language in the spirit of Rust, enforcing safe CPU and GPU memory management in the type system by tracking Ownership and Lifetimes . Descend introduces a new holistic GPU programming model where computations are hierarchically scheduled over the GPU’s execution resources : grid, blocks, warps, and threads. Descend’s extended Borrow checking ensures that execution resources safely access memory regions without data races. For this, we introduced views describing safe parallel access patterns of memory regions, as well as atomic variables. For memory accesses that can’t be checked by our type system, users can annotate limited code sections as unsafe . We discuss the memory safety guarantees offered by Descend and evaluate our implementation using multiple benchmarks, demonstrating that Descend is capable of expressing real-world GPU programs showing competitive performance compared to manually written CUDA programs lacking Descend’s safety guarantees.

Rewriting is a versatile and powerful technique used in many domains. Strategic rewriting allows programmers to control the application of rewrite rules by composing individual rewrite rules into complex rewrite strategies. These strategies are semantically complex, as they may be nondeterministic, they may raise errors that trigger backtracking, and they may not terminate. Given such semantic complexity, it is necessary to establish a formal understanding of rewrite strategies and to enable reasoning about them in order to answer questions like: How do we know that a rewrite strategy terminates? How do we know that a rewrite strategy does not fail because we compose two incompatible rewrites? How do we know that a desired property holds after applying a rewrite strategy? In this paper, we introduce Shoggoth: a formal foundation for understanding, analysing and reasoning about strategic rewriting that is capable of answering these questions. We provide a denotational semantics of System S, a core language for strategic rewriting, and prove its equivalence to our big-step operational semantics, which extends existing work by explicitly accounting for divergence. We further define a location-based weakest precondition calculus to enable formal reasoning about rewriting strategies, and we prove this calculus sound with respect to the denotational semantics. We show how this calculus can be used in practice to reason about properties of rewriting strategies, including termination, that they are well-composed, and that desired postconditions hold. The semantics and calculus are formalised in Isabelle/HOL and all proofs are mechanised.

We introduce the Bayesian Compiler Optimization framework (BaCO), a general purpose autotuner for modern compilers targeting CPUs, GPUs, and FPGAs. BaCO provides the flexibility needed to handle the requirements of modern autotuning tasks. Particularly, it deals with permutation, ordered, and continuous parameter types along with both known and unknown parameter constraints. To reason about these parameter types and efficiently deliver high-quality code, BaCO uses Bayesian optimization algorithms specialized towards the autotuning domain. We demonstrate BaCO's effectiveness on three modern compiler systems: TACO, RISE & ELEVATE, and HPVM2FPGA for CPUs, GPUs, and FPGAs respectively. For these domains, BaCO outperforms current state-of-the-art auto-tuners by delivering on average 1.36X--1.56X faster code with a tiny search budget, and BaCO is able to reach expert-level performance 2.9X--3.9X faster.

Structural subtyping and parametric polymorphism provide similar flexibility and reusability to programmers. For example, both features enable the programmer to provide a wider record as an argument to a function that expects a narrower one. However, the means by which they do so differs substantially, and the precise details of the relationship between them exists, at best, as folklore in literature. In this paper, we systematically study the relative expressive power of structural subtyping and parametric polymorphism. We focus our investigation on establishing the extent to which parametric polymorphism, in the form of row and presence polymorphism, can encode structural subtyping for variant and record types. We base our study on various Church-style λ-calculi extended with records and variants, different forms of structural subtyping, and row and presence polymorphism. We characterise expressiveness by exhibiting compositional translations between calculi. For each translation we prove a type preservation and operational correspondence result. We also prove a number of non-existence results. By imposing restrictions on both source and target types, we reveal further subtleties in the expressiveness landscape, the restrictions enabling otherwise impossible translations to be defined. More specifically, we prove that full subtyping cannot be encoded via polymorphism, but we show that several restricted forms of subtyping can be encoded via particular forms of polymorphism.

Inlining is a widely studied compiler optimization that is particularly important for functional languages such as Haskell and OCaml. The Glasgow Haskell Compiler (GHC) inliner is a heuristic of such complexity, however, that it has not significantly changed for nearly 20 years. It heavily relies on hard-coded numeric constants, or magic numbers, based on out-of-date intuition. Dissatisfaction with inlining performance has led to the widespread use of inlining pragmas by programmers.

Modern programming languages provide programmers with rich abstractions for data collections as part of their standard libraries, e.g. Containers in the C++ STL, the Java Collections Framework, or the Scala Collections API. Typically, these collections frameworks are organised as hierarchies that provide programmers with common abstract data types (ADTs) like lists, queues, and stacks. While convenient, this approach introduces problems which ultimately affect application performance due to users over-specifying collection data types limiting implementation flexibility. In this paper, we develop Collection Skeletons which provide a novel, declarative approach to data collections. Using our framework, programmers explicitly select properties for their collections, thereby truly decoupling specification from implementation. By making collection properties explicit immediate benefits materialise in form of reduced risk of over-specification and increased implementation flexibility. We have prototyped our declarative abstractions for collections as a C++ library, and demonstrate that benchmark applications rewritten to use Collection Skeletons incur little or no overhead. In fact, for several benchmarks, we observe performance speedups (on average between 2.57 to 2.93, and up to 16.37) and also enhanced performance portability across three different hardware platforms.

The continued specialization in hardware and software due to the end of Moore's law forces us to question fundamental design choices in compilers, and in particular for domain specific languages. The days where a single universal compiler intermediate representation (IR) was sufficient to perform all important optimizations are over. We need novel IRs and ways for them to interact with one another while leveraging established compiler infrastructures.

Halide and many similar projects have demonstrated the great potential of domain specific optimizing compilers. They enable programs to be expressed at a convenient high-level, while generating high-performance code for parallel architectures. As domains of interest expand towards deep learning, probabilistic programming and beyond, it becomes increasingly clear that it is unsustainable to redesign domain specific compilers for each new domain. In addition, the rapid growth of hardware architectures to optimize for poses great challenges for designing these compilers. In this paper, we show how to extend a unifying domain-extensible compiler with domain-specific as well as hardware-specific optimizations. The compiler operates on generic patterns that have proven flexible enough to express a wide range of computations. Optimizations are not hard-coded into the compiler but are expressed as user-defined rewrite rules that are composed into strategies controlling the optimization process. Crucially, both computational patterns and optimization strategies are extensible without modifying the core compiler implementation. We demonstrate that this domain-extensible compiler design is capable of expressing image processing pipelines and well-known image processing optimizations. Our results on four mobile ARM multi-core CPUs, often used for image processing tasks, show that the code generated for the Harris operator outperforms the image processing library OpenCV by up to 16× and achieves performance close to - or even up to 1.4 × better than - the state-of-the-art image processing compiler Halide.

CGO 2021 included two separate categories of papers—main conference papers and tool & practical experience papers—as it did in the previous year. However, this year the acceptance criterion of tool papers required the validation of the tool from the Artifact Evaluation Committee.

The software and hardware landscape of high performance computing is expanding faster than computational scientists can take advantage of new frameworks and platforms. In an ideal world, simulation codes would be written once in a high-level manner and achieve high-performance anywhere, but the reality is more complicated. Currently, high-level solutions lack support for sophisticated physical models across different parallel backends. Existing solutions with appropriate support are low-level and, therefore, tied to a specific hardware target. We present an approach that tackles this problem with a modularized separation of concerns: a middle layer separates the management of generating low-level optimized code from a high-level programmable layer. In this paper, we describe how our contributions to this hardware-agnostic, middle-layer language provide functionality for complex room acoustics simulations, a type of Finite Difference Time Domain (FDTD) simulation using stencils which is representative of many other 3D wave models. We show that we are able to develop performance-portable codes for these types of models which leads to performance on par with tuned hand-written implementations. Furthermore, we show how this approach is used to develop both host and device side code for multi-kernel applications, as is required for room acoustics simulations with complex boundaries.

Auto-tuning is a popular approach to program optimization: it automatically finds good configurations of a program’s so-called tuning parameters whose values are crucial for achieving high performance for a particular parallel architecture and characteristics of input/output data. We present three new contributions of the Auto-Tuning Framework (ATF), which enable a key advantage in general-purpose auto-tuning : efficiently optimizing programs whose tuning parameters have interdependencies among them. We make the following contributions to the three main phases of general-purpose auto-tuning: (1) ATF generates the search space of interdependent tuning parameters with high performance by efficiently exploiting parameter constraints; (2) ATF stores such search spaces efficiently in memory, based on a novel chain-of-trees search space structure; (3) ATF explores these search spaces faster, by employing a multi-dimensional search strategy on its chain-of-trees search space representation. Our experiments demonstrate that, compared to the state-of-the-art, general-purpose auto-tuning frameworks, ATF substantially improves generating, storing, and exploring the search space of interdependent tuning parameters, thereby enabling an efficient overall auto-tuning process for important applications from popular domains, including stencil computations, linear algebra routines, quantum chemistry computations, and data mining algorithms.

Usage of high-level intermediate representations promises the generation of fast code from a high-level description, improving the productivity of developers while achieving the performance traditionally only reached with low-level programming approaches.

Python is a popular, dynamic language for data science and scientific computing. To ensure efficiency, significant numerical libraries are implemented in static native languages. However, performance suffers when switching between native and non-native code, especially if data has to be converted between native arrays and Python data structures. As GPU accelerators are increasingly used, this problem becomes particularly acute. Data and control has to be repeatedly transferred between the accelerator and the host.

Optimizing programs to run efficiently on modern parallel hardware is hard but crucial for many applications. The predominantly used imperative languages - like C or OpenCL - force the programmer to intertwine the code describing functionality and optimizations. This results in a portability nightmare that is particularly problematic given the accelerating trend towards specialized hardware devices to further increase efficiency. Many emerging DSLs used in performance demanding domains such as deep learning or high-performance image processing attempt to simplify or even fully automate the optimization process. Using a high-level - often functional - language, programmers focus on describing functionality in a declarative way. In some systems such as Halide or TVM, a separate schedule specifies how the program should be optimized. Unfortunately, these schedules are not written in well-defined programming languages. Instead, they are implemented as a set of ad-hoc predefined APIs that the compiler writers have exposed. In this functional pearl, we show how to employ functional programming techniques to solve this challenge with elegance. We present two functional languages that work together - each addressing a separate concern. RISE is a functional language for expressing computations using well known functional data-parallel patterns. ELEVATE is a functional language for describing optimization strategies. A high-level RISE program is transformed into a low-level form using optimization strategies written in ELEVATE . From the rewritten low-level program high-performance parallel code is automatically generated. In contrast to existing high-performance domain-specific systems with scheduling APIs, in our approach programmers are not restricted to a set of built-in operations and optimizations but freely define their own computational patterns in RISE and optimization strategies in ELEVATE in a composable and reusable way. We show how our holistic functional approach achieves competitive performance with the state-of-the-art imperative systems Halide and TVM.

Stencil computations are a widely used type of algorithm, found in applications from physical simulations to machine learning. Stencils are embarrassingly parallel, therefore fit on modern hardware such as Graphic Processing Units perfectly. Although stencil computations have been extensively studied, optimizing them for increasingly diverse hardware remains challenging. Domain-specific Languages (DSLs) have raised the programming abstraction and offer good performance; however, this method places the burden on DSL implementers to write almost full-fledged parallelizing compilers and optimizers. Lift has recently emerged as a promising approach to achieve performance portability by using a small set of reusable parallel primitives that DSL or library writers utilize. L ift ’s key novelty is in its encoding of optimizations as a system of extensible rewrite rules which are used to explore the optimization space. This article demonstrates how complex multi-dimensional stencil code and optimizations are expressed using compositions of simple 1D L ift primitives and rewrite rules. We introduce two optimizations that provide high performance for stencils in particular: classical overlapped tiling for multi-dimensional stencils and 2.5D tiling specifically for 3D stencils. We provide an in-depth analysis on how the tiling optimizations affects stencils of different shapes and sizes across different applications. Our experimental results show that our approach outperforms existing compiler approaches and hand-tuned codes.

Heterogeneous accelerators often disappoint. They provide the prospect of great performance, but only deliver it when using vendor specific optimized libraries or domain specific languages. This requires considerable legacy code modifications, hindering the adoption of heterogeneous computing. This paper develops a novel approach to automatically detect opportunities for accelerator exploitation. We focus on calculations that are well supported by established APIs: sparse and dense linear algebra, stencil codes and generalized reductions and histograms. We call them idioms and use a custom constraint-based Idiom Description Language (IDL) to discover them within user code. Detected idioms are then mapped to BLAS libraries, cuSPARSE and clSPARSE and two DSLs: Halide and Lift. We implemented the approach in LLVM and evaluated it on the NAS and Parboil sequential C/C++ benchmarks, where we detect 60 idiom instances. In those cases where idioms are a significant part of the sequential execution time, we generate code that achieves 1.26x to over 20x speedup on integrated and external GPUs.

Stencil computations are widely used from physical simulations to machine-learning. They are embarrassingly parallel and perfectly fit modern hardware such as Graphic Processing Units. Although stencil computations have been extensively studied, optimizing them for increasingly diverse hardware remains challenging. Domain Specific Languages (DSLs) have raised the programming abstraction and offer good performance. However, this places the burden on DSL implementers who have to write almost full-fledged parallelizing compilers and optimizers.

Computer systems are increasingly featuring powerful parallel devices with the advent of many-core CPUs and GPUs. This offers the opportunity to solve computationally-intensive problems at a fraction of the time traditional CPUs need. However, exploiting heterogeneous hardware requires the use of low-level programming language approaches such as OpenCL, which is incredibly challenging, even for advanced programmers.

Graphics Processing Units (GPUs) are used as general purpose parallel accelerators in a wide range of applications. They are found in most computing systems, and mobile devices are no exception. The recent availability of programming APIs such as OpenCL for mobile GPUs promises to open up new types of applications on these devices.

Computers have become increasingly complex with the emergence of heterogeneous hardware combining multicore CPUs and GPUs. These parallel systems exhibit tremendous computational power at the cost of increased programming effort resulting in a tension between performance and code portability. Typically, code is either tuned in a low-level imperative language using hardware-specific optimizations to achieve maximum performance or is written in a high-level, possibly functional, language to achieve portability at the expense of performance. We propose a novel approach aiming to combine high-level programming, code portability, and high-performance. Starting from a high-level functional expression we apply a simple set of rewrite rules to transform it into a low-level functional representation, close to the OpenCL programming model, from which OpenCL code is generated. Our rewrite rules define a space of possible implementations which we automatically explore to generate hardware-specific OpenCL implementations. We formalize our system with a core dependently-typed lambda-calculus along with a denotational semantics which we use to prove the correctness of the rewrite rules. We test our design in practice by implementing a compiler which generates high performance imperative OpenCL code. Our experiments show that we can automatically derive hardware-specific implementations from simple functional high-level algorithmic expressions offering performance on a par with highly tuned code for multicore CPUs and GPUs written by experts.

SUMMARY: Next-generation sequencing (NGS) has a large potential in HIV diagnostics, and genotypic prediction models have been developed and successfully tested in the recent years. However, albeit being highly accurate, these computational models lack computational efficiency to reach their full potential. In this study, we demonstrate the use of graphics processing units (GPUs) in combination with a computational prediction model for HIV tropism. Our new model named gCUP, parallelized and optimized for GPU, is highly accurate and can classify >175 000 sequences per second on an NVIDIA GeForce GTX 460. The computational efficiency of our new model is the next step to enable NGS technologies to reach clinical significance in HIV diagnostics. Moreover, our approach is not limited to HIV tropism prediction, but can also be easily adapted to other settings, e.g. drug resistance prediction. AVAILABILITY AND IMPLEMENTATION: The source code can be downloaded at http://www.heiderlab.de CONTACT: [email protected].

Application programming for modern heterogeneous systems which comprise multiple accelerators (multi-core CPUs and GPUs) is complex and error-prone. Popular approaches, like OpenCL and CUDA, are low-level and offer no support for the two most complicated issues: 1) programming multiple GPUs within a stand-alone computer, and 2) managing distributed systems that integrate several such computers. In particular, distributed systems require application developers to use a mix of different programming models, e.g., MPI together with OpenCL or CUDA. We propose a uniform approach based on OpenCL for programming both stand-alone and distributed systems with GPUs. The approach implementation is based on two parts: 1) the SkelCL library for high-level application programming on heterogeneous stand-alone computers with multi-core CPUs and multiple GPUs, and 2) the dOpenCL middleware for transparent execution of OpenCL programs on several stand-alone computers connected over a network.