Many useful tasks in data science and machine learning applications can be written as simple variations of matrix multiplication. However, users have difficulty performing such tasks as existing matrix/vector libraries support only a limited class of computations hand-tuned for each unique hardware platform. Users can alternatively write the task as a simple nested loop but current compilers are not sophisticated enough to generate fast code for the task written in this way. To address these issues, we extend an open-source compiler to recognize and optimize these matrix multiplication-like tasks. Our framework, called Amulet, uses both database-style and compiler optimization techniques to generate fast code tailored to its execution environment on CPUs. Amulet achieves speedups on a variety of matrix multiplication-like tasks compared to existing compilers while handling a much broader class of computations compared to libraries.

Picking the right data structure for the right job is one of the key challenges for every developer. However, especially in the realm of object-oriented programming, the memory layout of data structures is often still suboptimal for certain data access patterns, due to objects being scattered across the heap. Therefore, this work presents an approach for the automated transformation of arrays of objects into a contiguous format (called columnar arrays). At run time, we identify suitable arrays, perform the transformation and use a dynamic compiler to gain performance improvements. In the evaluation, we show that our approach can improve the performance of certain queries over large, uniform arrays.

Dynamic taint analysis assigns taint labels to sensitive data and tracks the propagation of such tainted data during program execution. This program analysis technique has been implemented in various analysis platforms targeting specific programming languages or program representations and has been applied to diverse fields such as software security and debugging. While some of these platforms support customization of their taint analysis, such customization is typically limited to certain analysis properties or to predefined options. This limitation can require analysis developers to modify the analysis platform in order to adapt other analysis properties or to implement new taint analysis applications.

Dynamic taint analysis is a popular analysis technique which tracks the propagation of specific values while a program executes. To this end, a taint label is attached to these values and is dynamically propagated to any values derived from them. Frequent application of this analysis technique in many fields has led to the development of general-purpose analysis platforms with taint propagation capabilities. However, these platforms generally limit analysis developers to a specific implementation language, to specific propagation semantics or to specific taint label representations.

Today’s huge memories make it possible to store and process large data structures in memory instead of in a database. Hence, accesses to this data should be optimized, which is normally relegated either to the runtimes and compilers or is left to the developers, who often lack the knowledge about optimization strategies. As arrays are often part of the language, developers frequently use them as an underlying storage mechanism. Thus, optimization of arrays may be vital to improve performance of data-intensive applications. While compilers can apply numerous optimizations to speed up accesses, it would also be beneficial to adapt the actual layout of the data in memory to improve cache utilization. However, runtimes and compilers typically do not perform such memory layout optimizations. In this work, we present an approach to dynamically perform memory layout optimizations on arrays of objects to transform them into a columnar memory layout, a storage layout frequently used in analytical applications that enables faster processing of read-intensive workloads. By integration into a state-of-the-art JavaScript runtime, our approach can speed up queries for large workloads by up to 9x, where the initial transformation overhead is amortized over time.

Compiler optimizations are often based on hand-crafted heuristics to guide the optimization process. These heuristics are designed to benefit the average program and are otherwise static or only customized by profiling information. We propose machine-learning-based self-optimizing compiler heuristics, a novel approach for fitting optimization decisions in a dynamic compiler to specific environments. This is done by updating a machine learning model with extracted performance data at run time. Related work—which primarily targets static compilers—has already shown that machine learning can outperform hand-crafted heuristics. Our approach is specifically designed for dynamic compilation and uses concepts such as deoptimization for transparently switching between generating data and performing machine learning decisions in single program runs. We implemented our approach in the GraalVM, a high-performance production VM for dynamic compilation. When evaluating our approach by replacing loop peeling heuristics with learned models we encountered speedups larger than 30% for several benchmarks and only few slowdowns of up to 7%.

Dynamic taint analysis (DTA) is a popular program analysis technique with applications to diverse fields such as software vulnerability detection and reverse engineering. It consists of marking sensitive data as tainted and tracking its propagation at runtime. While DTA has been implemented on top of many different analysis platforms, these implementations generally incur significant slowdown from taint propagation. Since a purely dynamic analysis cannot predict which instructions will operate on tainted values at runtime, programs have to be fully instrumented for taint propagation even when they never actually observe tainted values. We propose leveraging speculative optimizations to reduce slowdown on the peak performance of programs instrumented for DTA on a managed runtime capable of dynamic compilation.

Code duplication is a major opportunity to enable optimizations in subsequent compiler phases. However, duplicating code prematurely or too liberally can result in tremendous code size increases. Thus, modern compilers use trade-offs between estimated costs in terms of code size increase and benefits in terms of performance increase. In the context of this ongoing research project, we propose the use of machine learning to provide trade-off functions with accurate predictions for code size impact. To evaluate our approach, we implemented a neural network predictor in the GraalVM compiler and compared its performance against a human-crafted, highly tuned heuristic. First results show promising performance improvements, leading to code size reductions of more than 10% for several benchmarks. Additionally, we present an assistance mode for finding flaws in the human-crafted heuristic, leading to improvements for the duplication optimization itself.

Optimizing machine learning (ML) workloads on structured data is a key concern for data platforms. One class of optimizations called "factorized ML" helps reduce ML runtimes over multi-table datasets by pushing ML computations down through joins, avoiding the need to materialize such joins. The recent Morpheus systemautomatedfactorized ML toanyML algorithm expressible in linear algebra (LA). But all such prior factorized ML/LA stacks are restricted by their chosen programming language (PL) and runtime environment, limiting their reach in emerging industrial data science environments with many PLs (R, Python, etc.) and even cross-PL analytics workflows. Re-implementing Morpheus from scratch in each PL/environment is amassive developability overheadfor implementation, testing, and maintenance. We tackle this challenge by proposing a new system architecture,Trinity, to enable factorized LA logic to bewritten only onceand easilyreused across many PLs/LA tools in one go. To do this in an extensible and efficient manner without costly data copies, Trinity leverages and extends an emerging industrial polyglot compiler and runtime, Oracle's GraalVM. Trinity enables factorized LA in multiple PLs and even cross-PL workflows. Experiments with real datasets show that Trinity is significantly faster than materialized execution (> 8x speedups in some cases), while being largely competitive to a prior single PL-focused Morpheus stack.

Modern database management systems employ sophisticated query optimization techniques that enable the generation of efficient plans for queries over very large data sets. A variety of other applications also process large data sets, but cannot leverage database-style query optimization for their code. We therefore identify an opportunity to enhance an open-source programming language compiler with database-style query optimization. Our system dynamically generates execution plans at query time, and runs those plans on chunks of data at a time. Based on feedback from earlier chunks, alternative plans might be used for later chunks. The compiler extension could be used for a variety of data-intensive applications, allowing all of them to benefit from this class of performance optimizations.

Developing software systems is inherently subject to errors that can later cause failures in production. While testing can help to identify critical issues, it is limited to concrete inputs and states. Exhaustive testing is infeasible in practice; hence we can never prove the absence of faults. Symbolic execution, i.e., the process of symbolically reasoning about the program state during execution, can inspect the behavior of a system under all possible concrete inputs at run time. It automatically generates logical constraints that match the program semantics and uses theorem provers to verify the existence of error states within the application. This paper presents a novel symbolic execution engine called SymJEx, implemented on top of the multi-language Java Virtual Machine GraalVM. SymJEx uses the Graal compiler's intermediate representation to derive and evaluate path conditions, allowing GraalVM users to leverage the engine to improve software quality. In this work, we show how SymJEx finds non-trivial faults in existing software systems and compare our approach with established symbolic execution engines.

Dynamic taint analysis is a popular program analysis technique in which sensitive data is marked as tainted and the propagation of tainted data is tracked in order to determine whether that data reaches critical program locations. This analysis technique has been successfully applied to software vulnerability detection, malware analysis, testing and debugging, and many other fields. However, existing approaches of dynamic taint analysis are either language-specific or they target native code. Neither is suitable for analyzing applications in which high-level dynamic languages such as JavaScript and low-level languages such as C interact.In these approaches, the language boundary forms an opaque barrier that prevents a sound analysis of data flow in the other language and can thus lead to the analysis being evaded.

Compilers perform a variety of advanced optimizations to improve the quality of the generated machine code. However, optimizations that depend on the data flow of a program are often limited by control-flow merges. Code duplication can solve this problem by hoisting, i.e. duplicating, instructions from merge blocks to their predecessors. However, finding optimization opportunities enabled by duplication is a non-trivial task that requires compile-time intensive analysis. This imposes a challenge on modern (just-in-time) compilers: Duplicating instructions tentatively at every control flow merge is not feasible because excessive duplication leads to uncontrolled code growth and compile time increases. Therefore, compilers need to find out whether a duplication is beneficial enough to be performed.

To cope with today's large scale of data, parallel dataflow engines such as Hadoop, and more recently Spark and Flink, have been proposed. They offer scalability and performance, but require data scientists to develop analysis pipelines in unfamiliar programming languages and abstractions. To overcome this hurdle, dataflow engines have introduced some forms of multi-language integrations, e.g., for Python and R. However, this results in data exchange between the dataflow engine and the integrated language runtime, which requires inter-process communication and causes high runtime overheads. In this paper, we present ScootR, a novel approach to execute R in dataflow systems. ScootR tightly integrates the dataflow and R language runtime by using the Truffle framework and the Graal compiler. As a result, ScootR executes R scripts directly in the Flink data processing engine, without serialization and inter-process communication. Our experimental study reveals that ScootR outperforms state-of-the-art systems by up to an order of magnitude.

Most high-performance dynamic language virtual machines duplicate language semantics in the interpreter, compiler, and runtime system. This violates the principle to not repeat yourself. In contrast, we define languages solely by writing an interpreter. The interpreter performs specializations, e.g., augments the interpreted program with type information and profiling information. Compiled code is derived automatically using partial evaluation while incorporating these specializations. This makes partial evaluation practical in the context of dynamic languages: It reduces the size of the compiled code while still compiling all parts of an operation that are relevant for a particular program. When a speculation fails, execution transfers back to the interpreter, the program re-specializes in the interpreter, and later partial evaluation again transforms the new state of the interpreter to compiled code. We evaluate our approach by comparing our implementations of JavaScript, Ruby, and R with best-in-class specialized production implementations. Our general-purpose compilation system is competitive with production systems even when they have been heavily optimized for the one language they support. For our set of benchmarks, our speedup relative to the V8 JavaScript VM is 0.83x, relative to JRuby is 3.8x, and relative to GNU R is 5x.

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.

The R language, from the point of view of language design and implementation, is a unique combination of various programming language concepts. It has functional characteristics like lazy evaluation of arguments, but also allows expressions to have arbitrary side effects. Many runtime data structures, for example variable scopes and functions, are accessible and can be modified while a program executes. Several different object models allow for structured programming, but the object models can interact in surprising ways with each other and with the base operations of R.

Originally developed with a single language in mind, the JVM is now targeted by numerous programming languages—its automatic memory management, just-in-time compilation, and adaptive optimizations—making it an attractive execution platform. However, the garbage collector, the just-in-time compiler, and other optimizations and heuristics were designed primarily with the performance of Java programs in mind. Consequently, many of the languages targeting the JVM, and especially the dynamically typed languages, are suffering from performance problems that cannot be simply solved at the JVM side. In this article, we aim to contribute to the understanding of the character of the workloads imposed on the JVM by both dynamically typed and statically typed JVM languages. To this end, we introduce a new set of dynamic metrics for workload characterization, along with an easy-to-use toolchain to collect the metrics. We apply the toolchain to applications written in six JVM languages (Java, Scala, Clojure, Jython, JRuby, and JavaScript) and discuss the findings. Given the recently identified importance of inlining for the performance of Scala programs, we also analyze the inlining behavior of the HotSpot JVM when executing bytecode originating from different JVM languages. As a result, we identify several traits in the non-Java workloads that represent potential opportunities for optimization. © 2015 The Authors. Software: Practice and Experience Published by John Wiley & Sons Ltd.

We present an approach to cross-compile Java bytecodes to Java-Script, building on existing Java optimizing compiler technology. Static analysis determines which Java classes and methods are reachable. These are then translated to JavaScript using a re-configured Java just-in-time compiler with a new back end that generates JavaScript instead of machine code. Standard compiler optimizations such as method inlining and global value numbering, as well as advanced optimizations such as escape analysis, lead to compact and optimized JavaScript code. Compiler IR is unstructured, so structured control flow needs to be reconstructed before code generation is possible. We present details of our control flow reconstruction algorithm. Our system is based on Graal, an open-source optimizing compiler for the Java HotSpot VM and other VMs. The modular and VM-independent architecture of Graal allows us to reuse the intermediate representation, the bytecode parser, and the high-level optimizations. Our custom back end first performs control flow reconstruction and then JavaScript code generation. The generated JavaScript undergoes a set of optimizations to increase readability and performance. Static analysis is performed on the Graal intermediate representation as well. Benchmark results for medium-sized Java benchmarks such as SPECjbb2005 run with acceptable performance on the V8 JavaScript VM.

When building a compiler for a high-level language, certain intrinsic features of the language must be expressed in terms of the resulting low-level operations. Complex features are often expressed by explicitly weaving together bits of low-level IR, a process that is tedious, error prone, difficult to read, difficult to reason about, and machine dependent. In the Graal compiler for Java, we take a different approach: we use snippets of Java code to express semantics in a high-level, architecture-independent way. Two important restrictions make snippets feasible in practice: they are compiler specific, and they are explicitly prepared and specialized. Snippets make Graal simpler and more portable while still capable of generating machine code that can compete with other compilers of the Java HotSpot VM.

An abstract syntax tree (AST) interpreter is a simple and natural way to implement a programming language. However, it is also considered the slowest approach because of the high overhead of virtual method dispatch. Language implementers therefore define bytecodes to speed up interpretation, at the cost of introducing inflexible and hard to maintain bytecode formats. We present a novel approach to implementing AST interpreters in which the AST is modified during interpretation to incorporate type feedback. This tree rewriting is a general and powerful mechanism to optimize many constructs common in dynamic programming languages. Our system is implemented in Java and uses the static typing and primitive data types of Java elegantly to avoid the cost of boxed representations of primitive values in dynamic programming languages.