Christian Wimmer’s research while affiliated with Oracle Corporation and other places

... It is present as a strategy to improve code quality and consequently optimize the Code Review process [7,8,9]. It is part of steps of modern compilers such as GraalVM [10], being necessary for the generation of optimized native code. It is present in popular tools like Github, which stores hundreds of millions of source code repositories and receives more than 1,000 pushes per minute, from more than 65 million developers. ...

Reference:

Evaluating Source Code Quality with Large Language Models: a comparative study

... Truffle [39,85,88] is a language implementation framework that enables guestlanguage integration via AST interpretation. At run time, the guest-language ASTs are partially evaluated and optimized based on profiling information accumulated during interpretation [42,50,87,89] and are subsequently compiled using the GraalVM Compiler. Truffle furthermore provides a standardized object storage model [86] that can be used across all language implementations. ...

Reference:

Control Flow Duplication for Columnar Arrays in a Dynamic Compiler

... Since then, multiple optimization methods were introduced, attempting to reduce the negative performance effect of object-orientation, including inline caching [13], object specialization [2,36], object inlining [6,14,15,39], devirtualization [23], semantic inlining [41], method specialization [33], object combining [37], object fusing [40], elimination of lambdas [28], and value fields inlining [5]. However, even though some of these optimizations were implemented in HotSpot [30], Java object methods are still slower than static methods, as it was demonstrated in Section 2. It is possible that they will never be equally fast due to fundamental differences between object-oriented and procedural programming paradigms [21]. ...

Reference:

Embracing Objects Over Statics: An Analysis of Method Preferences in Open Source Java Frameworks

... GraalVM [53,54,77] is a virtual machine and compiler infrastructure providing a shared platform for language implementations. It allows for dynamic optimizations where nodes in an AST structure may rewrite themselves during interpretation, alongside speculative optimization and deoptimization [78]. ...

Reference:

Quff: A Dynamically Typed Hybrid Quantum-Classical Programming Language

... GraalVM is a high-performance JDK distribution that makes it possible to build and run applications implemented in a wide range of high-level languages (e.g., Java, Scala, JavaScript, Clojure, Kotlin, etc.). Montsalvat leverages a GraalVM tool named Native Image supporting ahead-of-time (AoT) compilation of applications into native executables, called native images [18,35,57], which do not require a JVM at runtime. GraalVM native-image AoT compiles only reachable application methods, classes and fields, thereby excluding any redundant application logic from the final executable. ...

Reference:

Montsalvat: Intel SGX Shielding for GraalVM Native Images

... Specifically, there is a surge in the popularity of ahead-of-time (AOT) compilation. GraalVM [9] introduced a component: Native Image [10], which is a compiler for Java bytecode that uses a combination of points-to analysis, AOT compilation, and class initialization at build time to create self-contained binaries which start quickly and execute with a lower memory footprint compared to running on the Java Virtual Machine (JVM). These capabilities make it a perfect prospective for cloud environments. ...

Reference:

Software Architecture Reconstruction for Microservice Systems Using Static Analysis via GraalVM Native Image

... Partial evaluators have been developed for numerous languages [Alpuente et al. 1998;Andersen 1992;Elphick et al. 2003;Komorowski 1982;Lloyd and Shepherdson 1991;Meyer 1991] and for different domains [Berlin 1990;Berlin and Weise 1990;Futamura 1971]. The purpose of partial evaluation is to specialize code based on values known at partial-evaluation time in order to do less work at execution time, hopefully improving performance [Jones 1996;Würthinger et al. 2017]. Partial evaluation can be broken into online and offline techniques with [Jones 1996] explaining the differences the best. ...

Reference:

Practical compilation of fexprs using partial evaluation: Fexprs can performantly replace macros in purely-functional Lisp

... Specifically, our evaluation with a set of wellknown benchmarks shows significant reductions by up to 9x in program warm-up given that OSR is applicable. Note that our approach can be used by other Truffle bytecodebased implementations lacking high-level loop information, including GraalSqueak [19], Truffle Java [11] and Truffle CIL. Furthermore, it is applicable in any compiler with a background system which provides means for establishing mappings between extracted control flow. ...

Reference:

Supporting on-stack replacement in unstructured languages by loop reconstruction and extraction

... In the last decades, little progress has been made improving the encoding of floats, with most implementations either using boxed floats or suffering from the overhead of NaN-tagging on pointers. Thus, more general strategies that use data flow analysis [14,22,27,31] and partial evaluation [35] have been developed to find locations where the type of a value is known across its lifecycle. This allows the generation of specialized code that safely handles fully unboxed, untagged values. ...

Reference:

Float Self-Tagging

... If the dynamic typecheck fails, the code is either deoptimized, or the execution gets continued in baseline compiled code, which does not inline equals, but is less efficient [55]. ...

Reference:

Inlining-Benefit Prediction with Interprocedural Partial Escape Analysis