On the benefits and pitfalls of extending a statically typed language JIT compiler for dynamic scripting languages

Castaños, José G.; Edelsohn, David; Ishizaki, Kazuaki; Nagpurkar, Priya; Nakatani, Toshio; Ogasawara, Takeshi; Wu, Peng

doi:10.1145/2384616.2384631

Cited by 32 publications

(28 citation statements)

References 19 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In general, the benchmarks cannot be considered representative of real-world applications, yet significant performance differencens between Python compilers we considered indicative by the developers of the Fiorano JIT compiler [8], and benchmarks from the CLBG project have been used in the recent study of JVM languages by Li et al [16]. To provide complementary results for comparable workloads, we have decided to adopt the approach of Li et al, and based our study (mostly, but not completely) on 10 CLBG benchmarks, listed in Table 1 along with a brief description and inputs used.…”

Section: Benchmark Description Inputmentioning

confidence: 99%

“…Due to the high number of combinations and extensive duration of the experiments, we have not included different language runtimes among independent variables. Similarly, we have not varied the execution platform, because the metrics are defined at the bytecode level, and can be considered largely JVM 8 and platform independent.…”

Section: Measurement Contextmentioning

confidence: 99%

“…Specialization is considered to significantly aid performance of dynamic languages targeting the JVM [8]. Hot polymorphic call-sites are good candidates for optimizations such as inline caching [13] and method inlining [10], which specialize code paths to frequent receiver types or target methods.…”

Section: Call-site Polymorphismmentioning

confidence: 99%

“…Making the JVM perform well with various statically-and dynamically-typed languages clearly requires significant effort, not only in optimizing the JVM itself, but also, more importantly, in optimizing the bytecode-emitting language compiler, instead of just relying on the original JIT to gain performance [8]. This in turn requires that developers of both the language compilers and the JVM understand the characteristics of the JVM workloads produced by various languages or compilation strategies.…”

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Characteristics of dynamic JVM languages

Sarimbekov

Podzimek

Bulej

et al. 2013

Proceedings of the 7th ACM Workshop on Virtual Machines and Intermediate Languages

View full text Add to dashboard Cite

The Java Virtual Machine (JVM) has become an execution platform targeted by many programming languages. However, unlike with Java, a statically-typed language, the performance of the JVM and its Just-In-Time (JIT) compiler with dynamically-typed languages lags behind purpose-built language-specific JIT compilers. In this paper, we aim to contribute to the understanding of the workloads imposed on the JVM by dynamic languages. We use various metrics to characterize the dynamic behavior of a variety of programs written in three dynamic languages (Clojure, Python, and Ruby) executing on the JVM. We identify the differences with respect to Java, and briefly discuss their implications.

show abstract

Section: Benchmark Description Inputmentioning

confidence: 99%

Section: Measurement Contextmentioning

confidence: 99%

Section: Call-site Polymorphismmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 2 more Smart Citations

Characteristics of dynamic JVM languages

Sarimbekov

Podzimek

Bulej

et al. 2013

Proceedings of the 7th ACM Workshop on Virtual Machines and Intermediate Languages

View full text Add to dashboard Cite

show abstract

“…invokedynamic) or compiler alterations (e.g. [23]) can improve performance, the results still lag some way behind their meta-tracing equivalents [11]. We believe that language composition by meta-tracing has the potential to sidestep the semantic mismatch problem entirely by allowing each language's VM to be optimised specifically for that language.…”

Section: Related Workmentioning

confidence: 99%

Unipycation

Barrett

Bolz

Tratt

2013

Proceedings of the 7th ACM Workshop on Virtual Machines and Intermediate Languages

View full text Add to dashboard Cite

Language composition approaches have traditionally suffered from poor performance. In this paper we hypothesise that meta-tracing provides a means to compose independent language interpreters while retaining the performance levels of each. To study this approach, we compose Python and Prolog interpreters to form Unipycation. We present a case study of its use and a suite of micro-benchmarks which give us some understanding of its cross-language performance.

show abstract

Workload characterization of JVM languages

Sarimbekov

Mössenböck

Bulej

et al. 2015

Softw. Pract. Exper.

View full text Add to dashboard Cite

SUMMARYOriginally 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 optimizationsmaking 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.

show abstract

On the benefits and pitfalls of extending a statically typed language JIT compiler for dynamic scripting languages

Cited by 32 publications

References 19 publications

Characteristics of dynamic JVM languages

Characteristics of dynamic JVM languages

Unipycation

Workload characterization of JVM languages

Contact Info

Product

Resources

About