Why did you choose a single language only?

Strictly speaking, the Scala Benchmarking Project compares and contrasts two languages: Scala and Java. That being said, the reason for comparing a single language rather than several with Java is simple: Any benchmark suite on par with the Java benchmark suites used in JVM research needs to cover a broad selection of real‐world applications; choosing a single language makes it easier to develop a comprehensive yet cohesive suite.

[top]


Why did you choose Scala?

This research effort focuses on Scala, a statically‐typed language with roots in both functional and object‐oriented programming. The reason for preferring Scala over other possible candidates like Clojure, Groovy, JRuby, or Jython is as follows: Of the aforementioned five languages, four are dynamically‐typed. But this single language feature has significant impact on the performance of the JVM, a machine which, until recently, has been specifically tailored towards a single, statically‐type language, namely Java. In contrast to the dynamically‐typed languages, the execution characteristics of other statically‐typed languages like Scala on the JVM are less well understood, making the candidate most attractive for research.

Another interesting fact about Scala is that it also supports the Microsoft .NET platform. In future work, this will make it possible to verify whether the findings of the Scala Benchmarking Project are restricted to the JVM or can be generalized to other platforms as well.

[top]


Is Scala faster/slower than Java?

The goal of the Scala Benchmarking Project is explicitly not to compare the “performance of Scala” with the “performance of ‘raw’ Java.” Instead, this project aims at contrasting the execution characteristics of Scala and Java applications. The question is thus not which of the two languages is faster/slower, but whether they exhibit different characteristics during execution. If so, this knowledge can then be used by JVM researchers to optimize a virtual machine specifically with respect to Scala rather than Java applications.

[top]


Is there an official mailing list for the Scala Benchmarking Project?

Not yet. We continuously monitor the mailing lists of the DaCapo project, though.

[top]