The recognition of the program constructs that are frequently used by software developers is a powerful mechanism for optimizing and parallelizing compilers to improve the performance of the object code. XARK (EXtensible Compiler for Automatic Recognition of Computational Kernels) is a compiler framework that recognizes syntactical variations of a large collection of program constructs. Well-known examples are induction variables, reductions and array recurrences, which have been the focus of intensive research in the area of compiler technology during the 90's.

In addition, XARk is a platform for the development of compiler techniques that require advanced symbolic analysis. So far, it has been successfully used as a powerful information-gathering tool for parallelizing transformations applied to sequential source codes, and for building analytical models of the behavior of the cache memory.

RSS feed is provided so that users may be efficiently and easily informed of changes and new features, without having to check the web page looking for updates.