Halide - new programming language for applications for image processing


              Researchers from the "Computer Science and Artificial Intelligence Laboratory" (CSAIL) at MIT (Massachusetts Institute of Technology) have introduced a new programming language halides. It will facilitate the development of applications for image processing on modern computer systems. With this embedded DSL (Domain Specific Language) is the efficient use of algorithms, for example, Camera Raw pipelines, bilateral grid, Laplace filtering or image sharing possible.

              Halide-programs should be not only easier to write and to read as image processing code in other conventional programming languages. Rather, the promises automated code optimization in a significantly faster execution Halide (PDF), even compared to more complex optimization händischer - the researchers say up to one-third shorter code for up to six times higher processing speed. One of the major factors is a high degree of parallelization Maximization of individual processes.

               When embedded in C + + runs on platforms such as halides x86-64/SSE, ARM v7/NEON and CUDA. It supports SIMD units, multiple cores, and complex memory hierarchies. The compiler can be applied to Mac OS X systems (6.10 to 8.10, the team sets itself a) install and mainstream Linux distributions like Ubuntu (12.04). Builds for Windows should be technically possible, but the researchers tried or tested was. Halide is open source and licensed under the MIT license. The source code for the compiler and for halide itself can be downloaded from GitHub.

No comments:

Post a Comment