Java for the graphics card
Phil Pratt-Szeliga, a doctoral student at New York's Syracuse University, has released the source code to his Rotbeer GPU compiler on Github. He had introduced the software in late June at the High Performance Computing and Communications Conference in Liverpool, the content of this presentation is in the documentation is also available on Github directory (PDF).
The developer compares his compiler with the alternative of CUDA or OpenCL libraries in Java code, integrate, and refers to his approach to a higher degree of automation: Its compiler Serialize complex representations of objects and even arrays of primitive data types in. He also analyzed independently using the Java optimization framework Soot Java bytecode and create automatically CUDA code.
Pratt-Szeliga, who wants to maintain its achievement so long and evolve, "to the Java programming language is no longer popular," has given the compiler source code of some 21,000 lines of additional 39 test cases that comprise the 7,000 lines of code. It indicates that all tests on both Windows and Linux are successful. But would not support dynamic method calls native methods and reflection. Also, garbage collection is still way off.
Subscribe to:
Post Comments (Atom)
No comments:
Post a Comment