As GPGPU usage has become more main stream, the desire to utilize this type of data parallelism in computationally intensive applications has also increased. The early and current CUDA programming tools have allowed the development community to easily embrace this technology, and many different applications have resulted [1]. Utilizing the CUDA language extensions and runtime API is a decent method to incorporate GPU technology, however, porting existing applications can be somewhat labor intensive unless automated externally. Having compiler extensions (frontend or backend) that can identify and organize such data parallelism will accelerate the porting these existing applications and further acceptance of GPGPU usage.

As a means to eliminate hardware dependence per GPU generation the Parallel Thread Execution ISA was developed and utilized in the backend for GPGPU kernel compilation [2]. While this platform is useful for efficient code optimization and hardware portability, it also presents an opportunity to more efficiently add GPGPU extensions to existing compilers/language extensions. The purpose of this development effort is to demonstrate the process and issues associated with adding PTX extensions to the backend of an existing compiler frontend/backend.


Included in this site are the following:

1. Explanation of the test (pseudo) language
2. History/Background of PTX
3. Proposed Features of the backend being developed
4. Time line for development
5. Project Milestones
6. Pseudo Language Framework With PTX support
7. Performance Analysis and Results

Link to Proposal

Link to Progress Report (November 11)

Link to Final Report