AMD goes CUDA or parallel computing development
The quest for “true” heterogeneous computing, the GPU and CPU combo
First announced last month, NVIDIA’s CUDA toolkit for parallel programming development is now available for download but it is not without its competitors. As the debate continues to heat up around the programming standards and best practices used for heterogeneous computing, AMD is taking steps to foster some kind of leadership position through its partnership position with a non-proprietary alternative in the form of OpenCL.
Unlike CUDA, the chip manufacturer says that OpenCL enables developers to tap into the full power of the PC by accessing both the GPU and CPU for true heterogeneous computing. While CUDA is arguably the incumbent standard, AMD insists that OpenCL is closing the gap and is an industry-wide effort, supported by a range of companies including AMD, Apple, and Intel.
AMD says it has continued to expand its support of OpenCL with several recent announcements:
- OpenCL-based plug-in for Autodesk: This month AMD announced a technology demonstration of a Bullet Physics plug-in for Autodesk Maya 2011 software. The new plug-in is based on OpenCL industry standards and the open-source Bullet Physics Engine.
- OpenCL University Kit: AMD recently released the OpenCL University Kit, a set of materials that can be leveraged by any university to assist them in teaching a semester course in OpenCL programming.
- AMD APP SDK with Open CL 1.1 support: A new announcement that sees AMD APP technology enabling AMD GPUs and CPUs to work together to run many demanding computing tasks faster than possible with just the CPU alone.
Source: Dr Dobbs’s