Advanced Micro Devices and its ATI graphics business unit will not support Nvidia Corp.'s CUDA general-purpose computing on graphics processing units (GPGPU) environment. AMD/ATI believes that since CUDA (which was originally an abbreviate for compute unified device architecture) was developed by Nvidia, the latter will get unreasonable performance benefits in CUDA-powered applications, whereas performance of ATI graphics chips may be handicapped.
"[Nvidia] would intentionally damage performance to make Nvidia GPUs run the same app better. Even if it was not intentional, it would not be optimized for our instruction set architecture like our own SDK," said Gary Silcott, a spokesperson for AMD, reports The Inquirer web-site.
Earlier this week William Dally, chief scientist of Nvidia, said that in the future software developed using Nvidia’s proprietary CUDA software development kit will be able to work on a broader range of platforms.
“In the future you will be able to run C with CUDA extensions on a broader range of platforms, so I don't think that will be a fundamental limitation. I am familiar with some projects that are underway to enable CUDA on other platforms,” said Bill Dally at a “roundtable event”.
Despite of the fact that Mr. Dally did not name ATI Radeon graphics processing unit family as an alternative platform for Nvidia's CUDA, since presently there are no widespread stream computing platforms apart from Nvidia GeForce and ATI Radeon, it is logical to assume that Nvidia's chief scientist meant AMD's graphics processors.
AMD has always said that it plans to support only open, industry-wide GPGPU programming standards, which are OpenCL and Microsoft DirectX compute shaders.