Nvidia plans to offer CUDA architecture on their upcoming Tegra system-on-a-chip for mobile devices, along with all their GPU offerings. However, the first Tegra installment, due in mid-2009, won't carry CUDA support, said Jen-Hsun Huang, cofounder, president and CEO of Nvidia in an interview.

CUDA is a C language environment that aides software developers to write programs for solving complex computational problems using parallel processing power of multicore GPU.

Huang said that a new computer architecture based on GPUs and other types of parallel processors that work together with the conventional CPUs is emerging. "The CPU is excellent for sequential processing, but there are many types of problems that you can operate on in parallel," he added.

Nvidia is working with application developers to port their software to the CUDA architecture, Huang said. Nvidia has already made CUDA available for free.

Since the first batch of Tegra is yet to arrive, it's too early to say anything about future integration of Tegra and CUDA.

Source: Network World