Since 1986 - Covering the Fastest Computers in the World and the People Who Run Them

Language Flags
June 1, 2011

AMD Hitches Hopes to Fusion

Nicole Hemsoth

Some argue that AMD has a long road ahead as it seeks to push the OpenCL movement along. The claim goes that the ultimate success of the company’s Fusion APUs will depend in large part on how well GPU computing takes off, which means it is risky business to tie all hopes to how well this movement fares.

As Ben Hardwidge wrote this week, “Fusion is either going to carry AMD through to the victory parade, or drag it through the streets for a pelting in the village stocks.” He says that as of now, “the whole future of AMD’s CPU division rests on GPGPU computing being catapulted into the mainstream” If this happens, he opines, this means the company’s APU could theoretically knock Intel’s CPUs in GPU-accelerated applications and do the same for the Core i7.

To get to the heart of these questions about the future of the APU and GPGPU in general, he put some tough questions to AMD’s manager of Fusion software marketing, Terry Makedon and the head of product marketing for desktops and Fusion software, Sasa Marinkovic.

In this interview, Makedon notes that indeed, OpenCL is important for AMD and that they hope to bring some new opportunities for GPGPU development out of their upcoming Fusion Developer Summit in Seattle in a few weeks (we will be on site for that, by the way).

He gives kudos to Nvidia for “coming out with CUDA and pushing the whole GPGPU market in a certain direction” but says that he’s been of the opinion that “CUDA’s not going to exist forever, it’s going to go away—it’s proprietary.” He pointed to the recent announcement from Intel that it was opening an OpenCL certification program and says that “all of a sudden our friends at Intel have joined the industry standard OpenCL.”

Makedon sheds light on the OpenCL/CUDA debate for HPC specifically. He says that although CUDA won’t last in many segments, “it will exist for very specific high-performance computing applications—stuff like genetic mapping and specific things that universities are researching.” He claims that CUDA will be the norm in this segment,  in part because “places like that are happy with the infrastructure and support and they’ll just stick around.” He feels that CUDA was first to the game and it is here that their advantage lies.

For those interested in more about the desktop side of CUDA and OpenCL conversation, the interview is lengthy and detailed.

Full story at Bit-Tech