Graphics chipmaker nVidia has said it plans to license as IP cores some of its technology in the hope of building up a customer base among other chipmakers and systems houses developing their own SoCs.
Jen-Hsun Huang, president and CEO of nVidia told the Reuters Global Technology Summit in San Francisco, CA this week: “The bottom line is the world has changed and we’re expanding our business model to serve markets that we historically could not serve by selling chips alone.”
Compared to its chief competitor AMD, which bought graphics chipmaker ATI several years, nVidia saw the potential of making SoCs for the mobile and embedded market at an early stage. The company launched its first Tegra embedded SoC in 2009. The latest volume-shipping version, Tegra 3, combines a multicore ARM9 architecture with its own GPU-compute engines, last year. AMD recently launched an x86-based design with integrated graphics and plans to follow up with an ARM-based product line.
However, the Tegra family competes against a number of multicore SoCs from mobile and embedded chipmakers that license either Imagination Technologies’ GPU IP, which has built up a strong position in the market, or the Mali product line from ARM. So far, the main market for the Tegra 3 appears to be in automotive infotainment systems rather than phones – the product has design-ins at Audi and Tesla Motors.
“This is a way for us to engage customers who don’t like to buy chips because they like to create their own, because they have the capacity, creativity and now the scale to build their own,” Huang told the Reuters conference.
The graphics company has licensed its technology before – to Sony for the Playstation 3. This move marks the first time that the company plans to offer GPU IP on a more widespread basis and has picked the Kepler engine as the first to be made licensable. Although Kepler is used in its high-end graphics cards, nVidia plans to put the core into its Tegra 5 mobile and embedded product, expected to start shipping in 2014 aiming to take advantage of the trend towards GPU computing. The Kepler engine runs nVidia’s CUDA compute environment as well as OpenCL.