Nvidia's Tegra Line of SoCs
So far, Nvidia is the only company from the PC space working in the mobile SoC market and seeing at least some success. In light of its latest host and graphics processor designs, Nvidia appears to have a lot of potential in the tablet space, especially, that still hasn't been tapped.
Tegra is the name Nvidia gives to its mobile SoCs, dating back to the company's first push into the media player and smartphone markets. The first and only contract for Tegra was signed with Microsoft for its Zune HD media player.
It wasn’t until Tegra 2, with its first-ever dual-core mobile CPU and GeForce graphics, that Nvidia was taken more seriously in the mobile space. Unfortunately, that processor shipped later than promised. And by the time it attracted attention, competitors were offering their own dual-core chips as well. Additionally, Tegra 2 had some performance issues related to design decisions Nvidia made, which ended up frustrating the early adopters willing to give the newcomer a shot. This resulted in recommendations to stay away from the company's SoC.
Still, when Nvidia announced Tegra 3, the first quad-core Cortex-A9-based processor, there was hope that the company had ironed out its problems. After all, we've seen one GeForce card after another do battle in the hyper-competitive world of PC graphics. But Nvidia didn't end up conquering its struggle against delays. By the time Tegra 3 arrived, Nvidia was forced to market it as a fast mid-range device.
This strategy proved somewhat successful, and Tegra 3 won a place inside the first Nexus 7 tablet from Google and in the first Surface RT tablet from Microsoft. The Nexus 7 sales alone made up half of the SoC's sales. However, despite a moderate win, the fact that Nvidia didn’t really have a high-end product in the mobile market couldn’t be overlooked.
Tegra 4 appeared a year later with a four-core Cortex-A15 complex (plus another battery-saver -A15 companion), promising performance on par with some of the fastest SoCs out there, which it actually delivered. However, Tegra 4 also ran into a couple of issues. One was the fact that it used an old GPU architecture lacking unified shaders and limited to OpenGL ES 2.0-based games. Meanwhile, competitors were coming out with unified shader architectures and OpenGL ES 3.0 support.
Nvidia was actually supposed to deliver the Kepler GPU architecture with Logan in 2013. But because Tegra 2 and 3 were delayed so severely compared to their original schedules, Tegra 4 ended up rolling out in 2013, pushing what we now know as Tegra K1 to 2014.
The other major issue was that Tegra 4 isn’t terribly efficient, resulting in its use as a tablet-oriented SoC at best. Aside from a few Chinese OEMs, nobody's using it in a smartphone.
So, on one hand, execution problems and delays forced Nvidia to push a chip armed with obsolete graphics into 2013. And, on the other, design decisions Nvidia made regarding performance led to it not being well-suited for smartphones, barely qualifying it for tablet form factors. Unsurprisingly, Google dropped Nvidia as a partner for the 2013-edition Nexus 7, which was a major blow considering the previous-gen Nexus 7 represented so much of the SoC's sales.
Fortunately for Nvidia, Tegra K1 looks a lot more appealing to both consumers and OEM partners, and we’ll explore why soon enough.