As you may have seen with Tegra 3, NVIDIA won some “super-phone” contracts, and was included in many tablet designs. Despite the relative success of those products, the volume of Tegra 3 devices shipped remain a small percentage of the overall market (for more details on that, check my NVIDIA Shield is about shaping the mobile battlefield story).
One of the reason was because wireless carriers and handset makers wanted to have more cost-efficient handset designs with 4G LTE. To do that, one needs to integrate the 4G LTE modem into the main processor (or SoC) and reduce the power envelope in order to avoid shipping devices with a huge battery. The American Galaxy S3 uses a Snapdragon S4 chip, and many others made the same choice last year. With Tegra 4i and an integrated 4G LTE modem, NVIDIA can now address that same market, and this opens a new world of opportunities for the company. The fight will be very tough, but at least, NVIDIA has something on the table now.
From a high-level, Tegra 4i runs on the same principles than Tegra 4 does: Both chips use a 4 cores + 1 battery saver core architecture. The idea is that the four main cores take on the heavy workload, and the battery saver core is there for low-intensity tasks and states. But a closer look reveals that Tegra 4i has 60 graphics units instead of 72 on its older sibling. It also uses ARM R4 CPU cores (the design is derived from the Cortex A9) that are much more nimble and power-efficient. And as we said earlier, Tegra 4i has an integrated programmable 4G LTE modem.
The R4 cores offer better performance than the Cortex A9 used in Tegra 3, but are much smaller than the Cortex A15 featured in Tegra 4. And because R4 is so tiny (1.2 square mm) its chip design exhibits a higher than usual performance/size ratio. The bottom-line is: on paper, it is a great chip. Now, we’re really looking forward to getting our hands on it in the coming days.
What do you think? Is this how you expected Tegra to evolve?