Nvidia Launches Quad-core Tegra 4i SoC with Integrated 4G LTE Chip
Known for a while by its codename Gray, the first Nvidia chipset for smartphones and tablets that comes with an integrated 4G LTE chip was officially announced: Tegra 4i. With the launch of its first integrated platform, Nvidia enters in direct competition with Qualcomm and the smaller MediaTek, the only companies that have chipsets with complete solutions in their portfolio.
In January this year, Nvidia introduced its new Tegra 4 SoC syncronized with the i500 radio modem. Even though the new Tegra 4i is related to the Tegra 4, the new unit hides more differences underneath. The 2G/3G/4G radio modem is implemented within the same silicon pill, but the processing core is different.
While Tegra 4 is using ARM Cortex A15 cores in Nvidia’s proprietary 4-PLUS-1 configuration, the new Tegra 4i is based on something more… errr… “exotic.” The ARM Cortex-R4 core has specifications similar to the popular Cortex A9 core, but unlike it, it’s optimized for real-time processing by including bigger cache memories, hardware processing for instructions predictions, and better routines for out-of-order execution. The 4-PLUS-1 architecture was kept, it’s main purpose being to reduce the battery consumption in idle mode.
Nvidia might have chosen the ARM Cortex R4 over the Cortex A15 because the silicon pill is smaller and the cores are more battery friendly. Therefore, Tegra 4i has a surface of only 60 mm², compared to Tegra 4, with 80mm².
In order to compensate the power differences and taking advantage from an improved manufacturing process, Tegra 4i’s cores will be clocked at 2.3 GHz, compared to Tegra 4′s 1.9 GHz clock speed. By integrating the i500 modem, Nvidia can offer its partners a solution with reduced battery consumption, compared to the previous models that were requiring a separate modem, also keeping the production costs to a low level. Anyway the real benefits of the Tegra 4i platform will be see when the first terminals based on the new chipset will hit the market.
A concession made for energy consumption and head dissipation was to reduce the number of processing units of the GPU from 72 to 60, the maximum supported resolution dropping from 3,200 x 2,000 pixels to only 1,9200 x 1,200 pixels.
The other specifications like the video compression or decompression engine orComputational Photography Architecture (Chimera), the processing rutines that provide better speeds for synchronized HDR capture or processing for two photos, are similar.