Mobile devices are increasingly offering better photographic features and manufacturers of the mobile industry are constantly looking for new technologies. Nvidia with its Tegra K1 mobile chip, aims to raise the bar even further, thanks to the inherent power of the GPU.
The GPU (Graphical Processor Unit) are becoming the talk of engineers and software developers, and not just because they’re something new, but because of the increasingly central role they’re having. Originally, the majority of all computer tasks are performed by the CPU, but with the increase of specific graphic needs, devices in which such tasks are performed by the GPU were developed.
The CPU and GPU have lived long on computers, each performing specific tasks, but in recent years GPUs have experienced a huge increase in performance, especially thanks to Nvidia and ATI graphics card manufacturers, which has caused many developers and engineers to exploit these capabilities.
The big difference with most of the current solutions is that Nvidia uses the GPU for tasks other manufacturers typically performed with the CPU, and the result can be abysmal.
The Nvidia Tegra K1 mobile chip is possibly the most serious bid within the mobile industry to realize the potential of these GPU in order to provide a processing capacity superior to current solutions. Designed for smartphones and high-end tablets, the Tegra K1 uses the GPU to perform intensive calculations that would be normally done by the CPU.
Like Sridhar Ramaswamy (Senior Technical Marketing Manager of Nvidia Tegra) said, using the potential of the GPU offers great advantage for photographic capabilities. In general, most mobile device manufacturers choose to incorporate cameras with rather modest benefits, because they need to make heavy use of the software to try to improve them. A special case would be the focus of the Nokia Lumia 1020, which opts for a camera module with a Zeiss sensor.
The Tegra K1 is based on Nvidia Kepler architecture, and consists of two versions, 32-bit and 64-bit, the first based on a quad-core A15 CPU and the second on a Denver dual core CPU. But both share the 192 CUDA cores GPU, independent units that can perform different tasks.
One of the key features is precisely the CUDA, a language in which you can program and use a GPU like a CPU, and this allows you to use the vast computing power of the GPU for image processing, unlike the CPU, which is generally focused to do one task before the next.
Nvidia actually provides the appropriate API (Application Programming Interface) so that developers outside the company can use the capabilities of the Tegra K1 to design their own applications and features, something really needed to expand this ecosystem and attract other manufacturers to use the chip.
As for its video capabilities, the Tegra K1 includes codecs for encoding and decoding UHD 4k video via hardware, up to 24 fps and 30 fps respectively -remember that more power is needed to encode than to decode, hence the difference-, and for Full HD video (about 2K), these values are at 60 fps and 120 fps respectively.
With the Tegra K1, the graphical power of mobile devices -smartphones and tablets- is even close to that offered by current laptops.
In terms of power, according to Nvidia, it offers about half of the power of the GeForce 730M/740M graphics cards, since we are approaching the performance of laptops, which sooner or later will be replaced by tablets, at least for most non-professional users. Nvidia says the Tegra K1’s performance is above to those of their current counterparts, such as the Apple A7 chip or the Qualcomm Snapdragon 800 used in Android smartphones.
According to Nvidia, Tegra mobile chip K1 outperforms any competitive solutions, with its Kepler GPU architecture using a 192-core
The Tegra K1 and the intense use of the GPU can process HDR images. Traditionally, this technique combines three independent samples which are then merged into a single image, a task that takes time because it needs a lot of CPU resources. In the Nvidia chip, all HDR processing is performed by the GPU -they call it “always on HDR”- which delivers a unique image without any delay.
The “real time video filtering”, or applying video filters in real time, is another type of functionality that is only possible thanks to the computing power offered by the GPU. Another application that uses this great processing power is the "object tracking" or monitoring of objects. This last feature also has applications in the field of the automotive industry.
The Nvidia Tegra architecture also gives a new dimension to gaming, but scientific applications can also take advantage of the power of the GPU to its intensive calculations.
The Tegra chip K1 is now available on Nvidia products itself, but hopefully they will establish partnerships with other manufacturers later this year, without specifying any particular name.
More detailed information on the Tegra K1 Official File.