Google is unclear on the performance of the Tensor chip in Pixel 6 but here are some details

The Pixel 6 and Tensor chips just revealed tonight are mainly focused on its artificial intelligence-focused TPU (Tensor Processing Unit), and how this custom hardware will help Google stand out from its competitors. This is also a key point of Google’s today’s conference: the company called Tensor a milestone of device learning, this chip was designed by Google Research, allowing the company to easily convert advances in artificial intelligence and machine learning into actual consumption.

Google said that the Tensor chip will have the most accurate automatic speech recognition (ASR) provided by Google for quick queries and longer audio tasks for Google Assistant, such as live subtitles or voice recorder applications. A tensor can also implement the new features of Pixel 6, such as motion modes, more accurate face detection, and real-time translation of the text, which can be converted into different languages ​​while you enter text. Google also said that the Tensor chip will handle specialized machine learning tasks, and its power efficiency is much higher than previous Pixel phones.

Join RealMi Central on Telegram, Facebook & Twitter

Smartphone chips have many components besides artificial intelligence. With the exposure of Pixel 6, we finally have more details about other parts of the chip, including CPU, GPU, modem, and the main components that make Tensor work. As rumored, the Tensor chip uses a unique combination of CPU cores. There are custom TPUs (tensor processing units) for artificial intelligence and two high-power Cortex-X1 cores.

The two mid-end rumors are the old Cortex-A76 cores, and then four low-power efficiency cores (probably Arm’s usual Cortex-55 design). Graphics are provided by a 20-core GPU, in addition to an environmental center that powers environmental experiences (such as an always-on display), a private computing core, and a new Titan M2 chip for security. There is also a dedicated image processing core that helps Pixel’s iconic photography.

It’s not entirely clear why Google chose to use the Cortex-A76 core instead of the more modern Cortex-A78 (which is more powerful and power-saving), but it’s worth noting that the Pixel 5’s Snapdragon 765G also uses two A Cortex-A76 core as its main CPU core, so Google may tend to stick to the products it knows. Although the performance improvement is not as arrogant as Apple’s self-developed chips, the new phone should still be the fastest Pixel phone in the vertical comparison.

Google promises that the CPU performance is 80% faster than the Pixel 5 and the GPU performance is 37% faster. The real question is, how do the performance of Pixel 6 and its Tensor chip compare to other traditional Android flagships? Google’s self-developed CPU can be said to be a very unique configuration, especially when compared with the main chips of Qualcomm and Samsung. The traditional four high-performance and four efficiency cores are compared.

In theory, the number of X1 performance cores provided by Google is twice that of Snapdragon 888 or Exynos 2100-this is the most powerful Arm design, the latter all use one Cortex-X1, three Cortex-A78 and four Cortex- A55 cores. But Google also replaced the two high-end cores with mid-range cores, which may help battery life and performance…or it may just cause the overall device to be weaker. Once we have a chance to test Pixel 6 and Tensor, we will know soon.

Leave a Comment