Everything about the Tensor G5 has been leaked, the first designed entirely by Google for its mobile phones

In mid-September, some details of the Pixel 10 series were revealed to create some of the specifications of the upcoming Google phones. And now comes an important leak, since it concerns all the information about the Tensor G5, the first chip designed entirely by Google without depending on Samsung.

In previous Tensors, Google was behind the competition, so the wait is maximum for what it is capable of doing with the Tensor G5 and now you can check out some of the best specs of this SoC.

A massive leak from Google’s gChips division has allowed Android Authority to see some documents that show everything about the upcoming Tensor G5 chip. First of all, the previous Tensor G4 was characterized by an updated processor compared to that of its predecessor, although it did not really bring significant improvements and I was happy to be 6% faster in several tests nuclei the nuclei.

This is mainly because Google removed the middle cluster core to reduce the power consumption of Tensor G4. Tensor G5 updates CPU cluster againbut in a different way than one might expect, since it will retain the main Arm Cortex-X4 core in order to take advantage of the new Cortex-X925, which will allow important improvements in general terms.

The main clusters were also distributed differently. On the one hand, the supports now have five Cortex-A725 cores instead of the three Cortex-A720 of the Tensor G4. The other change concerns its smallest cluster which has been reduced to just two nuclei Cortex-A520.

The main objective is that Tensor G5 offers a decent multi-core performance boost. Worst of all is that the same Cortex-X4 is still used, although it must be made very clear that these are specifications and will need to be tested. on site the chip to verify that all these changes will lead to an improvement in the performance of Google’s next Pixel 10.

Graphics and AI

Let’s move on to the GPUGoogle decided to build one from Imagination Technologies (IMG), the DXT-48-1536 at a speed of 1.1 GHz. Little is known about this graphic, although there are two interesting details: support for ray tracing or. ray tracingand GPU virtualization, which allows the use of accelerated graphics in a virtual machine.

In AI processing The Tensor G5 doesn’t receive as many improvements and has a slightly faster TPU (Cloud Tensor Processing Unit) than that of the Tensor G4. The TOPS (trillion operations per second) is 40% higher, but remains at a final improvement of 14% according to internal tests carried out by Google.

The G5 tensor is designed in 3nm by TSMCsimilar to Apple’s A18 Pro, and although its specifications don’t seem like a huge improvement, for a mobile like the Pixel it may be more than enough to get a significant step up in performance. If Google is able to deliver great photos with sensors from a few years ago or more, the same is true with its chip and software, and that’s where the beauty of its Pixel lies.