Get the App
SLTechnology News&Howtos  ›  IT Information  › 

Google's third-generation self-developed processor Tensor G3 architecture exposure: unique 9-core CPU, supporting ray tracing

Shulou Source: shulou.com Published: 2023-11-24 17:15:47 09月12日 Update

CTOnews.com June 4 news, foreign media Android Authority recently learned through Google internal sources, Google's third generation of self-developed processor Tensor G3 will adopt a unique 9-core CPU architecture, GPU new ray tracing function.

It is reported that the Tensor G3 CPU will be composed of one 3.0GHz Cortex-X3 core + four 2.45GHz Cortex-A715 cores + four 2.15GHz Cortex-A510 cores, which will increase the frequency compared with previous generations and may bring some performance improvements. However, there may be a big gap compared to the latest Arm v9.2 architecture.

On the GPU side, Android Authority's sources couldn't provide an exact number of cores, but according to their calculations, the Tensor G3 will be powered by a 10-core Mali-G710 GPU Immortalis and support ray tracing technology.

In other configurations, Tensor G3 will support AV1 video encoding for the first time, upgrade with TPU processor codenamed "Rio"(running frequency of 1.1GHz), support UFS 4.0 storage, etc.

Tensor G3 is expected to be available in Google Pixel 8 phones.

▲ Google Pixel 8 rendering, from OnLeaksCTOnews.com with Tensor G3 compared to the previous two generations of CPU architecture:

Tensor CPU architecture: 2 Cortex-X1 cores at 2.8GHz + 2 Cortex-A76 cores at 2.25GHz + 4 Cortex-A55 cores at 1.8GHz

Tensor G2 CPU architecture: 2 Cortex-X1 cores at 2.85GHz + 2 Cortex-A78 cores at 2.3GHz + 4 Cortex-A55 cores at 1.8GHz

Tensor G3 CPU architecture: 1 3.0GHz Cortex-X3 core + 4 2.45GHz Cortex-A715 cores + 4 2.15GHz Cortex-A510 cores

Source: Android Authority

Source: Android Authority

Source: Android Authority

Tags: Core architecture support message light processor processing aspect source frequency unique third generation code kernel function gap performance mobile phone technology quantity Apple Docker Huawei Linux macOS MariaDB Microsoft MySQL NVidia OPPO Reno Shulou Information Docker Huawei MariaDB Xiaomi