Tensor G2 Is a 5nm Processor Just Like Its Predecessor: How Is It Different?

Google's latest flagship smartphones -- the Pixel 7 and the Pixel 7 Pro draw power from the in-house Tensor G2 processor, co-developed by Samsung. Several reports claimed that the Tensor G2 is fabbed by Samsung using a 4nm process. However, a Google spokesperson has confirmed otherwise, stating the Tensor G2 is fabbed using a 5nm process.

As of now, there is no clear information if the Tensor G2 is based on Samsung's 5LPE or if it is using the slightly improved 5LPP manufacturing process. In either case, the power efficiency difference between the two devices will be very negligible.

Official statement from Google's Spokesperson

"We purpose-built Google Tensor G2 for real-world use cases. Our final architecture, which includes 5nm, helped us reach that goal while increasing both performance and power efficiency. This approach also allowed us to add new capabilities while taking a step forward on machine learning with our next-generation TPU with G2"

Google Tensor G2 Details

The Tensor G2 is the latest flagship SoC from Google, which currently powers the Pixel 7 and the Pixel 7 Pro, and the same chipset will also power the upcoming Pixel Tablet. The Tensor G2 has an octa-core CPU design with two high-performance cores (Cortex-X1), two medium-performance cores (Cortex-A78), and four efficiency cores (Cortex-A55).

The SoC comes with the improved Mali-G710 MC10 GPU. Although the Tensor G2 uses the same Cortex cores as its predecessor, the Tensor G2 actually comes with a higher clock speed, delivering improved single-and multi-core performance.

Did Google Cheap-Out Or Is It A Sensible Decision?

Samsung's 4nm 4LPE node is a tried and tested manufacturing process. In fact, processors like Exynos 2200 and the Snapdragon 8 Gen 1 are actually fabbed using the same. However, these processors had some issues when it comes to heat management which can be linked to Samsung's 4nm manufacturing process.

Qualcomm hence shifted Snapdragon 8+ Gen 1 manufacturing to TSMC's 4nm process. As we have seen with the reports, the Snapdragon 8+ Gen 1 when compared to the Snapdragon 8 Gen 1 is not just powerful, but it is also power efficient.

It is possible that Google may have attempted to save a buck by manufacturing the Tensor G2 using an older 5nm node, as newer nodes are always expensive. However, we can't ignore the fact that Google might have also considered the poor thermal performance of Samsung's newer 4nm process and instead decided to use the older but reliable 5nm process to strike a balance between thermal performance and efficiency.

Source