Google has been using its specially designed Tensor chips to power Pixel smartphones since the introduction of the Pixel 6. However, until now, these chips have not matched the performance or efficiency levels of their rivals.
Upcoming Improvements
This situation might soon improve, as Android Authority has obtained leaked internal documents that outline Google’s intentions with the new Tensor G6 chip, codenamed “Malibu.” The primary focus for the Tensor G6 seems to be on enhancing thermal management and efficiency. Major performance upgrades don’t appear to be expected, and even the Tensor G5 is only anticipated to deliver a slight increase in performance over the current chip.
Changes in Strategy
It appears that Google’s system-on-chip (SoC) strategy is not currently aimed at achieving significant performance enhancements. In fact, according to Android Authority, the Tensor G6 chip might experience some "downgrades" in certain aspects.
GPU Decisions
For example, it is reported that Google will not include a new GPU with the Tensor G6, instead opting for an older design that was initially meant for the Tensor G4. If this is true, the Tensor G6 might miss out on ray tracing support, which the Tensor G5 is expected to offer. This could be considered a step backward in certain respects.
The choice to use the G4’s GPU in the G6 is thought to be driven by Google's aim to minimize the chip’s die area. Reports indicate that the Pixel 10’s chip will be relatively large at 121 mm², built using TSMC’s 3nm process, while Apple’s A18 Pro, which utilizes the same process, measures only 105 mm².
Google is targeting a die size of 105 mm² for the Tensor G6, aiming for a 4% area reduction from TSMC’s newer N3P process node (the G5 is built with N3E). Using the G4’s GPU is expected to help decrease the die area by around 12%. Additionally, Google is reportedly saving space by cutting down the DSP (digital signal processor) by one core and reducing the SLC (system-level cache) to 4MB. However, it seems that there will be no changes to the CPU size.
In conclusion, Google is concentrating on creating smaller, more efficient chips, prioritizing space-saving measures over performance advancements.
