Tensor G4 Running In Google’s Top-End Pixel 9 Pro XL Throttles To 50 Percent Of Its Maximum Performance, Indicating Poor Thermal Efficiency

Google’s executives praised the latest Tensor G4, which is featured in every single one of the newly announced Pixel 9 models, claiming better raw and AI performance, coupled with incredible power efficiency. However, these attributes were not on display in the latest stress test involving the Pixel 9 Pro XL, as the chipset throttled to 50 percent of its maximum performance.

New stress test also shows that the efficiency cores belonging to the Tensor G4 downclocked to as low as 0.57GHz due to thermal throttling

During Google’s Pixel 9 event, it was revealed that except for the base model, the Pixel 9 Pro, Pixel 9 Pro XL, and the Pixel 9 Pro Fold were outfitted with a vapor chamber, which would help control the Tensor G4’s thermals, resulting in better performance. This upgrade was one of the most notable changes the company introduced this year and a welcome one, given that last year’s Pixel 8 and Pixel 8 Pro suffered from severe overheating issues when running stress tests.

Related Story Google Pixel 9 And Pixel 9 Pro Are Official With A New XL Model, Feature Improved Cameras, Tensor G4 Chip, And More

Unfortunately, on this occasion, @callmeshazzam shows that the Pixel 9 Pro XL, Google’s top-tier and biggest model, cannot prevent the Tensor G4 from overheating. The stress test shows that the SoC’s performance cores dropped to 1.32GHz from 3.10GHz, with the efficiency cores slipping from 1.92GHz to 0.57GHz. This is a disappointing showing from the Tensor G4, but before readers start bashing Google for charging $1,099 for a flagship with sub-par performance, there are a few things to consider.

Firstly, we do not know the ambient temperature in which the Pixel 9 Pro XL was tested in, as that has a major effect on the chipset’s ability to prevent thermal throttling for as long as possible. Additionally, we reported a while back that even MediaTek’s Dimensity 9300, a chipset fabricated on TSMC’s 4nm N4P process, lost 46 percent of its maximum performance when running the