It's fine. The problem with samsung's exynos was that they charged the same price and shipped it together in the same model with the more powerful and efficient snapdragon chips. If they charged less or gave their buyers a choice in the firat place, it would have been simple.

Using off the shelf Arm IP isn't what makes the Tensor SoC interesting, it's the custom ML pipeline. If it's using an X1 of course it'll run hot and if it's on Samsung process nodes it'll run even hotter, nothing they can do about that.

Quote from: NA on August 08, 2021, 05:08:11
I work at Google and this article is very misleading.
Tensor is built by Samsung but tons of different features and components are baked into it that are made by Google. For example, so many machine learning workflows and algorithms work much faster and use much less power when running on Tensor versus Exynos chips

We'll wait for the actual product release before taking a side on this. For now, if Samsung was involved in the manufacturing, we cant expect it to run without heating up (as seen on the SD888).

I work at Google and this article is very misleading.
Tensor is built by Samsung but tons of different features and components are baked into it that are made by Google. For example, so many machine learning workflows and algorithms work much faster and use much less power when running on Tensor versus Exynos chips