Evolution of the Samsung Exynos CPU Microarchitecture [pdf]

Snapdragons are generally more performant with less consumption. Apart from obvious battery life effect, they tend to produce slightly better pictures due to heavy image processing - basically more data available within timeframe available for a snap, without annoying the user with slowness.

That being said, internet rumors are that over time samsung's phones on exynos tend to remain more snappy with various updates compared to snapdragon ones. I doubt anybody did serious comparison, I have s22 ultra with exynos (no other choice in europe) and so far so good.

Samsung is continuing with Exynos, but they're using ARM Cortex architectures nowadays, as opposed to the in house Mongoose Mx architecture described in this doc (the M6 never made it into a product, and the previous M5, M4, M3 cores were never any better than the contemporaneous ARM competition).

Qualcomm also just uses ARM Cortex nowadays, rather than in-house designs.

Their main issue is not even the design, but manufacturing process - Samsung's "5nm" process for example was underwhelming compared to its TSMC counterpart.

100% of the CPU design team at Samsung in Austin was laid off in 2019. That's why none of the authors in this 2020 paper have a Samsung email address.

I didn't fully read the paper you linked, but I think at a high level the difference is this: the Exynos approach only encrypts the data stored in the branch predictor, while the Arm approach additionally encrypts the index.

The Exynos approach means that if the attacker can find a branch of its own that has a hash collision with the desired victim branch during the course of the victim's lifetime, it can still perform a cross-training attack (however, if the victim exits and is relaunched then the mapping will change and the attacker must start over). This is perhaps unlikely, but the Exynos paper only suggests as a mitigation that the OS periodically change the key, at the expense of mispredictions (see the last paragraph of Section V).

The Arm approach solves this by using a "light-weight set update mechanism" that allows the hardware to automatically change the key periodically without incurring as much overhead. I'd have to read the paper more carefully to understand exactly how it works though.

They couldn't without some Rosetta-like layer. It'd lock their phones out of too many apps.