@AndrewReid_EOSHD

Haven't Intel and NVidia been playing the same tricks for years?

No, not same. And they did not have such wide, extremely coordinated and tightly controlled PR things.

Last Nvidia PR releases are similar, but much smaller in scale and not controlled after around 7 days.

Apple controls everything (for top influences) for months.

In real software I am all for optimisations if it improves genuine performance... With all new architectures there's the most room for new optimisation.

One of issues here is that Apple PR tell about ARM cores performance while up to 40% of performance are actually using GPU parts and public or secret coprocessors.

The M1 architecture is all about coprocessors though, if most of the die is sitting idle and you're only using the ARM cores, the software isn't optimised enough. Good software engineers will use the hardware available on the chip.

Yes the marketing is misleading though and should make clear what the GPU utilisation is when talking pure CPU benchmark figures.

As for user perspective, I care most about how well my specific software runs for certain tasks - and synthetic benchmarks are rarely a good indicator of that anyway, either on Intel or Apple.

@radikalfilm

I posted info from another topic, it is from reversers.

Apple libraries silently use this coprocessor, but it is not documented and mentioned anywhere, official APi is absent.