John Siracusa

That is a big thing, presumably due to improvements in the, I forget what they call it, like the neural whatever crap that they put into their GPU cores in the previous generation.

Yeah, they put some processing stuff in each GPU core that is useful for doing like machine learning AI stuff.

And I guess they improved that in... I don't know if they improved it or have more of them.

But anyway, Apple is making big claims about the GPU compute performance.

And when they say previous generation, I'm assuming they mean like the M5 Max versus M4 Max.

So that's a big number.

And that is significant because...

if you think about where, what kind of CPU or, you know, what kind of GPU bound workloads are people doing these days, AI is right up there, local, you know, on machine AI.

So, I mean, we'll have to see what the benchmarks bear out and how realistic that 4X number is, but it is the standout number in everything here.

Yeah, that's how they sell them.

I know.

But here's what it comes down to, is what Marco was saying.

The old M4 was 8 and 4, 8 big, 4 little.

The new arrangement is 5 and 10.

So it's pretty much flipped.

It used to be double the number of really good ones to the not as good ones.

Now it's the opposite.

It went from 8, 4 to 5, 10.

So, you know, it's really just total inversion.

And that's why the naming makes sense because, look,