Apple M5 rumors

[Jimmyjames]

In what ways? Lowering latency? Increasing speed? That all seems like standard incremental improvement stuff more so than any Apple magic sauce.

It seems like the reason AirDrop fails sometimes, is that the chipset has to sleep to save power. If Apple can make low power performance of their WiFi/bt chip better, overall reliability could improve. I don’t think this is something they can do with current bt modules.

That said, there’s another ‘next gen’ RF chip rumored for next year which I think might be a springboard for some Apple Magic - the second gen UWB chip. I could imagine UX for AirDrop that’s essentially ‘point your phone directly at someone else’s phone and flick whatever is on screen to them.’

More generally, precision location opens up a lot of opportunities for new UX. And who knows, maybe there’s some synergy between UWB and beam forming for WiFi/bluetooth…fails sometimes is that the Bluetooth chip

The second gen UWB chip came out with the iPhone 15 Pro and Apple watch 9.

 

[Jimmyjames]

From my experience with OpenWrt, AP and STA mode are different. And if an interface can act as AP, and STA, both can work simultaneously. And if I'm not wrong, Apple uses BCM WiFi and those are usually fully 802.11 compliant and can be used in routers or mesh solutions.

If you’re using OpenWRT, I assume you’re using it on a router or access point? The hardware (radios) in those are not the same as you find in client devices, eg laptops, HomePod, Apple TV.

 

[Artemis]

M5 Pro chip could separate CPU and GPU in 'server grade' chips - 9to5Mac

One of the key elements of Apple’s A-series and M-series chips is the System-on-a-Chip (SoC) design which tightly integrates all...

9to5mac.com

 

[Artemis]

Wondering how much this will impact latency and power. If Apple's doing it, presumably not enough to be a major net negative for users over the cost reductions. But this heuristic doesn't stand versus actual measurement etc, companies blow things of course.

Anyway my guess is it's measurably higher idle power and maybe also some hit to dynamic power, but whether or not that's enough to outweigh the advantage of being able to lower costs/raise the bar for the GPU size etc and improve their supply chain? I doubt it if they're doing it. Would be a great thing if chiplets at TSMC are finally good enough to go this route - wonder if we'll see it in base chips too eventually (suspect not but who knows).

 

[Artemis]

By base I mean M base. not A.

 

[diamond.g]

How would that impact the Unified Memory model?

 

[dada_dave]

How would that impact the Unified Memory model?

It shouldn’t - it would be very similar to the M1/2 Ultra if you think about it - at least in that respect.