I knew “scary fast” had to mean some sort of processor bump for hardware, but I was secretly hoping they’d kill off the remaining Lightning ports on their keyboards, trackpads, and mice.
And I was hoping they’d finally redesign that god awful mouse. I don’t know how people live with that thing.
From Wikipedia: The term "3 nanometer" has no relation to any actual physical feature (such as gate length, metal pitch or gate pitch) of the transistors.
I built a M3 MBP just to see how much money a mxed out unit would be.
M3 14" MBP
Max chip with all the cores
128GB RAM
8TB storage
$4700
That's about the cost of my last MBP and iPhone pair, two times over. At that point, why even go for a laptop, vs. what would clearly be a high end desktop station?
they are. (unless we're talking about Mac only, which are not as repairable or upgradable)
desktop allow for a far better modularity, and reparability, far more ports, PCIe expansions like sound cards, etc.
If my screen breaks or I'd rather use a bigger one, I just buy a monitor and plug it in. If my CPU dies or is no more enough for my use case, I'll just buy a better one while still using every other component. If I need more hard drives, I'll just buy more SATA cables. If I need better sound, I'll buy a sound card.
those features are dealbreakers. laptops will never be able to compete with a real desktop.
The memory maximums are going to be more and more important when it comes to local AI applications.
Take language models for an example
To run a 30b model, you need 24gb of video ram to do it fully on the video card. That's a nvidia 3090 or 4090 today. But in the grand scheme of things, 30b is small. They are going to get much bigger, especially when you want larger contexts which allow the AI to remember more about its interactions with you.
Apples memory is unified, so it can be system ram, or video ram. You'll be able to easily load a 70b model into a MacBook with 64gb of ram for example, where you'd need 2 3090s or 4090s and a hefty PSU on a current Gen non Mac PC (if you even can with just that)
For the moment, things are better optimized for windows and nvidia hardware, but Apple is encroaching on this space, and their huge amounts of video memory will begin to unlock using and training larger and larger models with each hardware generation.
Expect to see nvidia starting to offer higher video ram cards as well for this exact reason. Maybe even cards tailored to that instead of gaming with really high amounts of ram.
I can't see local models or hardware needing to scale much past the sizes we already have. Recent models like mistral have shown that we are still far from saturation at current model sizes.
Doesn't the m2 max allow 196gb of ram? Seems like an odd downgrade. The value in these for me is the unified memory for large ai models, but most consumers may not notice that. Who knows.
Based on my 60 seconds reading on it, onboard GPUs typically share the systems RAM. It is usually a fixed amount from my understanding. Dynamic caching seems to allow the GPU to only consume what it needs. Without knowing more, I'm guessing this means it frees up more RAM for the system instead of holding a fixed chunk in reserve for the GPU, or, on the other side, allows the GPU to use more RAM than some predetermined fixed amount.
According to Apple's press release, the GPUs in the new Macs are already faster and more efficient than those that came before them. But they go further thanks to their support for Dynamic Caching, a feature that "unlike traditional GPUs, allocates the use of local memory in hardware in real time."
What does that mean? Apple says that "with Dynamic Caching, only the exact amount of memory needed is used for each task. This is an industry first, transparent to developers, and the cornerstone of the new GPU architecture."
pity qualcomm has just wiped the floor with them. even with the new M3 it's not even close I believe. please correct me if I'm wrong (who am I kidding you're gunna correct me)