Bad news for anyone who buys the M5 MacBook Pro as an "AI" machine and finds it can't fit any of the more interesting LLMs!
At this point, I get the soldered on ram, for better or worse... I do wish at least storage was more approachable.
Exceptions apply to those running local LLMs.
Their sales copy for reference:
"M-series chips include unified memory, which is more efficient than traditional RAM. This single pool of high-performance memory allows apps to efficiently share data between the CPU, GPU, and Neural Engine.... This means you can do more with unified memory than you could with the same amount of traditional RAM."
Still not the fastest ram, that they use for dedicated GPUs, but faster than most x86 options.
I’m not trying into be a fanboy and maybe it’s a little bit “cope”, but apple has always put as much RAM as is necessary for the computer to work—and not a lot more—in their base models.
:)
You're not silly, you're just able to see reality.
Apple knows who is buying the bulk of their computers, and it isn't power users ... most people buying computers don't have a clue what RAM is even used for.
I'd hit beachballs, but macOS balances 8GB of RAM fine even with Tahoe for regular users
I know people complain at every release. But I look at the three choices presented and they are all disappointing to me. It's a huge turnoff to see the only initial differentiator presented to be a choice between "measly" amounts of RAM and storage to "barely acceptable" amounts.
To get even close to the specs on my Surface Pro I'd have to hit the configurator and spend at least $1000. Even more to hit the config of my work issued HP notebook.