Hacker News new | ask | show | jobs
by ThePhysicist 1 day ago
Apple RAM prices always had quite a bit of margin though, I think they charged around 4x the going market rate per GB (that said you can't fully compare their RAM to a loose DIMM stick). I was planning to pick up a new Mac Studio this autumn, now I'll have to see if I can afford it, though I have been spending 1,000 USD on LLM subscriptions in some months so I guess even a 10,000 USD Studio Mac amortizes quite fast if it allows me to run coding models locally.
5 comments

The funny thing is that currently Apple's RAM upgrades are cheaper than the loose dimm sticks.

I've just got a new MBP this month, because I expect the prices to rise significantly with the new macbooks in the fall.

I would wait until the next Mac Studio. Rumors are it will have 768GB max memory and with the M6 Max chip, even faster prefill. I feel like it’s the endgame.
Yeah for like $25,000+
Probably will be but I have a feeling the days of $20 and $100 coding plans are winding down. Now that people are so used to using agents, they will raise prices and tighten their grip. At $20k, I might be tempted.
That's an expensive looking 'if'.
> you can't fully compare their RAM to a loose DIMM stick

Why?

https://macmagazine.com.br/wp-content/uploads/2020/11/18-tea...

This image shows it best.

The memory is on the same package as the SoC. (It's not on the motherboard.)

IIRC modern Apple devices integrate the memory into the whole SoC instead of making it separate on the board and replaceable. It's definitely not swappable like a DIMM or CAMM module would be. Can't find a photo of a decapped M4 chip to prove it, though...
Check my sibling comment for a photo link (M1); or here:

https://macmagazine.com.br/wp-content/uploads/2020/11/18-tea...

Its not integrated to the SoC, it is soldered to the mainboard though.
The memory is not integrated into the SoC die itself, but it is packaged alongside the SoC rather than being separately mounted.
At Apple scale I would have thought that makes it cheaper, not more expensive.
there's no local AI model that comes even close to the ones you get access to by paying 1000 USD per month
The large models are really close in my experience. Just slower.