Hacker News Reader

Comment by eagleinparadise | Hacker News Reader

Anyone know when to expect the M5 Pros? I am on a base 16gb M1 and struggling hard in daily workloads. I am often running at 20gb of swap memory usage.

I don't really use local LLMs but think 32GB RAM would be good for me... but I am so ready to upgrade but trying to figure out how much longer we need to wait.

View on HN

First rule of mac world is get the most memory you can afford.

I got the cheapest m1 pro (the weird one they sold thats binned due to defects) with 32gb ram and everything runs awesome.

Always get the most ram you can in mac world. Running a largish local LLM model is slowish but it does run.

A mac out of memory is just a totally different machine than one with.

probably because most of the devs building the software are on the highest ram possible and there is just so much testing and optimization they dont do.

Part of me misses my OG base 14" M4 Pro. The battery on that thing was absolutely phenomenal - literal 12-14+ hours of real-world use. Not so much on the 14" M1 Max (64GB) that I upgraded to after about 2 yrs.

'Real-world idle' efficiency on the newer chips is the main reason I've got the (slight) itch to upgrade, but 64GB+ MBPs certainly don't come cheap.

Which is crazy because I recently moved to an M1 pro and the idle battery was at least twice as good as any windows x86 machine I've ever used.

IIRC on macbooks, memory is actually not that important because of the insanely fast SSD and swap. Though LLMs are probably a different story

There was a 6 month gap between the M4 and M4 Pro, so maybe a while.

This gap makes no sense to me. I wonder if Apple is just leaning into this cycle because it's easier to make M5s than more advanced processors, so you can sell this sooner?

From a buyer's perspective, I don't like it at all.

Different Chip SKUs are often a TON of work. By trying to release all of them at the same time, you'll have a chip pipeline where you need tons of work, all at the same time, all in the same stages of the process. By staggering them, you spread this work out across the year.

M series chips are ridiculously massive, as Apple apparently does not want to transition to chiplets, so they can’t easily compose CPU. Thus refining the process and improving yields on the smaller parts probably makes sense.

As an other example the current ultra part is the M3, and it was released early 2025, after even the M4 Pro/Max, and a good 18 months after the M3 was unveiled. We might not see an M4 Ultra until 2027.

Rumors suggest they might be early next year, or likely by spring.

Reply on news.ycombinator.com