[xyzsparetimexyz]

So the argument of this _very_ Claude written piece is that.. On apple devices you have GPUs with fast-enough unified memory and can link them together?

Okay, fair, but why would I get an Apple device when I can get a Linux device with some monster motherboard that has a dozen memory slots?

[cyanbane]

I submitted this ( I am not the author ) because I thought it was interesting take, I honestly didn't think it was AI, maybe AI polished, but not AI. Was surprised when it was flagged and I saw over on HN Removals . Ran through Claude and got this back:

"Overall, my honest read: this feels like a human-written opinion piece (possibly AI-assisted for polish/editing, which is extremely common now), not a fully AI-generated article. The specific personal history ("I've been saying this in group chats for months"), the running joke about being right before everyone else, and the reader engagement in the comments (with the author replying) all point to a real person with a strong personality behind it — though I can't rule out that Claude or another model helped tighten the prose."

It is interesting that everyone was so very certain it was AI written. Just read your comment about it being Claude based, now I am trying to figure out if Claude is trying to recuse itself in some way by giving me those results.

Still feel like the point it interesting and going to keep diving into it and I am surprised it was flagged. May have to up my reading on Removals more and more.

[brookman64k]

I don’t think that a generative model is a good judge to assess if a text was written by itself. This article has a super stong “written by AI” smell to it. Once you notice all the em dashes and “it isn’t X, it’s Y”patterns you can’t unsee it. That plus the sloppy images, make it really hard for me to finish reading.

[smw]

Because bandwidth between those sticks and the GPU isn't fast enough for fast inference.

[xyzsparetimexyz]

Even with an integrated gpu? Surely it'll be faster that using thunderbolt cables

[cyanydeez]

both AMD and Apple hit the unified memory goal. but its the bandwith thats bottlenecking their performance and onboarc memory is even worse.

[xyzsparetimexyz]

Is Apple in a better position than Amd as this article seems to claim?

[wtallis]

AMD's Strix Halo has a 256-bit memory bus. Apple's Max class chips have a 512-bit memory bus, Ultra class chips have a 1024-bit memory bus. Unless you want to go for datacenter hardware, AMD's offerings simply aren't in the same league in terms of memory bandwidth.

[xyzsparetimexyz]

I see, interesting thank you

[cyanydeez]

depends: if they can keep getting larger memory, yes

[adabyron]

This made me go down a rabbit hole.

Looks like Mac Minis had a 64 GB model (hard to get now) for roughly $1,600. You could get a 3 of these (192GB) & thanks to the Thunderbolt 5 connection, they out performed any single board AMD for a similar price that had more memory. If you tried to do so something similar with Nvidia GPU you're looking at a lot of watts, cooling & a much faster experience but only 48 GB of memory.

[bigyabai]

> only 48 GB of memory.

You can layer the LLMs across multiple GPUs, you can absolutely have 192gb of Nvidia dGPU memory with the right MOBO and CPU combo.

[adabyron]

I was trying to compare equal in price