Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Apple discontinues the Mac Pro

9to5mac.com

Post reply on HN
Print view View original

651–660 of 672 posts

Re: Apple discontinues the Mac Pro

#651

Post by matthewfcarlson »

asimovDev wrote: post #289
matthewfcarlson wrote: post #125

Earlier quoted context omitted.

As someone who worked on the M2 Mac Pro and has a real soft spot for it, I get it. It’s horrendously expensive and doesn’t offer much benefit over a Mac Studio and a thunderbolt pci chassis. My personal dream is that vms would support pci pass through and so you can just spin up a Linux vm and let it drive the gpus. But at that point, why are you buying a Mac? Opinions are my own obvs.

do / did you have to always work in the office or do you get to work from home by taking a test rig with you ? always been curious about this

Hardware generally isn't allowed outside of lockdowns. There are things you can do with dev fused hardware to remotely control it which make life easier. But most devs just come into the office since being there in person is nicer.

Re: Apple discontinues the Mac Pro

#652

Post by matthewfcarlson »

zeusk wrote: post #136
matthewfcarlson wrote: post #125

Earlier quoted context omitted.

As someone who worked on the M2 Mac Pro and has a real soft spot for it, I get it. It’s horrendously expensive and doesn’t offer much benefit over a Mac Studio and a thunderbolt pci chassis. My personal dream is that vms would support pci pass through and so you can just spin up a Linux vm and let it drive the gpus. But at that point, why are you buying a Mac? Opinions are my own obvs.

> My personal dream is that vms would support pci pass through and so you can just spin up a Linux vm and let it drive the gpus. SR-IOV is just that? and is well supported by both Windows and Linux.

Yes- that's what I referring to. Basically the virtualization framework supporting handing a specific PCIe device off to a VM. Link management is still handled by macOS but the actual PCIe packets are handled by the VM (which could be windows or linux, which would have a GPU driver)

Re: Apple discontinues the Mac Pro

#653

Post by dangus »

internet2000 wrote: post #20
bigyabai wrote: post #16

Earlier quoted context omitted.

> Apple’s desktop Macs actually all have modular SSD storage "Modular" does not mean that it's serviceable, repairable or upgradable. Apple's refusal to adopt basic M.2 spec is a pretty glaring example of that.

> Apple's refusal to adopt basic M.2 spec I get the ideological angle, but in practical terms that's not a barrier: https://www.aliexpress.us/w/wholesale-apple-ssd-adapter.html...

Even without an adapter there are 1st and 3rd party modules available.

Re: Apple discontinues the Mac Pro

#654

Post by CamperBob2 »

root_axis wrote: post #410
sipjca wrote: post #152

Earlier quoted context omitted.

I don’t think it’s about literally shrinking the models via quantization, but rather training smaller/more efficient models from scratch Smaller models have gotten much more powerful the last 2 years. Qwen 3.5 is one example of this. The cost/compute requirements of running the same level intelligence is going down

There are no practically useful small models, including Qwen 3.5. Yes, the small models of today are a lot more interesting than the small models of 2 years ago, but they remain broadly incoherent beyond demos and tinkering.

I don't think you can make that case for 35b and up, including the 27B dense model. A hypothetical Mac Studio with 512 GB and an M5 Ultra would be able to run the full Qwen 3.5 397B model at a decent speed, which is more like 12 months behind the current SoTA.

A lot of people got a bad first impression about the 3.5 models for a few different reasons. Llama.cpp wasn't able to run them optimally, tool calling was broken, the sampling parameters weren't documented completely, and some poor-quality quants got released. Now that these have all been addressed, they are serious models capable of doing serious business on reasonably-accessible hardware.

Re: Apple discontinues the Mac Pro

#655

Post by utopiah »

jedberg wrote: post #574
utopiah wrote: post #448

Earlier quoted context omitted.

> ...making the perfect hardware for home inference machines. I really don't get why anybody would want that. What's the use case there? If someone doesn't care about privacy, they can use for-profit services because they are basically losing money, trying to corner the market. If they care about privacy, they can rent cloud instances in order to setup, run, close and it will be both cheaper, faster (if they can affo…

I think the main use case is home automation. You don't want details of your home setup leaking out.

I've been running home automation 24/7 for few years now, using Mozilla WebThings then HomeAssistant. I indeed selected my hardware to rely on proper standards, e.g. ZigBee, so that I wouldn't have to use any proprietary app. For me it's mostly about resilience, e.g work offline, than privacy but of course I also appreciate that benefit.

... that being said, and I did read https://www.home-assistant.io/blog/2025/09/11/ai-in-home-ass... or even have use a local modal for STT but IMHO it's not good, definitely not worth buying a $4k device for. It doesn't generate good rules, it gets some commands but it's quite basic. So I understand the concept but so far I haven't seen anything good enough to warrant such a purchase for such a very very specific niche. I can imagine literally a couple of persons interested in this but I'd bet they'd also have a GPU laying around that could be available for that too.

TL;DR: technically not impossible, I doubt it's a popular use case.

Re: Apple discontinues the Mac Pro

#656

Post by aurareturn »

rudedogg wrote: post #586
aurareturn wrote: post #232

Earlier quoted context omitted.

Huh? What local LLM apis? It uses Metal.

The application development APIs, ie: https://developer.apple.com/documentation/technologyoverview...

Any serious LLM work isn't going to use that. They'll use Metal GPU. No one is going to inference using the NPU on a Mac.

OP said "work stations" which is implying Macbook Pros and Studios.

Re: Apple discontinues the Mac Pro

#657

Post by aurareturn »

root_axis wrote: post #133
readitalready wrote: post #59

Apple really dropped the ball here. They had every ability to make something competitive with Nvidia for AI training as well as inference, by selling high end multi GPU Mac Pro workstations as well as servers, but for some reason chose not to. They had the infrastructure and custom SoCs and everything. What a waste. It really could have been a bigger market for them than even the iPhone.

Nah, Apple made the right choice. Nobody except a niche market of hobbyists is interested in running tiny quantized models.

Small models keep getting smarter and local hardware keeps getting better.

At some point, they will converge and an inflection for local LLMs will happen. Local LLMs will never be as smart or fast as cloud LLMs but they will be very useful for lower value tasks.

Re: Apple discontinues the Mac Pro

#658

Post by foldr »

chirau wrote: post #527
pchristensen wrote: post #493

Earlier quoted context omitted.

$3.5k is a lot of money, but not a ton by American hobby standards. It's easy to spend multiples, even orders of magnitude more than that on hobbies like fishing, wine, sports tickets, concerts, scuba, travel, being a foodie, golf, marathons, collectibles, etc. It's out of reach for lots of people, even in developed countries. But it's easily within reach for loads of people that care more about computing than other…

I live in America, I am very well compensated. Have been for 15 years now. $3500 is a lot of money. A lot. There is a tiny bubble of us tech folks who think it is accessible to most people. It is not. It is also the same reason Macs are still a niche. Don't take your circles to be the standard, it is very very far from it, especially if you think $3500 is not a lot of money. It is easy to confirm this, just look at t…

Mac has about 15% of the market share in the US. It's not really a niche.

$3500 is more than I would spend on a hobby too, but there are, in absolute terms, a large number of Americans who can spend this much on their hobbies.

Re: Apple discontinues the Mac Pro

#659

Post by gloxkiqcza »

zozbot234 wrote: post #635
0x457 wrote: post #548

Earlier quoted context omitted.

What MoE has to do with it? Aside from Flash-MoE that supports exactly one model and only on macOs - you still need to load entire model into memory. You also don't know what experts going to be activated, so it's not like you can predict which needs to be loaded.

With proper mmap support you don't really need the entire model in memory. It can be streamed from a fast SSD, and this is more useful for MoE models where not all expert-layers are uniformly used. Of course the more data you stream from SSD, the slower this is; caching stuff in RAM is still relevant to good performance.

You can do this on a Mac as well tho, right? So that 128 GB unified memory becomes cache for very fast 1+ TB Apple SSD.

Re: Apple discontinues the Mac Pro

#660

Post by estimator7292 »

giwook wrote: post #407
sippeangelo wrote: post #394

Earlier quoted context omitted.

Some Chinese sources sell modded Nvidia GPUs with extra VRAM. They're quite affordable in comparison to even a Mac Pro.

And how much do you trust Chinese hardware?

Which of your devices weren't made in China?
Post reply on HN