Live data from Hacker News

macOS 26.2 enables fast AI clusters with RDMA over Thunderbolt

developer.apple.com

291–300 of 304 posts

Re: macOS 26.2 enables fast AI clusters with RDMA over Thunderbolt

#291

Earlier quoted context omitted.

Man you positively demolished that straw man. How much as a base model MacBook Air changed in price over the last 15 years? With inflation, it's gotten cheaper.

It's also gotten cheaper nominally. I just got a new base MBA for $750. Kinda surprised, like there has to be some catch.

Also, the MBA vs MBP lineup is different now. MBP was the default choice before even for students, so MacBooks sorta started at $1300. Now the MBA is decent, and the MBP is really only for pros who need extra power and features.

Re: macOS 26.2 enables fast AI clusters with RDMA over Thunderbolt

#292
post #235

Earlier quoted context omitted.

Why? AWS manages to do it ( https://aws.amazon.com/ec2/instance-types/mac/ ). Smaller companies too - https://macstadium.com Having used both professionally, once you understand how to drive Apple's MDM, Mac OS is as easy to sysadmin as Linux. I'll grant you it's a steep learning curve, but so is Linux/BSD if you're coming at it fresh. In certain ways it's easier - if you buy a device through Apple Business you can h…

Where the in the world are you working where MDM is the limiting factor on Linux deployments? North Korea? Macs are a minority in the datacenter even compared to Windows server. The concept of a datacenter Mac would disappear completely if Apple let free OSes sign macOS/iOS apps.

I’m talking about using MDM with Mac OS (to take advantage of Apple Silicon, not licensing) in contrast to the tools we already have with other OSes. Probably you could do it to achieve a large scale on prem Linux deployment, fortunately I’ve never tried.

Re: macOS 26.2 enables fast AI clusters with RDMA over Thunderbolt

#293

Earlier quoted context omitted.

eGPU with an Apple accelerator with a bunch or RAM and GPU cores could be really interesting honestly. I’m pretty sure they are capable of designing something very competitive especially in terms of performance per watt.

Really, that’s a place for the MacPro: slide in SoC with ram modules / blades. Put 4, 8, 16 Ultra chips in one machine.

You honestly don’t need extra CPUs in this system at some point do you?

Re: macOS 26.2 enables fast AI clusters with RDMA over Thunderbolt

#294
post #252

Earlier quoted context omitted.

You get occasional accounts of 3090 home-superscalers whereas they would put up eight, ten, fourteen cards. I normally attribute this to obsessive-compulsive behaviour. What kind of motherboard you ended up using and what's the bi-directional bandwidth you're seeing? Something tells me you're not using EPYC 9005's with up to 256x PCIe 5.0 lanes per socket or something... Also: I find it hard to believe the "performan…

I love your skepsis of what I consider to be a fairly normal project, this is not to brag, simply to document. And I'm way above 3 kW, more likely 5000 to 5500 with the GPUs running as high as I'll let them, or thereabouts, but I only have one power meter and it maxes out at 2500 watts or so. This is using two Xeons in a very high end but slightly older motherboard. When it runs the space that it is in becomes hot en…

Thanks for replying, and your power story does make more sense all things considering. I'm no stranger to homelabbing, in fact just now I'm running both IBM POWER9 system (really power-hungry) as well as AMD 8004, both watercooled now while trying to bring the noise down. The whole rack, along with 100G switches and NIC/FPGA's, is certainly keeping us warm in the winter! And it's only dissipating up to 1.6 kW (mostly, thanks to ridiculous efficiency of 8434PN CPU which is like 48 cores at 150W or sommat)

I cannot imagine dissipating 5 kW at home!

Re: macOS 26.2 enables fast AI clusters with RDMA over Thunderbolt

#295

Earlier quoted context omitted.

Really, that’s a place for the MacPro: slide in SoC with ram modules / blades. Put 4, 8, 16 Ultra chips in one machine.

You honestly don’t need extra CPUs in this system at some point do you?

They are inseparable for Apple. CPUS/GPUs/memory. They can use chipsets to tweak ratios, but I doubt they will change the underlying module format—everything together.

My suggestion is to accept that format and just provide a way to network them at a low level via pci or better.

Re: macOS 26.2 enables fast AI clusters with RDMA over Thunderbolt

#297
post #126

Earlier quoted context omitted.

A home PC has to have a SOTA gpu?

Probably upset that the high-end video game "hobby" costs more than it used to. Used to be $1-2K for the very best gaming GPU of the time.

I mean, yes. Very much so. People should be upset about a relatively affordable hobby getting to this point.

Re: macOS 26.2 enables fast AI clusters with RDMA over Thunderbolt

#299

Earlier quoted context omitted.

> You don't need to be a genius or a billionaire to realize that when most of the global supply of a product becomes unavailable the remaining supply gets more expensive. Yes. Absolutely correct if you are talking about the short term. I was talking about the long term, and said that. If you are so certain would you take this bet: any odds, any amount that within 1 month I can buy 32gb of new retail DDR5 in the US fo…

At this point I can't tell if you are arguing in bad faith, or just unfamiliar with how prime works. Just in case: You have cited the cost of prime for a full year. Oh for the love of fuck. I don't subscribe to Prime or pay any attention to how it's priced. I've gotten offers for free trials of Prime before, should I just ignore that for most people Prime is something they have to pay for?

Bear with me:

Back at the start of our discourse, you tried to prove that a sub $500 computer can't exist by citing that B+H is selling sticks of 32gb of DDR5 for $384 while ignoring, or failing to find that they had 2 x 16 kits of Crucial branded RAM for $270 in stock when you made the claim. That deal is gone now at BH, but Best Buy has the exact same kit for $260 in stock right now (https://www.bestbuy.com/product/crucial-pro-32gb-2x16gb-ddr5...). Rather critically, the original post did NOT make any claims about the price of a 32GB RAM stick, it made a claim about the existence of a You have said elsewhere that "a single entity bought more than 70% of the wafer production for the next year. That's across all types of memory modules." Based on reported industry rumors and press releases, OpenAI have made a non-binding agreement with two foundries that control 70% of the DRAM market for up to little more than half their output of raw DRAM wafers. That is a massive difference from buying 70% of the entire RAM market. Its a letter of intent for both foundries, and is very nebulous about that "up to" phrasing. Both of those deals was called a "Letter of Intent" by the foundry very specifically. If you aren't familiar, that specific phrase is typically used for a non-formalized agreement that has no legally enforceable provisions. No actual deal has been inked. I can understand how a misread happened, but not how you have such strong feelings on a story that you haven't understood the most basic details for. To summarize: not a deal in the legal/enforceable sense, not 70% of the global RAM or DRAM market, not all types of RAM, not even a firm commitment on the 40% of DRAM market from either party.

You say "I don't subscribe to Prime or pay any attention to how it's priced", but you somehow arrive on the only pricing option of the ~4 available options that undercuts what I am saying. I had to scroll to below the fold on google for "prime price" to get a single link that did not mention the lower monthly price in the search result. Even the google AI got it right. Yet, I am to give you the benefit of doubt that you have such specific, yet also profoundly limited, knowledge of the Prime program that you can cite the exact price of the yearly version of the product to prove a point, yet have no idea that a monthly subscription exists. A curiously specific ignorance.

There's more, but I'll move on.

I'm happy to accept your plea of ignorance, but it severely undercuts your arguments when you have to plea ignorance on the facts at the root of your arguments continually. It severely undercuts your plea of ignorance when every single number and fact you misquote happens to be an error in favor of your argument. People making mistakes in good faith tend not to make every single error in their own favor.

The worst part is, the position that RAM pricing will not drop has merit and is very arguable (although I don't agree given what I have seen so far). It is NOT a good thing for DRAM prices that OpenAI might have first dibs on a sizable minority of next years DRAM production. It is also not at all a given that that will be true. Continually using deceptively cherry picked, or outright wrong, numbers and info means that this conversation won't continue. I must insist on basing arguments in fact.

Thanks for the back and forth, for what it's worth. In any case, its always enlightening to get a good peak into how different people interpret numbers and facts, and arrive at their understanding of the world.

Re: macOS 26.2 enables fast AI clusters with RDMA over Thunderbolt

#300
post #294

Earlier quoted context omitted.

I love your skepsis of what I consider to be a fairly normal project, this is not to brag, simply to document. And I'm way above 3 kW, more likely 5000 to 5500 with the GPUs running as high as I'll let them, or thereabouts, but I only have one power meter and it maxes out at 2500 watts or so. This is using two Xeons in a very high end but slightly older motherboard. When it runs the space that it is in becomes hot en…

Thanks for replying, and your power story does make more sense all things considering. I'm no stranger to homelabbing, in fact just now I'm running both IBM POWER9 system (really power-hungry) as well as AMD 8004, both watercooled now while trying to bring the noise down. The whole rack, along with 100G switches and NIC/FPGA's, is certainly keeping us warm in the winter! And it's only dissipating up to 1.6 kW (mostly…

I stick the system in my garage when it is working... I very enthusiastically put it together on the first iteration (with only 8 GPUs) in the living room while the rest of the family was holidaying but that very quickly turned out to be mistake. It has a whole pile of high speed fans mounted in the front and the noise was roughly comparable to sitting in a jet about to take off.

One problem that move caused was that I didn't have a link to the home network in the garage and the files that go to and from that box are pretty large so in the end I strung a UTP cable through a crazy path of little holes everywhere until it reaches the switch in the hallway cupboard. The devil is always in the details...

Running a POWER9 in the house is worthy of a blog post :)

As for Frankie: I fear his days are numbered, I've already been eying more powerful solutions and for the next batch of AI work (most likely large scale video processing and model training) we will probably put something better together, otherwise it will simply take too long.

I almost bought a second hand NVidia fully populated AI workstation but the seller was more than a little bit shady and kept changing the story about how they got it and what they wanted for it. In the end I abandoned that because I didn't feel like being used as a fence for what was looking more and more like stolen property. But buying something like that new is out of the ballpark for me, at 20 to 30% of list I might do it assuming the warranty transfers and that's not a complete fantasy, there are enough research projects that have this kind of gear and sell it off when the project ends.

People joke I don't have a house but a series of connected workshops and that's not that far off the mark :)

Post reply on HN