that would likely only be present on the Max chip of the M5 generation

thanks I had always assumed it needed to be present in the base design of the chip

Soldered memory and no x16 PCIe slot on a desktop are interesting choices. Not sure who the target market is. Seems like the interconnect between boards is also pretty slow compared to Nvidia Digits or even thunderbolt 5.

Probably geared towards being a LLM workstation in a small format, similar to a Mac Studio.

Laptop chips often only have x8.

Looks like the Ryzen AI Max chips do have x16: https://www.amd.com/en/products/processors/laptop/ryzen/ai-3...

But you probably want M.2 slots over a single x16 slot.

They could have added a 16 lane upstream, 32 lane downstream PLX switch to packet switch between those devices.

PLX switches are too expensive for a consumer device. There's not enough competition in the PCIe switch market.

It's got a rtx 4070 laptop chip equivalent on chip, so most are not going to need more GPU or anything else that needs a x16 PCIe. Looks pretty nice for a small PC that's quiet, energy efficient, and don't flinch if you try to run something 3D intensive.

the cool thing about this is that it is the first mainstream PC chip with a pretty big integrated GPU + massive unified RAM pool.

See die shots here: https://videocardz.com/newz/amd-ryzen-ai-max-300-strix-halo-...

Ahhh this is so sad. So many of my favorite online spots are ending recently.

On a brighter note, Chips and Cheese are continuing the effort of quality technical journalism.

Metaverse was only mentioned 2 times.

I noticed that too, very interesting how it's not mentioned at all. Maybe Mark is getting a clue his little bet is a failure and a joke.

Reminds me of when Bill Gates wrote a book on the "Information Superhighway" (envisioned to be a MS-controlled walled garden like AOL) and then put out a revised edition a year later that replaced the term with "Internet". They still had their walled garden with Windows + IE for several years though.

you mean like github?

The message is loud and clear "As I’ve talked about efficiency this year, I’ve said that part of our work will involve removing jobs". & no surprise, "future" has only mentioned 4 times

Mr Zuckerberg, what is the future of social connection?

How many billion per mention did that cost?

And it was mentioned only as one item among others: “We do this with AI to help you creatively express yourself and discover new content, with the metaverse to deliver a realistic sense of presence, with new media formats to …” and “Our leading work building the metaverse and shaping the next generation of computing platforms ….”

Yeah, I noticed this. I wonder if it's next on the chopping block.

Thing is, I don't think that people trust FB/"meta" with the concept. Most of the feedback I've heard from people IRL has been to the effect of, "it could be cool, but not if Facebook is behind it."

Coincidentally that's also the number of times it was used today, and the number of people with any interest in it !

As someone working on mostly native iOS apps, Flutter apps, and React webpages, I’ve been looking for a solution to share common models and other utilities between clients.

Kotlin Multiplatform definitely looks like the most mature option, but I’m looking at Flutter/Dart as an option, too. Dart has support for method channels with Swift/Obj-C and Kotlin/Java plus there’s the official dart2js compiler.

I personally don’t think Flutter is the right choice for a best-in-class native mobile app and definitely not for a website, but I think there might be something worthwhile using Dart as the language of shared library like Kotlin Multiplatform.

Interesting, any resources on using Dart in this way? I also do native iOS development as a day job.

Edit: Pieced together how to do this from the official docs.

The M2 Air seems to have a considerably more comfortable design, which would be my main goal if I’m not buying a 14 or 16-inch Pro. Besides that, the Touch Bar has given me so much pain over the years. My 2018 Pro had parts of the Touch Bar OLED unit black out permanently even on a replacement model.

Glad to see this released publicly! I used it internally last year, and it was frustrating but useful (mostly relating to not having access rights and quickly changing versions).

Can you share some experiences like response times etc? Did you use Swift or the Rest API?

I used the Swift API. Response times were generally very fast, but it was also not a production build with a beta iOS version.

Apple probably chose a very niche workload to make that claim. However, these gaming tests that the article points out are pretty bad tests. You can't compare games that are running on an entirely different graphics API and running through the Rosetta translation layer.

You also can't compare games that don't run on MacOS :)

Most popular games used in a typical GPU review... simply don't have native versions to compare.

If Apple wants to look better in these reviews, they'll need to put some effort into getting native versions available.

(And yes, I understand it ultimately falls on game studios, they need the support from Apple to make it happen. Whether that's financial incentives, available libraries, or marketing that creates a market for gamers buying Apple products.)

Then look at blender which has a native metal renderer. Blender performance is also much less then advertised.