tht

mikethemartian said:

It seems that fabs more focused on the type of research and development that universities do is more fitting for the US than mass production.

We, the USA, needs to have state of the art CMOS mass production capacity. The problem is that Intel fucked up. Options are limited. Keep Intel afloat or have TSMC/Samsung build local fabs. 

9secondkox2 said:

blastdoor said:

netrox said:

I don't think a monolithic die makes sense. I am pretty sure they will be doing chiplet designs which makes economic sense and can offer impressive performance on some tasks. 

Agreed. The Max chips are already crazy big and I imagine pretty expensive. There’s a good reason AMD went the chiplet route. The rumors of Apple going the chiplet route with M5 pro and above make a lot of sense to me.

 I hope we get an M4 ultra but I’m also starting to think I shouldn’t count on it. I really like the idea of having an M4 ultra Mac Studio sitting on my desk but having a pair of M4 pro minis could also serve my purpose. If Apple would update the whole lineup at the same time I could pick the best computer, but with a huge delay for the ultra — and not even knowing if it will appear at all — the ultra becomes a harder option to choose 

The m series max chips are actually pretty small considering they pack not only the cpu, but the GPU and ram as well. M4 max is about 100 mm. Just an Nvidia 4080 GPU is about 400 mm by itself. A 4090 is over 600 mm. Two ultra chips wouldn’t even be that big. And that’s just s GPU. 

Apple can level the playing field and come out on top of literally every player with a monolithic die. The potential of the Mac Pro is seriously wasted currently. The equivalent of combining two ultra monoliths would hurt some serious feelings. The only barrier would be apathy on the part of apple. 

You have your die sizes mixed up. 

There isn't anything to crow about here, either from Apple or AMD. Transistor budgets roughly double every 2 years or so. Having more and more dedicated function units added to the "CPU" has been a process that is ongoing for the past 40 years. NPUs are just the latest thing, with a lot of hype. In another time, they'd just be called specialized SIMD units.

And, AMD has a lot of experience putting higher performance GPUs into an SoC. They provided the SoCs for both PS5 and XBox X-series, or in their language, APUs. There's a lot of technical heritage there.

Management likely needs to be convinced to make a high performance x86 APU for PC laptops, though. I'd bet AMD managers are thinking it is crazy to spend a lot of R&D on a high performance APU or SoC in the laptop market. AMD seems to be failing or having a down turn in this market. It's a hyper mature market where the vast majority of laptops are neutered $600 devices that doesn't require much of a GPU at all. They are competing for a pretty small slice with very little margins.

AMD is way way behind in the AI accelerator market. These are like $5000 modules that has been in hyper demand for the past couple of years, and is slated to grow. These are really high margin products. AMD not trying to cut into that market seems like a troublesome path. Those margins will pay for everything else. 

nubus said:

thrang said:

Umm, so we really want the EU or other governments to start directing product development? 

At times we do. The car industry didn't remove lead from gas. The tobacco industry invented nicotine products with watermelon flavor for kids. We got baby powder with asbestos as it was cheaper to produce. Food with carcinogenic ingredients. Building materials that burst into flames. Would you say Boeing did improve quality after government let Boeing control Boeing? That banks could do better without regulation?

This time it is about Apple producing e-waste and letting taxpayers pay. Apple agreed 15 years ago to fix it but did nothing. What would you have done?

Mandating charging standards is great. I'm all for it. Wish there were standards for battery storage, combiner boxes, EV chargers, plug-in solar+storage, etc. Also, getting rid of the middlemen and red tape is great too. Those two have to work together to become a force multiplier.

The EU's USBC mandate isn't doing anything regarding e-waste. In terms of e-waste, only difference is that what used to be Lightning cable e-waste in the EU has now turned into USBC cable e-waste. It won't be less wasteful until phone companies stop putting USBC cables in the box. Apple, who is the most likely to stop putting USBC cables in the box, probably will ship a USBC cable in the box for the next 3 to 5 years. The higher the cost Apple product, the longer it will ship with a USBC cable.

For any high cost product, a USBC cable is going to be in the box for the foreseeable future. Some low cost products may not come with USBC cables in the future, but that is a long road ahead. I just recently bought, and returned, a Bluetooth Keychron mouse for $40. It came with 2.4 GHz wireless USBA transceiver, a USBC cable, a USBC C to A adaptor, and a USB A to C adaptor. It was intended to be Bluetooth mouse.

The prevalence of wireless 2.4 GHz peripheral connections in the PC world is basically driven by Bluetooth not being available on all PCs in the past; and, Windows' Bluetooth UI probably sucks, and it is easier to just plug in the Wireless-G USBA dongle for wireless mice than to connect through Bluetooth. It seems every Bluetooth mouse for Windows comes with 2.4 GHz dongles. Meanwhile, virtually every computing device has Bluetooth! Argh!

Anyways, to reduce e-waste, gov'ts have to mandate OEMs to take back their used stuff and recycle them. The EU has something like this, but they need to be stronger measures. Or, have an actual gov't recycling program that they do themselves. Not ship and dump to some other country with a promise of recycling, but it needs to be done in our own backyards.

9secondkox2 said:

Doesn't make sense. It's already server-grade packaging. Perhaps better than other server-grade packaging.

I can think of 3 reasons this would make sense:

1. If Apple is combining their silicon with third party chips from Nvidia or semthing. 

2. Alternatively, another way it could make sense is if Apple is looking to add more GPU cores to various iterations of its chips, without increasign CPU core counts. i.e. having multiple sets of max/Ultra chips - one set for laptops, one for Mac Studio, and another for Mac Pro.

or...

3. Appel could be redoing the way it tiers its chip lineup. CPU could be the same, but GPU would be different for each tier. 

Interesting to see how this develops. 

As I understand it, the maximum sized chips that can be made with these more and more advanced nodes is decreasing.