I think Apple would love to show off the next level of their transition at WWDC. I think it will be the "Q" series.
There is no chance Apple would name a processor or product Q-anything. The letter is ruined.
The MBPs are the "developer machines". So it would fit to show them at WWDC. And then maybe throw in some Mac minis and iMacs for good measure.
This is the best reason to announce updated MBPs. Minis are also development machines--I build on one. I could see higher end iMacs, the recent update seemed a little incomplete at 24".
But I actually think we are more likely to see a Mac Pro for reasons I'll share below.
They are both the same architecture, just how M1, M1X, M2 are all on the arm architecture. Developers may implement new features found in the new OS, but unless they reveal some new hardware in the chip design that we haven’t seen before on A14 or M1 that requires specific code for that hardware (Unlikely), there’s really no need for this.
This is correct, there is no need for a new DTK for an A or M series chip.
However:
I have a theory that Apple could release one or more application-specific discrete accelerator cards under the
Apple Afterburner card product line that was
kicked off in the 2019 Mac Pro.
Such a product would also not warrant a DTK, but it would be worthy of an early announcement with fall availability.
The reason is Apple could share SDKs for such cards that would work (without acceleration) on existing AS machines but can ultimately target the new hardware in the fall.
As I understand it, even with more cores and memory, the M-series SoC design can not compete with the compute offered by discrete GPUs.
Apple's only foray into discrete GPUs that worked with laptops and the mini has been in their collaboration with Blackmagic for the RX580 and Vega 56 eGPU products. (I have the RX580 BM on my 2018 mini pushing to the XDR Pro and use it for general purposes.)
That said, those discrete, general purpose cards are not only currently incompatible with the M-series Mac mini and laptops, the cards are very weak in the context of today's GPU cards in a very important areas of software development: machine learning and artificial intelligence.
Much focus has been put on NVIDIA and their major strides in delivering powerful high performance computing (HPC). In fact,
NVIDIA announced the ARM-based Grace CPU just a few weeks ago, which is focused on helping "scientists and researchers to train the world’s largest models to solve the most complex problems." That sounds like the kind of space Apple would want to dominate.
NVIDIA's
Grace CPU product page has the quote: "Continuing to scale [AI models] for accuracy and usefulness requires fast access to a large pool of memory and a tight coupling of the CPU and GPU." Doesn't that sound familiar?
I realize this is going out on a limb a bit, but I think Apple could demo a Mac Mini and/or Mac Pro featuring an AI/ML optimized discrete "accelerator" card. Something like
Afterburner ML that lives in an external enclosure or as a Mac Pro Expansion (MPX) module.
Beyond the desire to get developers ready to take advantage of such a card by writing software for it, if the performance of Apple Silicon purpose-built accelerator cards is as mindblowing as the M1 was, the news could immediately freeze hardware spending and force IT managers to re-evaluate plans for capital investment budgets for 2022.
And if in some super rare case they did do this, they’d likely not announce a new product but they’d announce the feature coming to new chips
To some extent what you say here fits with the theory an Afterburner ML card is a component
to a new product (MX). Which is sort of a
feature coming to new chips.