[url=https://meincmagazine.com/civis/viewtopic.php?p=40765969#p40765969 said:
Desktop and M1 GPU is not even close. M1 and discrete laptop GPU is not even close (although it's damned good compared to both desktop and laptop iGPU).
I guess I should phrase it this way: the size of the market that can be addressed by desktop GPUs but for which mobile GPUs are insufficient keeps on shrinking.
This is true of laptops too. There are plenty of people who only own a smartphone, and that percentage of the population is growing. Soooo.... let's not talk about desktop performance in a thread about desktops? I don't really get your point. Should we not talk about Photoshop as relevant in a thread about Photoshop, because the number of use cases where Affinity Photo is enough keeps increasing?
We're to the point now where the only desktop parts that can't be replicated in a laptop are barely obtainable at all unless you're willing to pay literally four figures for a GPU. That same shift happened for storage (in terms of speed, not raw capacity) years ago.
GPUs are expensive if you want to buy one as an aftermarket part. As part of a prebuilt OEM system, it's quite easy and affordable to buy a desktop PC that absolutely demolishes anything that Apple has, including the graphics card in the Mac Pro. An MXM discrete laptop GPU would similarly be as expensive as a retail desktop PCIe GPU is currently, if such a thing were available at retail.
The high price of loose GPUs is not because there is no use case for these GPUs, it's because of extraordinary once-in-a-century supply chain issues. Retail GPUs are expensive because OEMs negotiated prices for GPUs that go into prebuilts long ago, so they are getting all those GPUs first, at the pre-negotiated affordable rate. What's leftover goes to retail, and the supply is almost nil. That's why they are expensive. Not because there is no demand for them.
I don't know what shift you are talking about WRT to storage, because the exact same storage available to desktops are available to laptops (with the exception of exotic server grade stuff like U.2), so there is no market segmentation there. They share a common market for parts.
Speed between laptop and desktop storage has been the same for at least 20 years, only capacity has differed, and it still differs today. So I don't really know what lessons are to be learned today from something that happened 20 years ago for entirely different reasons: the cheap availability of 7200 RPM in 2.5" drives.
The question is how much of the addressable market wants components that can't be cooled in the laptop form factor. As the components can do more within the laptop power budget, that portion will continue getting smaller. I brought up my own example as a case in point—I'd be happy to replace both a desktop and a laptop with just a laptop if the laptop can cool what I need adequately, which many reports on the M1 Max MBP say it can.
How much of the addressable market even wants a Mac? Why bother discuss anything unless it has absolute market dominance, and unless it appeals to 99% of the population? The same thing is already happening between laptops and smartphones, so I guess we can have this same discussion about those two form factors in 3-7 years, as components can do more and more within a smartphone's power budget. But I don't think that's very helpful to people who need a laptop. Macs and iPhones already use pretty much the exact same SoC, the only real difference is power budget and number of cores.
Maybe the lineup could look like this (all w/512 GB storage and 10G ethernet):
