By “have the bandwidth” I meant in terms of speed of the RAM data rate. In terms of fitting the bus width, no lol, we agree.
LPCAMM2 with LPDDR is really about workstation and replacing SODIMMs.
LPDDR as in non-LPCAMM and on-PCB or on-package is here to stay, because LPCAMM is still too big. DIY guys don’t get this. They think LPCAMM is the end of soldiered RAM — hardly. It’s the beginning of more bandwidth and lower power for workstation stuff, but I don’t think we’re going to see like wide memory bespoke stuff switch from LPDDR on-package to LPDDR LPCAMM or even 28W mainstream AMD/Intel CPUs switch from LPDDR (regular and off-package) to LPDDR LPCAMM.
It’s still too big! Which sucks, but for now it’s the truth, and probably will be in the future.
You know I have since softened my stance on this, apparently there is going to be further work on CAMM modules, I don't know if they'll call it CAMM3 or not, to make them tool-less to replace (and maybe smaller?). There was also apparently work in the original CAMM modules to make a quad-data channel version in the same size which obviously didn't survive conversion to JEDEC standard in LP/CAMM2. However, if that were to happen, then depending on the memory configurations offered, right now the only offerings are 32 to 128GB, you could certainly envision a "modular" memory version of the Max. I put it in quotes because truthfully you'd still need to populate all available slots (2 for quad channel memory) and if you wanted aftermarket upgraded you'd have to resell your old ones, not just add new ones. Which of course is not quite the same as buying the minimum number of SODIMMs and then adding your own or getting more later. I don't know that this will happen and as you say it'll still be bigger and somewhat more power-hungry than on-package (or even on-die if Apple goes there) memory (which are naturally the only reasons Apple wouldn't consider it, no other reasons whatsoever, no sir). But it could be practical.
Let's posit that someone made a quad channel 32 or even better 24 LPCAMM3 module, then two of the latter would be 48GB of memory on a 512-bit bus, exactly the same as the current M3 Max. You'd have to be able to fit 4 LPDDR 6GB modules on this hypothetical LPCAMM3 module. Sounds doable in theory? And with only two such modules, the space taken up on the motherboard wouldn't be
that bad especially if LPCAMM3 modules were also smaller.
Tbqh, I would be significantly less interested in this other than for academic and “spread the Apple IP heck yeah, break the X86 wall on computing” element if Apple’s RAM/storage margins and starting prices weren’t so patently insane. I can suck it up etc depending on the case but it really does piss me off as a matter of principle lol.
The whole “but it’s good RAM” etc thing is also B S, it’s always been exorbitant on the margin even pre-on package or big bus! Lol. And even LPDDR5x-7000++ or 8000 is being sold at really reasonable marginal prices in the PC world, or LPDDR5 was even from Dell like a year or two ago. Talking like + $400 to get to 64GB from 8 with an already low entry, even with a storage upgrade (and that’s another thing, lol, god).
In the iPhone and iPad I can tolerate Apple being Apple on modest starting storage/RAM stuff, really. Hasn’t been unacceptably low in a while too. Android offer better sheet hardware values still like with LTPO, 8GB LPDDR5x & 256GB for 900 and all, but it doesn’t feel that crazy tangible like the compute stuff does. It’s more a “ya well Apple is usually conservative and needs its premiums”.
But the computer stuff is just insulting, and I am not someone who whines about removable RAM/SSDs! That stuff is stupid. LPDDR = awesome.
The problem is just that Apple has a monopoly on their market and people kinda put up with it, which I get but still. In the Windows [or with any other hypothetical desktop OS licensed] world you just can’t get away with the same kind of crazy marginal pricing or starting pricing, because competition exists — removable often misses the plot.
I don't think you'll find much disagreement here that Apple's base specs and upgrade pricing can be pretty exorbitant - especially on the base M-series. While storage never really gets good (some models starting at 1TB are fine for their price points, but even for those the upgrade to 2TB is still really bad), I feel like the RAM gets better on the higher model - like the M3 Max, even to a lesser extent the M3 Pro, especially since it doubles as high bandwidth GPU memory which is normally even pricier. If you think of it as CPU memory, then yeah it's pretty pricy, but if you think of it as VRAM it ranges from decent to absolutely amazing (and the higher tiers on the Max are nowhere near as bad - $800 for to add 64GB to 64GB or 60 to 36 is a lot better than $400 to add 18 to 18). So somewhere in between terrible and amazing for the totality?
This is one thing I'm hoping that further competition in the PC space will rectify for everyone (GPU VRAM pricing and Apple's, especially base, RAM - the tea leaves for the M4 are looking half-positive that base storage and maybe RAM bumps are coming to the Mac).
Yep, if I do later this year I’d have an open season of tests if you guys want to know anything haha
Yeah the thing about my tests are that it may not even work as I'm not sure about how the base clocks respond to multiple threads which adds an extra confounding factor that you'd have to keep track of during the runs (i.e. record clock speeds during the tests) and each Cinebench run takes like 10 minutes. So it'd be a pain and if you were into it, I'd more than understand. Still I think it's a neat question though if you were up for it.
