Does the writer really think laptops are not upgradeable because of technical issues? Laptop makers convice people to buy more computer than they need as they can't be upgraded, making more money and waste in the process.
The first point is that LPDDR5 chip's operating frequency today has reached an all-time high of 9600 MT/s, and the latest off-the-shelf laptops (Intel Meteor Lake) are running at 7467 MT/s. This kind of high bandwidth is especially preferred in laptops for improved graphics and gaming performance. The iGPU has to compete with CPU for system DRAM bandwidth (which is already only a fraction of what a dedicated GPU has), every GB/s helps. Meanwhile, SO-DIMM is hitting its physical speed limit due to signal integrity issues (I heard that the impedance mismatch at the connector is too great to go faster). So yes, if you want high operating frequency, soldered RAM is the only option.
The second point is that, it's not always needed for every laptop, of course it's just a matter of tradeoff, many users would prefer lower DRAM bandwidth in exchange for better upgradability. On the other hand, hardware vendors can use this technical limitation as a justification to design machines with non-upgradable memory. You hit two birds with one stone: you can achieve better performance and force customers to pay extra for RAM at the same time.
If CAMM sees mass adoption, it's still up to the vendor to decide if they're willing to actually provide upgradability - but at least it'll be made possible at the technical level.
I hear what you’re saying, but AI makes memory bandwidth much more important in computers. As I understand it, it’s much easier to increase memory bandwidth by soldering ram adjacent to the CPU - like GPUs have been doing for decades.
Of course it’s suspicious that computer manufacturers can make more money in the process. But that doesn’t invalidate the technical argument.
If this standard solves the technical problem, I’m all for it. We’ll have to wait and see which manufacturers adopt it.
AI has been around for how many years, and it's still as niche and gimmicky as it's ever been. Especially the heavyweight kind that does need that bandwidth.
The only "AI" feature I use on macOS with any kind of regularity is the built-in OCR to copy text from images. But then my weak-ass Pixel 4a can do the same thing in the app switcher, so I assume it needs neither a fast CPU nor high-bandwidth memory.
The one we're talking about these days which requires high memory bandwidth only existed since 2023. 2022 was GPT 3.5, but locally Llama was the threshold when local llm really took off and that's early 2023. That's in the ballpark of time you normally need to design a new laptop and get it into production.
> Laptop makers convice people to buy more computer than they need as they can't be upgraded, making more money and waste in the process.
Honestly I don't think that really factors into it, simply because the vast majority of buyers have no idea what they need to begin with and really don't care about upgradability as a consequence. Why are they supposed to care about whether the RAM can be upgraded when they don't know what the RAM does or how much is a reasonable amount? (and let's not even mention different speeds...)
Frankly, the average customer does not need to be concerned with RAM speed. If he doesn't know, he doesn't need fast RAM.
RAM capacity on the other hand is something everyone needs to care about, because having insufficient RAM brings the entire computer to a screeching halt. Being blunt, I consider it a travesty that computer vendors are selling machines with 8GB of RAM today. 16GB is the new minimum for practical use, with how bloated Windows, Chrome, and honestly everything is getting.
Don't get me started any further on why we even need at least 16 gigabytes of RAM to do the same things we were doing with 16 megabytes back in the day. Bloody insane, this timeline is.
Meanwhile my 8GB Chromebook runs 3 VMs (Android VM, Linux dev VM, SteamOS VM) and Chrome without breaking a sweat. There is a reason why Google developed MGLRU.
Of course it's both but if you look at Apple's M1-M3, which are a SOC combining CPU, GPU and RAM, you can see it has technological merit. The things are known to be real fast.
There is an issue if you want to create a small laptop with a long battery life as SODIMM will use more space and power as LPDDR. (The SODIMM module and connector are very bulky and have a bad electical layout. CRAMM will be somewhere between the two)
Technical reasons certainly play a part in it. Thin and light and high performance is objectively harder with the currend SODIMM arrangement for replacable RAM. And it's not a slam-dunk win for the laptop manufacturers financially either, because it makes their manufacturing output less fungible for them as well. If they misestimate the demand for one particular configuration that loses them money.
This model forces users to accept the overpriced RAM and SSD options the manufacturer offers as well, previously you could buy the base model and swap in cheap parts but not anymore. The mark-ups are obscene, for example Apple charges $800 for a 2TB SSD when a top-of-the-line 2TB PCIe4 NVMe drive can be had for $140 at retail, or even cheaper if you settle for something just "very fast" instead of "extremely fast".
I also made the observation that more expensive laptops tend to have fewer replaceable parts vs the cheaper models. There is clearly also a manufacturer difference. Out of all the OEMs it's HP that often offers laptop with replaceable memory (especially on the low end).
Playing devil's advocate it might be that the more compact you try to make a laptop the less servicable it gets. I don't buy it tho.
