I'm stunned that "mapping out" blocks of bad RAM is a thing you can do. It reminds me of the old days when hard disks had to be "low-level formatted" first (ask your parents, kids) and it would ask you for a list of bad sectors to map out. Sometimes hard disks would even come with a list of known bad ones printed on the top.
"Bad" ram re-mapping is handy for equipment that has special areas of memory shared with hardware like accelerator ASIC, other cpu, or unknown features. Also useful in extreme environment contexts, where the chips may get heavily damaged while in use.
>hard disks would even come with a list of known bad ones printed on the top.
HDD still do, but the flash on the controller quietly handles sector sparing, and remapping. This is why one can't just swap unmodified controller PCB anymore for data-recovery without secret vendor software. =3
The race to make laptops thinner is pointless. I wish I could buy a thicker laptop that is more robust, repairable, expandable, and has better cooling, with the extra space filled with batteries.
Perhaps for you, but I very much care about my laptop's size and especially it's weight, as I carry it around on a daily basis.
Repairing or upgrading it aren't really of interest to me, as I wouldn't be able to aquire the parts where I typically live anyway. And performance and cooling have already been more than good enough for me in most laptops for the last decade.
The post you replied to literally doesn't care about upgrades.
More enforcement also doesn't matter if its what you need to get specifically _what was asked for_, which is a thinner device in this case.
I also wouldn't like an X60 if it meant I have to leave things out of my bag.
What is the profit for a manufacturer if you buy one laptop and use it for 10 years straight instead of buying a new one every 3 years?
Also they not only solder DRAM, now they solder keyboards so that you cannot replace it easily. My garbage keyboard broke after just 2 years of using and I had to solder ~100 plastic pins to replace it (and the replacement keyboard turned out to be not of a great quality).
Until very recently, the low power version of RAM had to be soldered to the motherboard.
If you want the battery life improvements, bear in mind that there is a hard limit on how large a battery you are allowed to fly with, so low power components are important.
I wish that my MBP14 was as thin as my partner’s MBA13.
Not to mention how often I long for my MBA11 with the tapered body which almost made it feel like it was floating in my lap when working from the couch.
My lenovo work laptop is crippled by poor cooling and can only sustain ~1GHz clocks on extended tasks. An old personal laptop 10 years older outperforms it.
Framework does exactly what the GP claims to want.
Unfortunately, not many people making those claims _actually_ want them by revealed preferences. They overwhelmingly want the lower price laptops with similar performance specs.
Or… there’s a level of value that people will assign to the repairability offered by Framework, but the current cost is too high?
I like and support what Framework offers. But when I last bought a laptop (a couple of years ago now) the FW13 was almost double the price of a similarly-specc’d and thinner/lighter Asus. (From €1000 to ~€18-1900).
I’d probably pay a ~40% premium for FW, all else being equal, but 80-90% is just too much: it just doesn’t make sense to whatever value-assessment algorithms I’ve developed over time.
But a 40% premium is not enough to sustain framework imho. The business model of framework is that you don't buy a new laptop, but just upgrade. Therefore, the lifetime value of one customer on framework is less than that of an Asus buyer (who would have to buy a new laptop every new CPU generation or when batteries die, which is probably at most 5 years).
Therefore, framework has to charge a lot more per customer to make the business make sense. 2x premium is not unreasonable under this sort of assumption.
I personally do think they're doing a good job of course, but i do also wonder the long term viability because their customers are so niche.
> But a 40% premium is not enough to sustain framework imho. The business model of framework is that you don't buy a new laptop, but just upgrade. Therefore, the lifetime value of one customer on framework is less than that of an Asus buyer (who would have to buy a new laptop every new CPU generation or when batteries die, which is probably at most 5 years).
> Therefore, framework has to charge a lot more per customer to make the business make sense. 2x premium is not unreasonable under this sort of assumption.
If this is their model, maybe it makes sense from an MBA perspective, but I'm more in the MBAs-are-ruining-the-world camp, and I hate it.
A few points:
- I suspect that private buyers of laptops rarely have strong loyalty to a single brand, so Asus et al can't assume a long chain of purchases over time - they have to continue to compete on price and specs. Framework can probably assume stronger brand loyalty than every other manufacturer out there, based on a) if you've paid their hefty premium, you're really into what they're offering, and b) you've paid the high price of entry for the promise of lower-cost upgrades down the line, which you can only get from Framework. So Framework can probably assume a longer guaranteed income from each customer than any other manufacturer, even if the actual value decreases over time (due to lower cost of upgrades vs. full systems). However, if the upgrades are profitable as individual items, they still win.
- I'd prefer to see Framework sell computers and components at the lowest possible price they can while still being a viable business, with the margin of those items each justifying their existence individually. I appreciate that their economies of scale are different to larger manufacturers and their designs may be more expensive to deliver - this would be covered by a (modest) premium.
- Ultimately, I'd characterise the business model you described as "we'll f*k you on entry, and you might claw back some value over time". I'd rather aim for "we'll sell you a good product for a reasonable price today, and then delight you with the bonus of upgradability later on".
- I wonder whether their issue might actually be scale of manufacturing: that they can't deliver enough systems quickly/reliably enough to compete for a broader market (which would require lower margins to attract more price-sensitive customers) thereby driving them to target a more limited market of nerds who love the repairability concept, and also price higher per system to make the numbers work?
Speaking as someone who did buy a framework as my primary computer (and has been happy with it), my hesitation had much more to do with uncertainty about the execution than with price, which is pretty competitive when you diy. Especially because for the value proposition to pay off, the company/ecosystem has to survive for several years at least and continue making compatible product in the categories I would want. It's a big investment for most people and it's not surprising most go with a safe/known option.
13 seems to have emerged as the generally best compromise for using at your desk and traveling. Even a 15” is still small relative to an external monitor and is getting large to fly with.
"bestest" CPU is not required, but with 13 inch screen you have to either make everything tiny and strain your eyes, or have only 10-15 lines of code in the screen. Also, modern websites and apps use gigantic font size and you will have like 3 paragraphs of text in one screen.
Companies have standardized on, mostly, 13” laptops with whatever accessories. But, honestly, for some people who never leave their desk, a desktop of some sort is probably a better choice— though just connecting to a server works too.
It is fairly simple business to desolder the old BGA chip and replace with a new one.
There are specialized shops willing to do this around the world. This is especially common in third world countries, Eastern Europe, China etc.. There are people who would do this change for around 5-10 USD in India (as long as you can find the replacement yourself).
I am amazed at how many people on this forum think of this as some scary dark magic.
I read that Chinese found a way to upgrade DRAM on GPUs, for example, making RTX 2080 Ti with 22 Gb DRAM, or RTX 4090 with 48 Gb. The modded cards are better (by price and RAM size) than the new ones manufactured by NVIDIA, although some say they are not reliable.
I am so glad that people like Louis Rossmann[1], Electronics Repair School[2], NorthridgeFix[3] and countless others have worked to change that perception.
Chip-level repair SHOULD be the norm.
China has been the master at this stuff for a while with institutes like Wantong Training School, Shenzhen etc.
It is a real shame that much of this knowledge is either on Chinese or Russian Internet forums or buried in obscure and unsearchable YouTube videos.
We used to buy Dells with minimum RAM and add quality third party RAM for way less than Dell wanted for it. They also use ram quantity to leverage other specs we're not bothered about. Of course they are moving to soldered.
the fact no component supplier, not even alibaba sells LPCAMM2 sockets tells you everything you need to know. The only prices I can find are in $30 range per socket
Connectors are expensive, high-speed ones particularly. $30 is fine, though it makes a giant difference at what quantity that is. Did you look at a single one or 1k?
I've been eyeing LPCAMM2 interposers for non-RAM uses - as an off the shelf connector for stacked PCB designs. The conclusion was that it wasn't produced or used at sufficient scale for it to work.
DRAM problems used to be a common thing, at least AFAIK in the 90s up to early 10s. I wonder if that still the case or is it now a non-issue for the first 5 years of usage. Our Watch, iPad, Phones all have more memory than old computers. Even Apple Watch Ultra now has 4GB of memory which I thought was crazy.
If it is not, how would all these appliance with memory doubt with bad memory?
I'm looking to replace my "eats batteries inside 4 months" (set to charge to no more than 90%, health is 60% in 4 months after less than 20 cycles) second-hand Carbon, and urgh, so many soldered RAM or memory on package designs. Some of Lenovo's 14 inch laptops are still SODIMM or LPCAMM2, but most appear to have gone the non-maintainable route.
Tried a Tuxedo laptop, lovely 120Hz display, but keyboard layout and I disagree. Turns out I really like the older ThinkPad inverted T with pgup/pgdown bracketing, and the 4 keys to the right of the F keys. Problem is, that locks me to ThinkPad/Lenovo, as everyone else seems to use the split up/down cursor with Fn to get page keys. Also really prefer physical mouse buttons over the trackpad, and the Tuxedo was haptic only.
The upgrade cost on the LP RAM though.. 940 EUR for 32 -> 64. Sure, 23% is VAT, but damn the hardware costs hurt right now. At least SODIMM is still cheaper.
The RAM situation is beyond helpless. I was meaning to buy another 32GB stick for my Framework, but they're out of stock and so are any validated alternatives.
I've found just one store that carries them, but seeing as they reported "50+ units in stock" I called the store and learned that "the item is available through a third party vendor within four business days", but looking at that the ratings it would be more like three weeks.
All manufacturers have just decided not to produce any this year. Think about how absolutely crazy that is. 100% of the year's production is for AI. The good news is in 3 years when dollars are worthless, there will be a glut of also worthless server RAM.
Gone back to my old T480 here for that. I don’t need a lot of CPU but I do need a lot of RAM. Managed to get 2x16Gb sticks for not much money. And a better screen than the newer T14 I have.
T480 also has dual batteries with one being removable. I upgraded the top cover to magnesium, added WiFi 7, upgraded to 32 GiB, swapped from a HDD to a SSD, and changed the motherboard from i3-7300U to the i7-8650U. The only repair I ever did was replace the screen because I messed up once and dropped it. Oddly enough, it supports an eGPU over Thunderbolt 3 and 25W TDP unlocking so it was once used as my compact gaming machine. Mine also has 3 spare extended batteries and a custom external charger for them.
T480s and T490 are inferior compared to the T480's usability of ports and batteries.
Bash me all you want, but after CPU, GPU, memory controller & a host of other things integrated in modern System-on-Chips (SoC), I've long held the belief that next logical step would be to integrate RAM.
Optimal: piecemeal. As in: X sockets on a board, each socket takes a chip that adds CPU core(s), GPU cores, and some amount of main RAM. More chips, more powerful computer.
Keyword: socket. Read: replaceable. Bad chip out, good chip in. Slow chip out, faster chip in. Soldered onto board = reliable, but a defect could junk the whole machine.
Considering the current trend is for SoCs to deintegrate: don't count on it.
Putting RAM inside the logic die is crazy expensive - just look at how much footprint L3 cache takes up on a modern chip. It'll always make more sense to put it in a separate die, made using RAM-specific processes. With Zen the memory controller and GPU has (once again) moved to a separate die, and with X3D we're now even seeing separate cache dies.
There's of course HBM if you want to consider that "integrated", but the large cost bump, additional manufacturing risks, and loss of flexibility don't make that an attractive proposition unless you really need the extra speed. Regular DDR dies? You gain basically nothing by co-packaging those.
Manufacturing processes for RAM and processor cores are not one and the same and use different tradeoffs. It's the main problem in-memory processing is running into, which is akin to what you're describing here. The die space required for the same level of performance makes the idea of on-die RAM intractible. The best they're able to do with RAM optimised manufacturing processes are tiny RISC-V cores that lack most of the feature sets. What apple does with moving the RAM chips on to the package really is the only viable option.
> Bash me all you want, but after CPU, GPU, memory controller & a host of other things integrated in modern System-on-Chips (SoC), I've long held the belief that next logical step would be to integrate RAM
Haven't Apple (A-series, M-series) and others been copackaging memory for a while? (SiP or MCM?)
It is - anything short of HBM can be replaced that way.
The issue, of course, is just how much time, effort and skill is required to do it.
It might become a viable commercial process outside third world countries if RAM prices spike even more - or if someone finds a way to teach an off the shelf $1000 robot arm advanced autonomous microsoldering.
Then there's the problem of forcing the device itself to recognize the new RAM. It's a device-specific trick. RAM sticks have a little tag chip that tells a PC "how much RAM this stick has", but RAM chips themselves don't - so the device has to infer its RAM capacity from elsewhere. Like strap resistors on the board. Which you have to find and adjust for the new RAM capacity. That, or fix the RAM values in firmware, which is also device-specific.
> RAM sticks have a little tag chip that tells a PC "how much RAM this stick has", but RAM chips themselves don't - so the device has to infer its RAM capacity from elsewhere
I'm surprised that in 2026 this is still true.
You could easily fit a little microcontroller on the die of each RAM chip which reports its make/model/manufacturing date, capacity and capabilities, can be triggered to run self-tests, etc. If they aren't doing that yet, I'm surprised that they aren't.
Well, it turns out they actually are bothering – while in DDR thru DDR4, the mode registers are write-only, LPDDR2 onwards have an MRR (Mode Register Read) command, which the CPU/chipset can use to inspect the manufacturer and capacity; and DDR5 has copied the same facility from LPDDR.
If you live in Shenzhen, it's a subway trip away. Good repair services elsewhere can do it too. Either way any warranty you have is probably getting called into question after that, even if that's supposed to be illegal in some places (e.g. the US - they're supposed to prove that your modification/repair is what caused the warranty fault, but in practice…).
Yes, of course. It says - ball grid (array) in the article. Doing it by hand, e.g. with just hot air is proper hard. It's not hard with specialized equipment (that cost way more than the laptop, itself). Sourcing the chips is not easy either.
As for the videos and the Macs, you can see the same people (likely) soldering RAM as well.
Since the OP mentions Lenovo, look at their other product lines for ones that still have socketed RAM. I'm currently running an E16 with 40GB RAM, 8GB soldered, 32GB socketed. And aside from that it leaves my previous X1 for dead in terms of performance. Lenovo's replacing the carbon fiber case, the whole reason the X1 line is called "Carbon", with crappy black plastic was the last straw.
To counterweight all the cynical takes about soldering RAM being a conspiracy to force device obsolescence, I can think of at least one other dynamic that could have resulted in today's scenario: smartphones.
Smartphones also need RAM and are especially subject to requirements of always being thinner, lighter, and have a durable battery, yet have spacious amounts of RAM.
Growing demand for smartphone RAM pushed for its optimization and production scaling and since all its requirements also match laptops, this convergence was inevitable.
Hopefully the growing demand for datacenter RAM density will make modular interface standards like LPCAMM2 more available (soldered has no place in datacenter).
Am I the only one who wonders if there is a way to get the kind of performance we want out of non-soldered components/SOC designs? Like from what I've heard, it's a "physics says no" problem, but I'm not sure if that's actually true (not being an expert in this technology) or if it's just the excuse many companies use to justify soldering everything to make serviceability impossible. And it would be really neat if this is actually being looked into by labs right now (possibly with AI's help), since areas like that are where I think (non) LLM AIs might be very useful. Same thing for medicine, materials, etc.
The boards using LPDDR don't have socketed CPUs. Also, every connector is a huge hit to your signal integrity, so one versus two might indeed be a big deal.
And yeah, it can be, but that's how you end up with the far slower regular DDR. Fast, repairable, cheap: pick (at most) two.
Effectively there are no socketed CPUs for laptops, though, for way more than a decade already. Saying that as I'd fully support modular components, as in desktops.
While it's true that you can get better performance out of soldered memory than a socket, that's also not why vendors use it. Desktop computers, with their very much not soldered memory, get great performance after all. The laptop vendors solder the memory because it lets them make more money off selling units with more memory soldered in. That's all.
> Besides, since RAM isn't getting cheaper for a while, this isn't a big issue now.
I would argue the exact opposite, actually! My home machine has 32GB of RAM and is already over 7 years old. If it were to die right now I'd ideally go for an upgrade with 64GB so it would be good for another 7 or so. But that's completely unaffordable, and I could survive with 16GB for a year or two, so going for the bare minimum now and upgrading it later is looking very attractive...
I don't know if they're asking for them, but they sure seem to appreciate them and see them as "better".
At work, the standard issue laptop is some form of HP Elitebook. These have been getting thinner, and people seem to like that. We're not a tech shop, and even the IT dept. is mainly barely-technical project-managers. An 8th gen i5 would be plenty powerful for what they do, but fortunately for them we have to replace the laptops regularly because for some reason we buy them with the least amount of RAM possible (even before the prices exploded).
Anyway, at the start of the year I was fed up with the ridiculous screen quality of the HPs and other annoyances, so I got myself an Lenvo p14 something-or-other. It's almost twice as thick and much heavier than the HPs, and everyone who sees it makes fun of it.
"Smartphone makers glue them because it's cheaper and they can make more money ..." No. It's what the consumers want. Thinner phones, water resistant phones, and thinner laptops.
I can't recall the last time someone mentioned a phone or laptop being too thick as a downside in a review. Thin and light devices have been widely available for well over a decade, even to the point of caricature like the iPhone Air. It just isn't a feature worth caring about anymore.
Water resistant? Those ratings are lies anyways, and nobody is going to bring their expensive smartphone on a diving trip to take pictures - or even take it surfing without a waterproof case. In practice "water resistant" means "won't die from an occasional splash", and that can be achieved perfectly fine by repairable phones.
People seem to care about "thin and light" laptops, at least my colleagues do. But since they don't pay for these out of pocket, I can't tell how important it is to them, measured in currency.
Now, I don't know about surfing, but I regularly strap my iphone to my motorbike's handlebars to use as a GPS. No case. I don't particularly enjoy doing this under pouring rain, but it does happen, and the phone gets drenched. However, to the point about people wanting this, yes, I do. I would also love a replaceable battery. My old Galaxy S5 had both features, so saying that waterproofing means non-user-replaceable battery is a lie.
I'm washing my expensive flagship phone with water and soap every week, for almost a decade now. Never had a problem because of this (except the USB port getting disabled for a few hours until it dries).
>hard disks would even come with a list of known bad ones printed on the top.
HDD still do, but the flash on the controller quietly handles sector sparing, and remapping. This is why one can't just swap unmodified controller PCB anymore for data-recovery without secret vendor software. =3
Perhaps for you, but I very much care about my laptop's size and especially it's weight, as I carry it around on a daily basis.
Repairing or upgrading it aren't really of interest to me, as I wouldn't be able to aquire the parts where I typically live anyway. And performance and cooling have already been more than good enough for me in most laptops for the last decade.
Thinner devices need more solid enforcement to stay stiff. My Thinkpad X60s weighs the same as a Macbook Air and it's infinitely more upgradable.
Also they not only solder DRAM, now they solder keyboards so that you cannot replace it easily. My garbage keyboard broke after just 2 years of using and I had to solder ~100 plastic pins to replace it (and the replacement keyboard turned out to be not of a great quality).
If you want the battery life improvements, bear in mind that there is a hard limit on how large a battery you are allowed to fly with, so low power components are important.
Could it be that thin laptops are largely an excuse to provide a need to solder RAM; which in turn encourages consumption through obsolescence?
Not to mention how often I long for my MBA11 with the tapered body which almost made it feel like it was floating in my lap when working from the couch.
Unfortunately, not many people making those claims _actually_ want them by revealed preferences. They overwhelmingly want the lower price laptops with similar performance specs.
I like and support what Framework offers. But when I last bought a laptop (a couple of years ago now) the FW13 was almost double the price of a similarly-specc’d and thinner/lighter Asus. (From €1000 to ~€18-1900).
I’d probably pay a ~40% premium for FW, all else being equal, but 80-90% is just too much: it just doesn’t make sense to whatever value-assessment algorithms I’ve developed over time.
and you'd be one of the higher ones.
But a 40% premium is not enough to sustain framework imho. The business model of framework is that you don't buy a new laptop, but just upgrade. Therefore, the lifetime value of one customer on framework is less than that of an Asus buyer (who would have to buy a new laptop every new CPU generation or when batteries die, which is probably at most 5 years).
Therefore, framework has to charge a lot more per customer to make the business make sense. 2x premium is not unreasonable under this sort of assumption.
I personally do think they're doing a good job of course, but i do also wonder the long term viability because their customers are so niche.
> Therefore, framework has to charge a lot more per customer to make the business make sense. 2x premium is not unreasonable under this sort of assumption.
If this is their model, maybe it makes sense from an MBA perspective, but I'm more in the MBAs-are-ruining-the-world camp, and I hate it.
A few points:
- I suspect that private buyers of laptops rarely have strong loyalty to a single brand, so Asus et al can't assume a long chain of purchases over time - they have to continue to compete on price and specs. Framework can probably assume stronger brand loyalty than every other manufacturer out there, based on a) if you've paid their hefty premium, you're really into what they're offering, and b) you've paid the high price of entry for the promise of lower-cost upgrades down the line, which you can only get from Framework. So Framework can probably assume a longer guaranteed income from each customer than any other manufacturer, even if the actual value decreases over time (due to lower cost of upgrades vs. full systems). However, if the upgrades are profitable as individual items, they still win.
- I'd prefer to see Framework sell computers and components at the lowest possible price they can while still being a viable business, with the margin of those items each justifying their existence individually. I appreciate that their economies of scale are different to larger manufacturers and their designs may be more expensive to deliver - this would be covered by a (modest) premium.
- Ultimately, I'd characterise the business model you described as "we'll f*k you on entry, and you might claw back some value over time". I'd rather aim for "we'll sell you a good product for a reasonable price today, and then delight you with the bonus of upgradability later on".
- I wonder whether their issue might actually be scale of manufacturing: that they can't deliver enough systems quickly/reliably enough to compete for a broader market (which would require lower margins to attract more price-sensitive customers) thereby driving them to target a more limited market of nerds who love the repairability concept, and also price higher per system to make the numbers work?
I'm pretty sure my memory is correct, and for a DIY Ryzen 8840/16GB/1TB it was roughly an 80-90% premium, which I'd not describe as competitive.
I used to use an 11" as my daily driver (and that was before retina-resolutions were a thing). It wasn't ideal, but it was fine for development work.
There are specialized shops willing to do this around the world. This is especially common in third world countries, Eastern Europe, China etc.. There are people who would do this change for around 5-10 USD in India (as long as you can find the replacement yourself).
I am amazed at how many people on this forum think of this as some scary dark magic.
Chip-level repair SHOULD be the norm.
China has been the master at this stuff for a while with institutes like Wantong Training School, Shenzhen etc.
It is a real shame that much of this knowledge is either on Chinese or Russian Internet forums or buried in obscure and unsearchable YouTube videos.
[1]: https://www.youtube.com/channel/UCl2mFZoRqjw_ELax4Yisf6w
[2]: https://www.youtube.com/@electronicsrepairschool
[3]: https://www.youtube.com/@NorthridgeFix
One can also DIY such mods/repairs but it requires some extra equipment and supplies.
Searching for "bga soldered ram chip change laptop" on YouTube returns like thousands of results of people doing this.
Although I now realize how skewed my perception of microsoldering and other such skills because such skills seem to be extremely niche in the west.
https://frame.work/blog/framework-laptop-13-pro-deep-dive-lp...
I've been eyeing LPCAMM2 interposers for non-RAM uses - as an off the shelf connector for stacked PCB designs. The conclusion was that it wasn't produced or used at sufficient scale for it to work.
https://www.koreatimes.co.kr/business/tech-science/20260420/...
If it is not, how would all these appliance with memory doubt with bad memory?
Tried a Tuxedo laptop, lovely 120Hz display, but keyboard layout and I disagree. Turns out I really like the older ThinkPad inverted T with pgup/pgdown bracketing, and the 4 keys to the right of the F keys. Problem is, that locks me to ThinkPad/Lenovo, as everyone else seems to use the split up/down cursor with Fn to get page keys. Also really prefer physical mouse buttons over the trackpad, and the Tuxedo was haptic only.
The upgrade cost on the LP RAM though.. 940 EUR for 32 -> 64. Sure, 23% is VAT, but damn the hardware costs hurt right now. At least SODIMM is still cheaper.
The LP in LPDDRx isn't about creating thinner devices.
The LP stands for Low Power and is aimed at making devices that draw less power resulting in a longer battery life.
I've found just one store that carries them, but seeing as they reported "50+ units in stock" I called the store and learned that "the item is available through a third party vendor within four business days", but looking at that the ratings it would be more like three weeks.
T480s and T490 are inferior compared to the T480's usability of ports and batteries.
Agree non user service parts is not good for consumers.
Use tools as d search Youtube videos memory upgrade device name before you buy a laptop to check if the memory is user upgradeable without soldering.
Optimal: piecemeal. As in: X sockets on a board, each socket takes a chip that adds CPU core(s), GPU cores, and some amount of main RAM. More chips, more powerful computer.
Keyword: socket. Read: replaceable. Bad chip out, good chip in. Slow chip out, faster chip in. Soldered onto board = reliable, but a defect could junk the whole machine.
Putting RAM inside the logic die is crazy expensive - just look at how much footprint L3 cache takes up on a modern chip. It'll always make more sense to put it in a separate die, made using RAM-specific processes. With Zen the memory controller and GPU has (once again) moved to a separate die, and with X3D we're now even seeing separate cache dies.
There's of course HBM if you want to consider that "integrated", but the large cost bump, additional manufacturing risks, and loss of flexibility don't make that an attractive proposition unless you really need the extra speed. Regular DDR dies? You gain basically nothing by co-packaging those.
Haven't Apple (A-series, M-series) and others been copackaging memory for a while? (SiP or MCM?)
Then there is the HBM which is an expensive process to mount on the chip.
Can't you select and copy the text in the photo (on most smartphones and PCs at least)?
The issue, of course, is just how much time, effort and skill is required to do it.
It might become a viable commercial process outside third world countries if RAM prices spike even more - or if someone finds a way to teach an off the shelf $1000 robot arm advanced autonomous microsoldering.
Then there's the problem of forcing the device itself to recognize the new RAM. It's a device-specific trick. RAM sticks have a little tag chip that tells a PC "how much RAM this stick has", but RAM chips themselves don't - so the device has to infer its RAM capacity from elsewhere. Like strap resistors on the board. Which you have to find and adjust for the new RAM capacity. That, or fix the RAM values in firmware, which is also device-specific.
I'm surprised that in 2026 this is still true.
You could easily fit a little microcontroller on the die of each RAM chip which reports its make/model/manufacturing date, capacity and capabilities, can be triggered to run self-tests, etc. If they aren't doing that yet, I'm surprised that they aren't.
I'm assuming if it were still under warranty, dead RAM chip would be enough to get a free repair/replacement from the manufacturer
As for the videos and the Macs, you can see the same people (likely) soldering RAM as well.
Good BGA experience and a steady hand seems required
Smartphones also need RAM and are especially subject to requirements of always being thinner, lighter, and have a durable battery, yet have spacious amounts of RAM.
Growing demand for smartphone RAM pushed for its optimization and production scaling and since all its requirements also match laptops, this convergence was inevitable.
Hopefully the growing demand for datacenter RAM density will make modular interface standards like LPCAMM2 more available (soldered has no place in datacenter).
And yeah, it can be, but that's how you end up with the far slower regular DDR. Fast, repairable, cheap: pick (at most) two.
I agree there is draw back when using socket(mostly size), but performance is not the big issue.
MRDIMM RAM can reach even higher Transfer rate than LPDDR5, yet still socktable.
CPU and RAM are socketed. Clevo had socketed graphics cards too
Besides, since RAM isn't getting cheaper for a while, this isn't a big issue now.
I would argue the exact opposite, actually! My home machine has 32GB of RAM and is already over 7 years old. If it were to die right now I'd ideally go for an upgrade with 64GB so it would be good for another 7 or so. But that's completely unaffordable, and I could survive with 16GB for a year or two, so going for the bare minimum now and upgrading it later is looking very attractive...
At work, the standard issue laptop is some form of HP Elitebook. These have been getting thinner, and people seem to like that. We're not a tech shop, and even the IT dept. is mainly barely-technical project-managers. An 8th gen i5 would be plenty powerful for what they do, but fortunately for them we have to replace the laptops regularly because for some reason we buy them with the least amount of RAM possible (even before the prices exploded).
Anyway, at the start of the year I was fed up with the ridiculous screen quality of the HPs and other annoyances, so I got myself an Lenvo p14 something-or-other. It's almost twice as thick and much heavier than the HPs, and everyone who sees it makes fun of it.
"Smartphone makers glue them because it's cheaper and they can make more money ..." No. It's what the consumers want. Thinner phones, water resistant phones, and thinner laptops.
Water resistant? Those ratings are lies anyways, and nobody is going to bring their expensive smartphone on a diving trip to take pictures - or even take it surfing without a waterproof case. In practice "water resistant" means "won't die from an occasional splash", and that can be achieved perfectly fine by repairable phones.
Now, I don't know about surfing, but I regularly strap my iphone to my motorbike's handlebars to use as a GPS. No case. I don't particularly enjoy doing this under pouring rain, but it does happen, and the phone gets drenched. However, to the point about people wanting this, yes, I do. I would also love a replaceable battery. My old Galaxy S5 had both features, so saying that waterproofing means non-user-replaceable battery is a lie.