Dell Will Have Custom DDR5 Memory Module for its Upcoming Laptops

[Valantar]

I never saw a set of DDR4 for $600. The price of 32GB of DDR4 today would be the same as 16GB of DDR4 in 2017

Then you haven't looked very hard - Newegg has tons of kits above $1000 right now (those are 8-DIMM kits for HEDT, mostly). Your DDR4 pricing is also way off - you can get 32GB of DDR4 for about $100 now, which is significantly less than what 16GB cost at the peak a couple of years ago. Btw, what capacity is that $600 DDR5 kit? Also, using the top range of prices as a point of comparison is a very dubious approach, as there is essentially no price ceiling for premium high end products, so where the most expensive ones land is effectively random. You need to look at either entry prices or prices for a representative selection of products.

 

[Mussels]

Hm, that's interesting. Guess it was more of a gradual change than a hard cut-off then, as I don't recall seeing any socketed Ivy Bridge or Haswell laptops. Thanks for clearing that up

I have both (and thats also when i last bought a laptop, as upgrading CPU SSD and RAM in those laptops has got them still working well to this day)

 

[trsttte]

So rumor-ish info, it's possible this will become a new standard. On the WAN show from LTT they mentioned Dell is in talks with JEDEC and Intel to make this a standard (they also mention some of the points that other people mentioned here like z-height and some possible hard capacity limits because of interference and such with sodimms at current speeds)

The timestamp is at 4536s (&t=4536s)

 

[goodeedidid]

Yeah, so this way instead of changing just a bad motherboard, CPU, or RAM sticks, we get to throw out the whole system and buy a new one. Like getting a flat tire and scrapping the entire vehicle, then buying a new one. Good business, but only in a fantasy world where resources are infinite and shortages don't exist.

Weak argument. When does ever memory fail?!? Probably in the sub one percentile if ever. This isn't the nineties you know? You're in 2022 fam

 

[Valantar]

Weak argument. When does ever memory fail?!? Probably in the sub one percentile if ever. This isn't the nineties you know? You're in 2022 fam

Repair is still important, even if component failures like this are rare. Upgrades are a better argument, but it really depends on the component too. I'm fine with soldered CPUs for mobile, simply because socketed CPUs aren't doable in thin-and-light form factors without significant compromises (especially thickness). But there's another angle to this: we need more well-trained microsoldering/board level repair shops with good access to spare parts. Even with easily replaceable parts things on the motherboard can go wrong that are "easily" (with the correct skills, tools and knowledge) fixable, but that can't simply because the infrastructure doesn't exist. Replacing a CPU is still beyond nearly all people capable of doing this, but even soldered RAM is relatively easily replaced if you know what you're doing and can get a replacement part.

 

[goodeedidid]

Repair is still important, even if component failures like this are rare. Upgrades are a better argument, but it really depends on the component too. I'm fine with soldered CPUs for mobile, simply because socketed CPUs aren't doable in thin-and-light form factors without significant compromises (especially thickness). But there's another angle to this: we need more well-trained microsoldering/board level repair shops with good access to spare parts. Even with easily replaceable parts things on the motherboard can go wrong that are "easily" (with the correct skills, tools and knowledge) fixable, but that can't simply because the infrastructure doesn't exist. Replacing a CPU is still beyond nearly all people capable of doing this, but even soldered RAM is relatively easily replaced if you know what you're doing and can get a replacement part.

The way I see it, MOBO, memory, SSD, and, of course, the CPU are all destined to become one whole/complete package. Everything will be unified as this is going to become the most mobile and efficient solution (Aren't components the most efficient when as close as possible physically to each other?). For example like Apple silicon with unified memory. Manufacturing will become so complicated that nothing will ever be user repairable. It's a matter of when IMO

 

[chrcoluk]

Yes: size constraints, design goals and customer wants. If not for that, every laptop could be 10cm thick, use only desktop parts, and have 10m battery life, but... well, that's not what people want. And there are plenty of popular product categories now where two SO-DIMMs just isn't a sensible use of space, impinging on battery, keyboard, cooling, etc.

Never known a single person that wants a thin laptop, so I guess you mean marketing needs, kind of like how the manufacturers think we want slim phones but then most people thicken them with covers.

 

[Valantar]

Never known a single person that wants a thin laptop, so I guess you mean marketing needs, kind of like how the manufacturers think we want slim phones but then most people thicken them with covers.

*raises hand* I'm definitely one. And judging by sales figures, I'm definitely not alone. 12-14" thin-and-lights dominate the market these days (even if thicker and heavier cheapo education and enterprise laptops do bring the 15"-class into a majority of products offered). I did love my old, thick Thinkpad X201 (35.3mm thick at the back), but moving to something thinner was a godsend for portability. My current work-issued Latitude 7390 2-in-1 is much thinner, and that is a major benefit to me. The only people I know who don't care how thick their laptop is are people who don't care about portability at all, and just want something they can stow away more easily than a desktop. The benefits of thin laptops are very real for a huge number of people. That obviously doesn't mean "thinner is always better", and I've seen quite a few laptops too thin for my tastes (a couple of Samsungs, among others), as it has either made them flimsy or just feel bad in use. But a well built ~15mm thick laptop is vastly superior for my need than anything even 5mm thicker.

The way I see it, MOBO, memory, SSD, and, of course, the CPU are all destined to become one whole/complete package. Everything will be unified as this is going to become the most mobile and efficient solution (Aren't components the most efficient when as close as possible physically to each other?). For example like Apple silicon with unified memory. Manufacturing will become so complicated that nothing will ever be user repairable. It's a matter of when IMO

For mobile, that's likely, though not the SSD. The potential benefits from integrating storage on-package are tiny - even Apple isn't doing that. NAND and a controller need a ton of space that would just balloon pacakaging costs for no real benefits (the controller would still consume the same amount of power, after all). But I'm definitely thinking we'll see on-package memory proliferate in the coming years. Still, I think we'll likely see that paired with some kind of multi-tier memory setup, as a system like that will be both expensive and inflexible, and unlike Apple, neither Intel nor AMD have the integration needed to make just a few SKUs of their chips. Just imagine the inventory management nightmare if every mobile CPU SKU now instead came in 8GB, 16GB, and 32GB RAM SKUs - just keeping track of all of that would be a significant cost increase.

But that just increases the need for a proliferation of board-level repair shops and availability of parts for these. Heck, ideally every single PC that is scrapped should have every onboard component removed and tested for re-sale as spare parts. I hope someone with some clout, like the EU, gets around to setting up a system like this. It would be a massive benefit to both users and the environment.

 

[TheoneandonlyMrK]

I don't like this as others have said it's a ewaste generating idea, even when a laptop has died beyond repair I have often taken it's memory to upgrade others.
I won't be buying a Dell laptop, not that I did anyway though.
I do have a work dell i5, but wouldn't touch it's innards too so I'm not too bothered really I suppose.

 

[Valantar]

I don't like this as others have said it's a ewaste generating idea, even when a laptop has died beyond repair I have often taken it's memory to upgrade others.
I won't be buying a Dell laptop, not that I did anyway though.
I do have a work dell i5, but wouldn't touch it's innards too so I'm not too bothered really I suppose.

Wait, so making a new, better form of removable memory is worse to you than having it soldered to the PCB? Because that's the alternative here, not SO-DIMMs. You're starting from a presumption that is the opposite of the likely reality. After all, this is removable, and can thus be sold on or re-used in any compatible laptop, unlike soldered memory. According to the Dell rep LTT mentions speaking to in the video linked above, they're even working on making this into an official standard, in part because SO-DIMMs are reaching the end of their useful life due to signalling issues with higher speed memory. If that pans out, or even if it just leads to a new standard that cuts down on Z-height and keeps other benefits from this, that's still necessary progress - the number of laptop designs skipping SO-DIMMs because they don't fit their design requirements is staggeringly high, and this can likely fit in a lot of those.

 

[TheoneandonlyMrK]

Wait, so making a new, better form of removable memory is worse to you than having it soldered to the PCB? Because that's the alternative here, not SO-DIMMs. You're starting from a presumption that is the opposite of the likely reality. After all, this is removable, and can thus be sold on or re-used in any compatible laptop, unlike soldered memory. According to the Dell rep LTT mentions speaking to in the video linked above, they're even working on making this into an official standard, in part because SO-DIMMs are reaching the end of their useful life due to signalling issues with higher speed memory. If that pans out, or even if it just leads to a new standard that cuts down on Z-height and keeps other benefits from this, that's still necessary progress - the number of laptop designs skipping SO-DIMMs because they don't fit their design requirements is staggeringly high, and this can likely fit in a lot of those.

I'm ok with standardised multi OEM ,easy to buy upgrades is all, better, yes please too.

 

[goodeedidid]

Heck, ideally every single PC that is scrapped should have every onboard component removed and tested for re-sale as spare parts. I hope someone with some clout, like the EU, gets around to setting up a system like this. It would be a massive benefit to both users and the environment.

I doubt this will ever be a thing. Old parts aren't going to be compatible with newer tech and also I can imagine that the labor for something like this is going to be colossal and not profitable. The better solution would be just to destroy old discarded parts in a way that has a low carbon footprint.

Just imagine the inventory management nightmare if every mobile CPU SKU now instead came in 8GB, 16GB, and 32GB RAM SKUs

Isn't Apple doing that already?

 

[Valantar]

I doubt this will ever be a thing. Old parts aren't going to be compatible with newer tech and also I can imagine that the labor for something like this is going to be colossal and not profitable. The better solution would be just to destroy old discarded parts in a way that has a low carbon footprint.

You don't need it to be compatible with new tech to be useful. Used RAM is still usable as replacement parts for repairs, or can be re-used in products using older but still useful tech. The same is true for CPUs and GPUs - heck, look at the wave of Chinese motherboards using older mobile CPUs, as well as those new boards using older chipsets (including some really cool HEDT configurations). Just because a part is a few years old doesn't make it useless - and there's a reason the mantra goes "reduce, reuse, recycle" - recycling is the last step, as it's the most wasteful, with reuse being far better. And, ultimately, this could even amount to reduction, as making new PCs with older but still good components can cut the need for production of new parts outside of highly performance-sensitive situations.

Of course, when it comes to smaller components like capacitors, resistors, mosfets, chokes, and all the various controllers found in a PC, the opportunities for reuse are vast. This is obviously labor-intensive, but ... it needs to be done. We literally can't keep relying on digging new materials out of the ground forever. It simply isn't possible. And we need to face the realities of the continued usefulness of older products and the inherent benefits to keeping them in use vs. replacing them with nominally "better" things that perform the exact same task with minimal perceptible difference.

Also: this is already done quite a bit. Used electronic components are quite easily found in China, including in new products. The problem with how this currently works is compound: there's litte quality control involved; the parts are often rebranded or just cleaned up and sold as new; there's generally a lot of shady business involved in this. All of these are solveable problems if the scale of this was increased and the products came with the necessary warranties and QC to make manufacturers trust them (which of course also means strict requirements for the disassembly lines to ensure components aren't damaged when they are removed).

Also: f**k profits. Profits are wholly unnecessary in this. As with all highly valuable environmental and pro-consumer programmes, something like this should be heavily subsidized, at least until it's built up to a scale where it can be self-sustaining. And even at that level you don't need to be profitable - breaking even is good enough to stay afloat. The desperation for profit is a huge part of why capitalism has turned into the death cult it is today - highly beneficial things are seen as "not good enough" unless they can show continuous growth (which, in a finite universe and on a planet with finite customers and finite resources, is ... well, a logical impossibility), or are just dismantled and sold off by profiteering owners looking for a bigger payout. We really, really, really need to stop demanding that everything be profitable. In the end, that is a deeply harmful way of thinking.

Isn't Apple doing that already?

...so you entirely missed the part where vertical integration essentially removes this problem? I mean, I can elaborate if you need it, but that should be pretty clear. Apple gets to determine on their own the number of SKUs offered, and that's that. They don't have to listen to the desires of OEMs or partners, don't have to adapt their SKUs unless they want to, and can delimit things essentially entirely freely. Neither Intel nor AMD have that luxury - hence why they also have far more CPU SKUs than Apple does, even counting the different memory amounts. Of course vertical integration alone isn't necessarily enough - and Apple is also helped by their iron grip on some core customer groups, especially in the creative industries, where they also then get to set the parameters for software development and hardware needs.

 

[goodeedidid]

...so you entirely missed the part where vertical integration essentially removes this problem?

Why would I have missed the point? Apple is integrating vertically and they are doing exactly what I pointed out about, pretty much to a degree, a complete SoC, which is where the future is heading. Although they aren't really vertically integrated, IMO, because Apple isn't really producing anything, they are employing third parties throughout their whole production process. Apple is a design company. They design everything but they aren't producing anything, as far as I'm aware. I'm pretty sure Intel/AMD can produce their own memory and SSD with they idea of making better and more efficient system. I'm confident that in the future CPU, MEM, DISK, GPU.. are all going to be unified on single system package and the whole idea of using reusable parts is going to be technically impossible, but that isn't going to matter because a complete unified system isn't going to fail at all and the failure rates are going to be non-sexitent because you wouldn't have different components that increase the risk of failure.

Also: f**k profits.

That's not how the real world works. People and companies don't perform for free. Everything is about profits. Every company exists to make a profit out there. Discarded components have to be destroyed in some clean way that isn't going to cause too much pollution on the environment. I can't imagine what kind of a colossal mess reusing old parts that have been discarded is going to look like. It just makes no sense. Innovation and new tech can't use old components. I guess there could some market for reused parts but that would be a small percentile fraction of the whole industry and only small time repair shops are going to be into that kind of niche market. It is going to be much cheaper to make new iPhones and iPads with new components than using reused components if that is actually possible at all. Not to mention that such big companies as, for example, Apple or Samsung need to plan production, and nobody can't wait on volatile supply of old components. The whole idea just sounds like nonsense.

 

[Valantar]

Why would I have missed the point? Apple is integrating vertically and they are doing exactly what I pointed out about, pretty much to a degree, a complete SoC, which is where the future is heading. Although they aren't really vertically integrated, IMO, because Apple isn't really producing anything, they are employing third parties throughout their whole production process. Apple is a design company. They design everything but they aren't producing anything, as far as I'm aware. I'm pretty sure Intel/AMD can produce their own memory and SSD with they idea of making better and more efficient system. I'm confident that in the future CPU, MEM, DISK, GPU.. are all going to be unified on single system package and the whole idea of using reusable parts is going to be technically impossible, but that isn't going to matter because a complete unified system isn't going to fail at all and the failure rates are going to be non-sexitent because you wouldn't have different components that increase the risk of failure.

... so you missed the point. Didn't need that paragraph, a sentence would have sufficed. But to be clear: what AMD and Intel lack compared to Apple is the following: they are component manfucaturers; Apple is a systems manufacturer. That means Apple has full determining power over the specifications of the systems in which their products are used. Neither Intel nor AMD have anything approaching this, beyond setting minimum and maximum supported amounts for things. Whether Apple outsources production or not is entirely irrelevant, as Apple still has full determining power over the specifications of what is produced.

What are the effects of this? AMD and Intel need to provide a broad range of SKUs for OEMs to select from (and the option for custom SKUs for those who want it, like Apple used to do with their Iris Pro Core chips), while Apple can make however many they want and that's that. This is easily demonstrated by Apple having far fewer SKUs across the board. Apple has two CPU bins for its M1 - with 7 or 8 GPU cores active - and a choice of 8 or 16GB RAM for that, across MBA, MBP and Mac Mini. 4 SKUs. In comparison, Intel has 12 U-series 12th gen Core chips in their roster. Twelve. Apple has two M1 Pro models - 8/14 CPU/GPU cores and 10/16 - each of which again come in 16 or 32GB of RAM. Intel has 5 P-series 12th gen SKUs - close, but that's without accounting for memory. And the list goes on. The point being: having full control of final product specifications allows you to drastically reduce the number of SKUs you need to put out, which is a significant cost savings. And remember that if memory was integrated onto the chip, even for two tiers Intel's SKU numbers would double.

I also think your vision of unification is too extreme - as I said above, the benefits of packaging storage on an SoC are very small and pale in comparison to the drawbacks. Memory and GPU are another matter, but are also subject to practical considerations (number of SKUs that need to be produced, packaging size and cost, etc.). I definitely think we're going to see an increased architectural unification going forward, but I think your vision is far too limited here.

That's not how the real world works. People and companies don't perform for free. Everything is about profits. Every company exists to make a profit out there. Discarded components have to be destroyed in some clean way that isn't going to cause too much pollution on the environment. I can't imagine what kind of a colossal mess reusing old parts that have been discarded is going to look like. It just makes no sense. Innovation and new tech can't use old components. I guess there could some market for reused parts but that would be a small percentile fraction of the whole industry and only small time repair shops are going to be into that kind of niche market. It is going to be much cheaper to make new iPhones and iPads with new components than using reused components if that is actually possible at all. Not to mention that such big companies as, for example, Apple or Samsung need to plan production, and nobody can't wait on volatile supply of old components. The whole idea just sounds like nonsense.

That kind of myopic, calcified thinking is precisely why the world is barreling at highway speeds towards a civilization-derailing ecological disaster. I would really, really suggest you take some time to ask yourself why you're so set on maintaining systems that are not only by far not the only alternative we have, but are actively harmful to both people and the environments we live on a staggering number of levels. I would strongly recommend Mark Fisher's Capitalist Realism as an entry point into thinking on this, as it does an excellent job explaining how the ideology of late-stage neoliberal capitalism manages to convince those living within it that there is no possible alternative way of managing the world (despite this being factually untrue).

One example of your blind spots: "People and companies don't perform for free" is in no way an argument against not prioritizing profits. Profits are what is left after paying your employees and other expenses. It is entirely possible to run a non-profit organization where every employee has a good salary, and where the business itself is sustainable. The only people benefiting from a company being profitable are shareholders/owners as well as executives who are typically awarded bonuses for profitability. Profits themselves are not at all necessary for a stable and sustainable business. Revenue is necessary for that.

Further: just the fact that you keep saying nonsensical stuff like "discarded components have to be destroyed in some clean way" - what does "destroy" mean? Because the implication of the word is that it just goes away, which ... well, that's not how physical reality works. What happens to what is left over? That is the pollution, after all. Doesn't harvesting those raw materials and re-using them sound like a good idea for that, rather than "storing" (read: dumping) them somewhere? And again: it's first reduce, then reuse, then recycle. Recycling is the most wasteful and least efficient way of cutting emissions - it's resource intensive, expensive, and inefficient. Reuse of working components is an entirely acceptable way of going forward, and with sufficient QC and well established standards can be perfectly safe. The line "innovation and new tech can't use old components" is not only factually untrue (why can't a new laptop have some proportion of used, tested, known-good resistors, mosfets, etc?), but just serves as a rhetorical device to derail useful discussion by hand-waving at "progress" and "innovation". The infrastructure needed for something like this to work doesn't currently exist, and would take time and be expensive to build up, but that doesn't make it an impossibility.

As for your corporate apologism, which essentially boils down to "boo hoo corporate profits might shrink" - again, that's not an argument for or against anything. Shrinking access to resources is a fact that we and they have to deal with. Increasing raw material prices as well. As is the necessity of keeping toxic materials out of landfills and our environment more generally. We need to work towards fixing these problems. That this might be inconvenient to some of the wealthiest companies on the planet really shouldn't bother any of us. They can afford it, easily, and their only reason for complaining is their deep-seated ideological fetishization of profits for their own sake. And, just to be clear: companies exist first and foremost to provide useful products and services, and secondly to provide income to people to ensure a decent quality of life. Profits come, at best, third in line. And if the first two points can't be done without driving us towards ecological disaster, well, then good riddance. We can live without them and their greed.

 

[R0H1T]

I'm confident that in the future CPU, MEM, DISK, GPU.. are all going to be unified on single system package and the whole idea of using reusable parts is going to be technically impossible, but that isn't going to matter because a complete unified system isn't going to fail at all and the failure rates are going to be non-sexitent because you wouldn't have different components that increase the risk of failure.

The good thing is neither Intel, AMD nor Nvidia have to listen to you. There's good reason to go SoC or PoP route, namely efficiency, but there's enough good reasons to have them separate on a mainboard for desktop/workstation or even servers. I'll give you hint what that is ~ probably the biggest cost of running a DC, & no more components on a single chip will not solve that!

 

[iiee]

Who really does that anymore?

Of course there are people who do that. Why buy expensive RAM from Dell ?? Just buy a minimum 4GB ram then go after market to buy 64GB, much more cheaper if buy 64 from Dell. Same as CPU (before soldered, buy cheapest and replace with extreme), SSD, graphic card etc. I do this for every single precision I bought over the years.

To claim warranty, just put back original parts when do the diagnosis or when the technician comes.

 

[goodeedidid]

But to be clear: what AMD and Intel lack compared to Apple is the following: they are component manfucaturers; Apple is a systems manufacturer. That means Apple has full determining power over the specifications of the systems in which their products are used.

It isn't necessary for you to repeat the things I brought up. I'm not making a case Apple vs Intel/AMD... no need for you to derail the subject. Apple is the good example I gave for the benefits of unified system. Intel/AMD are a completely another matter and not relevant to the discussion unless you want to change the subject.

the benefits of packaging storage on an SoC are very small and pale in comparison to the drawbacks

How are the benefits paling in comparison to the drawbacks?

That kind of myopic, calcified thinking is precisely why the world is barreling at highway speeds towards a civilization-derailing ecological disaster. I would really, really suggest you take some time to ask yourself why you're so set on maintaining systems that are not only by far not the only alternative we have, but are actively harmful to both people and the environments we live on a staggering number of levels.

You confuse myopic with realistic. I get the sense that you're very idealistic and idealism never really leads up to good workable solutions. It's easy to say that nothing should be outright thrown away, components should be reusable, and the user should have the absolute right to repair... blablabla. That's all fine and dandy to envision it over a beer and to talk about it but right now this isn't how things work out in the real world.

One example of your blind spots: "People and companies don't perform for free" is in no way an argument against not prioritizing profits. Profits are what is left after paying your employees and other expenses. It is entirely possible to run a non-profit organization where every employee has a good salary, and where the business itself is sustainable. The only people benefiting from a company being profitable are shareholders/owners as well as executives who are typically awarded bonuses for profitability. Profits themselves are not at all necessary for a stable and sustainable business. Revenue is necessary for that.

This is exactly the right argument because this is the core goal of every business and corporation out there. Why do you think that, for example, tobacco companies are spending billions for lobbying in the pursuit of their own interests? Also I didn't say anything against NPOs.. I'm all for NPOs but I don't see how you envision NPOs recycling components.

The only people benefiting from a company being profitable are shareholders/owners as well as executives who are typically awarded bonuses for profitability.

You somehow forgot the fact the companies have employees, such as engineers, administration, and general staff. Do you think those people go there in the morning to work and get paid lol? Guess what, in order for a company to pay well their employees those companies have to make a profit. I don't see what's so hard to understand, this is literally how the world works anywhere even in "socialist" countries like China.

 

[Valantar]

It isn't necessary for you to repeat the things I brought up. I'm not making a case Apple vs Intel/AMD... no need for you to derail the subject. Apple is the good example I gave for the benefits of unified system. Intel/AMD are a completely another matter and not relevant to the discussion unless you want to change the subject.

... you're making broad-reaching statements about the future direction of chip development. Thus I don't care about your examples - your statements are formulated so that they must apply to the industry more broadly. And I'm trying to get through to you that for various reasons, that is ultimately quite unlikely, as these companies operate in very different ways.

How are the benefits paling in comparison to the drawbacks?

For storage? Lack of upgradeability, lack of data recovery options, thermal stress, and last but not least the massive area necessary to add a meaningful amount of flash to a CPU package. Compare the area of an m.2 drive to the area of a CPU - the SSD is larger, and only does storage. Sure, integrating the controller will save some area, and using HMB will save some more, but sticking flash packages onto a CPU package is still a bad idea in terms of cost. It will necessitate absolutely massive CPU packages for no real benefit. Because the benefits will be ... uh ... nothing? Some minuscule power savings from integrating the controller into the SoC? Some purported but irrelevant to 99.99999% of users security advantages? Performance won't change meaningfully, nor will access times - flash is the bottleneck there.

You confuse myopic with realistic.

No. Seriosuly. Go to your local book store, buy a copy of capitalist realism. Now. It will be good for you. You are literally acting out the mechanisms it describes, demonstrating an absolute incapability of imagining alternative ways of organizing the world. That you're describing it as "realistic" just drives home Fisher's point.

I get the sense that you're very idealistic and idealism never really leads up to good workable solutions. It's easy to say that nothing should be outright thrown away, components should be reusable, and the user should have the absolute right to repair... blablabla. That's all fine and dandy to envision it over a beer and to talk about it but right now this isn't how things work out in the real world.

Jesus, man... have I ever said that "right now this is how things work out in the real world"? Please put away your straw man building kit and actually read. I'm saying that if we actually make relevant efforts, this is possible, and would bring about positive change. I have never said every new component could be replaced with a used one; I have never claimed all components can be reused; I have simply claimed that building up a large-scale and strictly regulated electronics recycling industry is a necessary and highly beneficial thing to try and do on many levels. This isn't idealistic, it is looking for opportunity while also trying to work towards solving or at least alleviating several major global problems. Has anything I have said not accounted for this necessarily being a massive and expensive undertaking? No; I have been very explicit about it necessarily being so. I still think it is worth doing, as our current way of treating these things is running us into the ground.

This is exactly the right argument because this is the core goal of every business and corporation out there. Why do you think that, for example, tobacco companies are spending billions for lobbying in the pursuit of their own interests? Also I didn't say anything against NPOs.. I'm all for NPOs but I don't see how you envision NPOs recycling components.

So ... just to be clear: because in a world where corporate lobbyism for more than half a century has led to the entrenchment in law that publicly traded companies need to chase profits above all else, we necessarily need to accept this law as just and fair? Sorry, but I just don't buy that. This thinking is harmful. Heck, look at how the distribution of wealth has developed in the same period; how wages for ordinary empoloyees have changed. These laws do one thing, and one thing only: enrich the rich, making them more powerful in fundamentally antidemocratic ways. This, again, is harmful. It literally ruins the lives of millions if not billions of people. Your fatalist "but that's just how things are" reasoning doesn't change any of this. And no, it isn't realism, it is just an absolute and utter failure at imagining alternative modes of organizing the world. Again: go read Capitalist Realism. Please. Seriously - you're enacting the exact ideological blindness Fisher describes. Just because corporations and rich people like the way the world is organized does not mean it is the only, nor even the best way of organizing it. Capitalism works really, really hard to make us blind to other modes of organizing society, but that doesn't make the messages it puts out any less untrue.

You somehow forgot the fact the companies have employees, such as engineers, administration, and general staff. Do you think those people go there in the morning to work and get paid lol? Guess what, in order for a company to pay well their employees those companies have to make a profit. I don't see what's so hard to understand, this is literally how the world works anywhere even in "socialist" countries like China.

Okay, apparently I do have to repeat myself: revenue pays employees' wages. Profits are what is left over after expenses. Is this so hard to grasp? I literally addressed this in the post you're quoting.

 

[trsttte]

It isn't necessary for you to repeat the things I brought up. I'm not making a case Apple vs Intel/AMD... no need for you to derail the subject. Apple is the good example I gave for the benefits of unified system. Intel/AMD are a completely another matter and not relevant to the discussion unless you want to change the subject.

How are the benefits paling in comparison to the drawbacks?

You confuse myopic with realistic. I get the sense that you're very idealistic and idealism never really leads up to good workable solutions. It's easy to say that nothing should be outright thrown away, components should be reusable, and the user should have the absolute right to repair... blablabla. That's all fine and dandy to envision it over a beer and to talk about it but right now this isn't how things work out in the real world.

This is exactly the right argument because this is the core goal of every business and corporation out there. Why do you think that, for example, tobacco companies are spending billions for lobbying in the pursuit of their own interests? Also I didn't say anything against NPOs.. I'm all for NPOs but I don't see how you envision NPOs recycling components.


You somehow forgot the fact the companies have employees, such as engineers, administration, and general staff. Do you think those people go there in the morning to work and get paid lol? Guess what, in order for a company to pay well their employees those companies have to make a profit. I don't see what's so hard to understand, this is literally how the world works anywhere even in "socialist" countries like China.

You want to use Apple as a good example of integration? Fine, even they recognized it made no sense to integrate the memory into the rest of the system and have replaceable memory modules (the controller is integrated in the main soc which is kinda dumb since it makes data recovery in a failure scenario damn near impossible but anyway, their users don't seem smart enough to care and Apple pretty much fully pivoted away from any professional market outside creative areas). It's baffling to me how we're moving away from basic concepts of modularity and reparability and somehow trying to justify that as a good thing because laptops will be half a centimeter smaller for example. This is not even about idealism, making components that can work together and can be reused is good engineering, making things super integrated for integration sake is about shortcuts and guaranteeing future sales.

You also seem to misunderstand the difference between profit and revenue. A company can operate with no profits without any problems and continue to pay it's employees, heck that's what most big tech companies have been doing (technically, they do have profits but write them all off as investment in the business), as long as they have enough revenue to pay the liabilities (including salaries) everything's fine (they probably aren't growing but that's a different conversation that would lead to sustainable growth for example and different definitions on that)

 

[Valantar]

You want to use Apple as a good example of integration? Fine, even they recognized it made no sense to integrate the memory into the rest of the system and have replaceable memory modules (the controller is integrated in the main soc which is kinda dumb since it makes data recovery in a failure scenario damn near impossible but anyway, their users don't seem smart enough to care and Apple pretty much fully pivoted away from any professional market outside creative areas). It's baffling to me how we're moving away from basic concepts of modularity and reparability and somehow trying to justify that as a good thing because laptops will be half a centimeter smaller for example. This is not even about idealism, making components that can work together and can be reused is good engineering, making things super integrated for integration sake is about shortcuts and guaranteeing future sales.

IMO, some of the blame for the moves towards non-replaceable components can be placed on outmoded standards that are poorly suited to new tasks and implementations. SO-DIMMs are an excellent example of this - they haven't changed since ... DDR2?, yet laptops now look ... well, rather different than in 2007. Fitting an SO-DIMM back then was trivial; today it's impossible in many designs. This new form factor looks to alleviate thickness and area pressures from an old standard, which is an excellent initiative (especially when it also removes a future performance bottleneck). Hopefully it sees broader adoption or leads to the creation of a new standard.

 

[goodeedidid]

No. Seriosuly. Go to your local book store, buy a copy of capitalist realism. Now. It will be good for you. You are literally acting out the mechanisms it describes, demonstrating an absolute incapability of imagining alternative ways of organizing the world. That you're describing it as "realistic" just drives home Fisher's point.

If capitalism with that zest neoliberalism isn't, today, the only viable economic system then what is the alternative? We all have seen what a total failure the USSR was and what happens when the individual has no responsibilities and when the government tries to control your life. A total and utter failure fueled with corruption, total lack of responsibilities, and paying colossal taxes, which kind of drives home my point that your ideas of "we all coming together" (I don't know what you mean by we...) to fix our problems isn't a realistic and working view of the world.

For storage? Lack of upgradeability, lack of data recovery options, thermal stress, and last but not least the massive area necessary to add a meaningful amount of flash to a CPU package.

The only point I can really agree on is that storage isn't going to be upgradable, but having in mind that backup drives and USB4/TB4 are so fast now system storage drive does not need to be upgradable because for mass storage and backups you would use external drives. I just don't see what really warrants nowadays in 2022 upgradable OS storage. You point out thermal stress but have you seen is that really true? If the controller is in the CPU anyway I don't think it will run that hot because the CPU will be cooled anyway, just as Apple Silicon is running x8 times more efficiently than Intel/AMD counterparts.

'm saying that if we actually make relevant efforts, this is possible, and would bring about positive change. I have never said every new component could be replaced with a used one; I have never claimed all components can be reused; I have simply claimed that building up a large-scale and strictly regulated electronics recycling industry is a necessary and highly beneficial thing to try and do on many levels. This isn't idealistic, it is looking for opportunity while also trying to work towards solving or at least alleviating several major global problems. Has anything I have said not accounted for this necessarily being a massive and expensive undertaking? No; I have been very explicit about it necessarily being so. I still think it is worth doing, as our current way of treating these things is running us into the ground.

Again you're broadly and subjectively saying things that aren't true. Tell me, if all is true what you say then why is it not the case now? Why aren't we making today the relevant efforts to bring beautiful Earth-loving changes? It's all "if" with you. But that's okay. IF we all come together as people we could end all wars, poverty, and hunger but that is an IF and this is not how things work out obviously and have never ever worked in human history neither. So your points are just IF-talk and nothing more.

Okay, apparently I do have to repeat myself: revenue pays employees' wages. Profits are what is left over after expenses. Is this so hard to grasp? I literally addressed this in the post you're quoting.

If profits are low then that directly affects wages, company expenses, and even employment. So what is your point?

This is not even about idealism, making components that can work together and can be reused is good engineering, making things super integrated for integration sake is about shortcuts and guaranteeing future sales.

For building your own computer system that is okay or for upgradable servers I would assume to elongate the life of server infrastructures but that use case scenarios are highly industrialized and I would agree to that. I do build my own rigs and I like it. But there is a broader argument to be made with devices such IoT devices and all the mobile devices such as phones, tablets, and ultrabooks all would benefit from a complete SoC.

even they recognized

The M1 has unified memory, what are you saying?!?

You also seem to misunderstand the difference between profit and revenue. A company can operate with no profits without any problems and continue to pay it's employees, heck that's what most big tech companies have been doing (technically, they do have profits but write them all off as investment in the business), as long as they have enough revenue to pay the liabilities (including salaries) everything's fine (they probably aren't growing but that's a different conversation that would lead to sustainable growth for example and different definitions on that)

This isn't entirely true. No company has a long-term plan of functioning without profits. What you're talking about are just short-term strategies to make ends meet. Would be good to give examples.

 

[Mussels]

If they get JEDEC support behind it, at least dell wont be the only one selling the parts and it'll all work out more or less ok for upgrading as consumers

 

[Valantar]

If capitalism with that zest neoliberalism isn't, today, the only viable economic system then what is the alternative? We all have seen what a total failure the USSR was and what happens when the individual has no responsibilities and when the government tries to control your life. A total and utter failure fueled with corruption, total lack of responsibilities, and paying colossal taxes, which kind of drives home my point that your ideas of "we all coming together" (I don't know what you mean by we...) to fix our problems isn't a realistic and working view of the world.

Sorry, but your failure of imagination is not my responsibility. And you just keep playing out the exact rhetorical and logical mechanisms Fisher describes. It's uncanny, really. Pointing backwards to failed attempts in specific sociohistorical-cultural contexts and pretending that they are universally applicable; pretending that coming up with new ideas and solutions is impossible. None of this is true. I could absolutely go into specific detail as to why the USSR economic system failed (in short: corruption, though it's far more complex than just that, and ties directly into when and where it came to pass), but that's way beyond the point here. The point is: I'm suggesting alternative ways forward, and your response is a myopic, locked-down "no, that's impossible, because that's not how we currently do things". Your failure of imagination is on you, not me. It has been imprinted on you by the dominant ideology in the society that you live in, but it is nonetheless entirely possible for you to break those destructive and fatalist patterns of thinking.

The only point I can really agree on is that storage isn't going to be upgradable, but having in mind that backup drives and USB4/TB4 are so fast now system storage drive does not need to be upgradable because for mass storage and backups you would use external drives. I just don't see what really warrants nowadays in 2022 upgradable OS storage. You point out thermal stress but have you seen is that really true? If the controller is in the CPU anyway I don't think it will run that hot because the CPU will be cooled anyway, just as Apple Silicon is running x8 times more efficiently than Intel/AMD counterparts.

Backups? Where did I mention backups? I said data recovery. Good luck restoring data from on-package flash when a hardware failure occurs ... And yes, everyone loves being dependent on external storage devices dangling off their laptops at all times, that's such a great solution. It's also rather interesting that you entirely failed to address the largest concern by far here: packaging area. Again: how would you propose fitting, say, 1TB of flash with a wide enough interface for it to be fast onto a CPU/SoC package without that package becoming massive and thus ridiculously expensive?

Again: there are major drawbacks to on-package storage, and essentially zero advantages. Even in the predatory neoliberal late-stage capitalist hellscape we are currently heading towards (and partly living in), where does it make sense for companies to spend tons of money integrating storage onto packages when there are no benefits to this?

Again you're broadly and subjectively saying things that aren't true. Tell me, if all is true what you say then why is it not the case now?

Because global politics is a complete f*****g mess, and there are few organizations or bodies capable of enacting something like this on a sufficient scale for it to work well. Thankfully the EU has been working towards being such a global actor, though as with all politics, these things take a lot of time.

Why aren't we making today the relevant efforts to bring beautiful Earth-loving changes?

There are a lot of people trying to do so. But they are up against massively wealthy corporations with extremely powerful lobbyists in extremely important parts of the world (the US in particular, where corporate lobbying is completely out of control, but it's not much better elsewhere). This means that even when there is overwhelming evidence for why something would be near-universally beneficial, it often doesn't get implemented if some major corporate actor objects to it. Just look at how many decades it has taken to fight tobacco companies! It was widely known that smoking tobacco causes lung cancer in, what, the 1960s? 50s? Something like that. Yet it took 3-4 decades to even get tobacco companies to admit this, let alone get them to take any kind of responsibility for pushing their poison on people. E-waste is a far, far newer problem than tobacco.

It's all "if" with you. But that's okay. IF we all come together as people we could end all wars, poverty, and hunger but that is an IF and this is not how things work out obviously and have never ever worked in human history neither. So your points are just IF-talk and nothing more.

You know, it would be nice if you could refrain from the ridiculous infantilization and plain-faced bad-faith "arguments" here. Framing this as if what I'm proposing is some utopian pie-in-the-sky idea that could never be put into practice is just plain-faced not true - you're just operating as if the current way of the world is some sort of natural base order, rather than an ideological construction that has been built by large-scale concerted efforts over decades or even centuries. If an effort even .1% of that spent for the past 50 years in shoring up corporate profits and entrenching the political power of corporations was spent on fixing E-waste, this would be solved in a few decades. Entirely. Seriously, you need to take your blinders off and accept that the way things work today is not only deeply flawed, but not the only way things can work. This isn't the gargantuan undertaking you're making it out to be. It would take time, money, and a lot of effort, but it's entirely doable - but it isn't doable if all you're focused on is "but shareholders need to get richer!" When you're starting from that premise, you've already given up on so much of what is possible in the world.

If profits are low then that directly affects wages, company expenses, and even employment. So what is your point?

No. That is not a necessary causal relation. That is a causal relation within a specific subset of capitalist systems, in which profits are treated as a more important goal than providing employment, making good products, or fulfilling a useful function in society. I do not subscribe to this antidemocratic, harmful extremist belief, and thus do not agree that low profits necessarily directly affect wages. As I've said: you can have zero profits and still have a stable business. Stable business means no growth, but no losses either. No cutting of wages, no cutting of production, but no major investments (that aren't funded by grants or loans) either. In a system that isn't dead-set on maintaining the ridiculous, anti-scientific idea of infinite growth, this would not be problematic whatsoever.

For building your own computer system that is okay or for upgradable servers I would assume to elongate the life of server infrastructures but that use case scenarios are highly industrialized and I would agree to that. I do build my own rigs and I like it. But there is a broader argument to be made with devices such IoT devices and all the mobile devices such as phones, tablets, and ultrabooks all would benefit from a complete SoC.

To some extent, yes. As I've said before, I agree that there's a general trend towards integration, and I think we'll see more on-package memory in the future. (I also hope that means we'll see more two-tier memory setups, with both on- and off-package memory in the same system - but that would need low level OS support, of course.) But I do not believe we'll ever see flash storage on-package for anything but the most extremely compact microcontroller-like implementations. Smartphones stack RAM, but have off-package storage. As do IoT devices (except those that make do with some tiny amount of on-die storage). It just doesn't make sense to integrate this.

The M1 has unified memory, what are you saying?!?

They're talking about storage. The M1 integrates the controller, but keeps the flash off-package, as putting flash on the SoC package just makes no sense.

This isn't entirely true. No company has a long-term plan of functioning without profits. What you're talking about are just short-term strategies to make ends meet. Would be good to give examples.

Essentially every software/service startup operates this way. Twitter, Uber, Twitch, there are dozens and dozens of examples. Most of them never make any money whatsoever. Facebook operated like this for years until they built up their ad sales, hemorrhaging investor money. Twitter has been doing this to this day, living off investor money granted to them based on stock valuations and vague profits of "at some point we'll figure out how to make money off of this". But you know the real kicker here? Their problem isn't profits - their problem is that they don't have a revenue stream. Again: revenue is what you base things on. Profits are what is left over when you've paid your expenses. Failed businesses don't fail because they're not profitable - that's just a figure of speech, mainly - they fail because they have insufficient revenue.

 

[kanecvr]

How can you be so shortsighted? People don't use laptops unplugged all the time; I travel a lot and still spend 80-90% of the time plugged in with mine knowing that when on batteries I can't do anything intensive without limiting myself to a couple of hours battery life.

Thin and lights aren't exclusively for use on the go, they're usually someone's main computer and they need to be able to do everything.

What are you even talking about? I never mentioned anything about being plugged in. And being plugged in has noting to do with it.

I've tried thin and light machines - anything from ultra low power 15W i7's to the 14" 8 core Asus ROG Zephirus - compared to a workstation or gaming laptop performance is a noticeably reduced. Even in the case of "performance" (notice the quotes) oriented t&l machines, there is simply not enough thermal headroom and power circuitry to enable any serious computing*.

I'll give an example - take the HP x360 - at 15W ryzen 3750h will perform significantly worse then the same CPU but with a higher power limit - 65W like in the Asus TUF gaming. You can add as much ram to that as you want, it's still a dumpster fire.

*And by serious computing I'm talking rendering, compiling and so on - tasks that would benefit from having large amounts of ram. I've seen this kind of talk back in the 2000's - MORE RAM IZ MOAR POWR!!!! - marketing BS designed to make people pay for stuff they don't need. For every day computing 8GB is more then enough. In fact, the PC I use to study and write research papers on is an ancient i7 950 with only 6GB of ram, and it's excellent. Not even having dozens of tabs in edge open at the same time while doing image optimization and OCR in Acrobat DC will slow it down. I've been thinking of upgrading this PC as it's been in use since my collage days, but it's been running so well for the tasks it's expected to do that I feel no need to do so, and it has all the software I'm used to on it, as well as all my research papers and materials from studies and whatnot, so instead of throwing more ram at it or swapping it out for a newer gen machine I simply installed a PCI-E USB 3.1 card and upgraded the CPU cooler.

I brought this PC up because it has about the same performance level as a modern quad core thin and light laptop running a modern i5 or ryzen 5 "U" skew CPU - and 6Gb is not a hindrance. It it were I would have dug into my box 'o' ram and installed an additional 6 GB.

Also I formulated my reply as a serious valid QUESTION - what do you need that much ram for? I'm still waiting for an answer. I gave you a practical example of using 6Gb in 2022 for Acrobat DC, image recognition, OCR, photoshop, document digitalization and on line research (witch often involves having over 50 open browser tabs or multiple browser instances running at the same time) with no slowdowns or other issues.