[Neywiny]

HBM is also DRAM. I also think it's kind of a weak argument to say that them discontinuing ddr3 (which while in use still today in industrial/embedded was on the way out for consumers 10 years ago) and ddr4 which last had consumer CPUs for it 3 years ago is meaningful. What we need now is ddr5. Turning off the old fabs and moving those resources including people to ddr5 is a good thing. That's not price fixing. It's possible price fixing is in play, but discontinuing products people objectively don't use as much anymore isn't it.

[cosmic_cheese]

Switching off DDR3 manufacturing I can understand, but DDR4 machines are still quite relevant and usable… Ryzen 5000 series boxes for example don’t feel meaningfully weaker than they did when new. My 5950X tower certainly doesn’t, and it’d really be nice to be able to upgrade its RAM should I need to because it will continue to be useful for quite some time.

AMD just re-released their 5800X3D for AM4 board users who wish to upgrade which is further evidence that shutting off DDR4 production is premature.

[zarzavat]

They're running a business not a charity. Their job is to manufacture what the market as a whole demands. If they can make more money making HBM than DDR4 then they have to make HBM. Why would a business go out of its way to make less money?

[cosmic_cheese]

1. The AI bubble is an insane distortion and the gravy train isn’t going to last forever. Betting the farm on everlasting datacenter demand is myopic.

2. In a healthy, competitive market there would be smaller manufacturers that’d be happy to take up the big guys’ discarded business.

[Chyzwar]

1. But it might last for at least few more years, see Nvidia 1 trillion backlog.

2. Semiconductor manufacturing is the most complex industrial process in the world. You need billions of capex and decades of experience. Even existing semi players like intel cannot switch production to memory.

China CXMT is gaing traction in DDR market. New fabs from all players wil come online in the next two years.

[fennecbutt]

>the gravy train isn’t going to last forever.

Have you seen how the modern stock markets works lmao? It hasn't been based on reality in a long time.

Hell just look at Trump, should've run outta money from all his bad deals ages ago but the grift continues.

[cosmic_cheese]

It's true that markets have no need to be rational, but everything has an end and it's usually sooner than anybody ever expects.

[mschuster91]

People have been saying AI is a bubble for at least a year now and, unfortunately, it still hasn't collapsed.

[iwontberude]

Deleted

[dktbs]

Not sure I understand this statement. Apple is affected by DRAM price fixing the same as any other PC manufacturer.

[iwontberude]

Ahh you are right, not sure where I got that from…

[toast0]

DDR4 machines are relevant and usable, but it's pretty unusual that new systems on older ram are still big sellers. The 5800x3d was originally released in 2022, AM5 processors were also originally released in 2022. We'll probably see Zen6 soon, the third generation on AM5... I don't think AMD released new AM3 processors after the introduction of AM4, certainly not when Zen3 was coming soon.

When making long term plans in 2022, I don't think anyone expected DDR4 to need significant production in 2026. Since ram makers can pretty much sell whatever they make in today's marketplace, it makes sense (for those fabs that can) to stop making DDR4 and repurpose those fabs to make newer generation ram.

[thewebguyd]

> My 5950X tower certainly doesn’t

That's what I have in my gaming tower, and yeah I feel zero pressing need to upgrade. I did manage to put 64GB of DDR4 in it just before prices went totally bonkers, thankfully. Where I'm falling behind is my GPU I'm still on an nvidia 1660 super, but I just can't justify paying what they cost right now.

I would gain pretty much nothing moving to a newer board w/ DDR5.

[cosmic_cheese]

Exactly what I'm feeling. The CPU is great for what I'm using it for and I have a killer motherboard. Have a nice GPU capable of current gen games, too. The equivalent current gen board and CPU cost a small fortune for marginal readily visible gain.

Unfortunately, I opted for only 32GB of RAM because at that point more felt like overkill, which as a decision has aged poorly. I should've gotten more while it was still cheap.

[Neywiny]

Not to be a bootlicker but AMD releasing a product doesn't mean another company should make more DDR4. That's not price fixing. In the embedded space it's sadly very common for a part to be compatible with a very low number of options (shout out to cellular/admux RAM on the STM32H745 nucleo). That's just the way the cookie crumbles.

[pdimitar]

Dropping DDR-4 is anything but meaningful. It'll easily last 10 more years, machines from this gen are still much more affordable and quite powerful. In fact for most dev and gaming workflows the difference between the DDR-4 and DDR-5 generation of hardware is more or less negligible. I am exaggerating a bit -- but really, not too much.

Of course it might be a ploy to sheep-herd consumers and companies towards the expensive DDR-5. I would not put that below the ring of RAM producers.

[gruez]

>Dropping DDR-4 is anything but meaningful. It'll easily last 10 more years, machines from this gen are still much more affordable and quite powerful. In fact for most dev and gaming workflows the difference between the DDR-4 and DDR-5 generation of hardware is more or less negligible. I am exaggerating a bit -- but really, not too much.

How much % of the DRAM market do you think is made from computer enthusiasts upgrading their Zen 1/2 CPUs to Zen 3? Note intel and AMD both switched to DDR5 well before the exit from DDR3/DDR4 ("2024-2025", according to the complaint).

[cosmic_cheese]

Note though that for Intel, the first gen of DDR5 CPUs also supported DDR4, and many buyers bought the DDR4 versions of their boards because at that point DDR5 RAM was much more expensive for gains that were marginal at best, which effectively makes Intel’s following generation of DDR5 CPUs the actual transition point.

[pdimitar]

The mere "enthusiast" word in your question suggests the percent is not too big. But I am not sure I get your point -- elaborate, please?

[gruez]

The point is that just because there's a handful of people (relatively speaking) looking to buy more RAM to upgrade their last gen systems, doesn't mean there's robust demand for DRAM manufacturers to keep DDR4 manufacturing lines going. It's like arguing Sony shouldn't have exited the CRT business because there's retro enthusiasts on youtube scouring the earth for CRT monitors.

[pdimitar]

For the reasons stated above -- that DDR5-gen hardware is expensive right now -- I'd think the DDR4-gen market will remain alive for quite a big longer. Though that's likely much more on the second-hand market side of things.

While I wouldn't necessarily agree with "a handful of people", the fact is that neither of us can prove their lean -- so no point pursuing that argument thread.

So you might be right that it's a pure numbers/statistics decision. Or I might be right that they want to herd people into the more expensive hardware while forcing them to do so by phasing out production of the cheaper hardware.

No way to truly know IMO. We are exchanging hypotheses.

[cosmic_cheese]

I’m not sure that comparison makes much sense. By the time CRTs were phased out, demand was down to almost nothing and what little existed was confined to the extreme budget market. While I don’t have industry insights or anything I don’t think demand for DDR4 is anywhere near the bottom yet, and the remaining demand is centered on premium product (nobody running cheap DDR4 is upgrading). In a more normal market would be more than enough to justify continued production for several more years.

DDR4 production is likely still quite profitable, just not drowning-in-money AI-bubble profitable. If smaller foundries existed they’d be happy to take up the business.

Maybe really what needs to happen is some busting up of the giants…

[ksec]

There is a minimum amount of production volume for it to fit the price equation. If the market doesn't have that demand, it is fundamentally no different to CRT.

Otherwise they could continue to make DDR4 at a higher cost and sold at a higher price to which people will complain price fixing again.

[15155]

> objectively don't use as much anymore

Consumers use these every single day in embedded devices without knowing it.

I wouldn't be surprised in the slightest if the embedded DDR3/DDR4 market greatly exceeds the number of consumer desktop computing devices in terms of "devices with memory" (not in sheer IC count or nominal size though.)

The level of design effort and PCB expense to go from DDR3 to DDR5 is enormous.

[nubinetwork]

I was hoping it was a typo, and they actually meant 4 and 5... I didn't even know people made products that still use ddr3.

[Neywiny]

Especially industrially, ddr3l is just fine. At a certain point you don't need the speed of newer generations, and buying new idk lpddr5x/t controllers for 30 year old process nodes just isn't worth it. Until ddr3/l to EOL from everybody. The big 3 aren't the only memory fabs.

[cududa]

Embedded devices absolutely need DDR3 and DDR4

[bogwog]

It's easy to see why your argument is wrong with a simple hypothetical: what if they were still making DDR4 today? Would people still buy it?

The answer is an obvious "fuck yeah", even if you ignore the DDR5 price gouging. People will buy it because people still have DDR4 hardware, and that hardware is still extremely relevant.

So if there's a market for it, but none of the suppliers are trying to sell to it... Wtf is happening? Basic capitalism logic says any rational supplier would sell DDR4 for easy profits, meeting an unmet demand. That it isn't happen points to some kind of collusion, IMO.

[revolvingthrow]

Because the market pays less for DDR4 than for HBM (or DDR5), and since HBM is heavily modified, vertically stacked DRAM, it competes for the same raw inputs and fab space than DDR4 used.

If I can produce DDR4 for modest profit or HBM for a lot more profit I will obviously produce HBM. And given physical realities producing HBM takes from existing DDR4 production capacity. Worse still, it takes roughly 3GB of ram to produce 1GB of hbm iirc.

[dcrazy]

> People will buy it because people still have DDR4 hardware

The question is whether there’s enough meaningful demand for aftermarket DDR4 upgrades to make it worthwhile to a manufacturer to keep producing DDR4 instead of switching to HBM and DDR5.

Micron claimed retail is a rounding error, a market not worth serving. So you’d presumably need to find industrial buyers who would be willing to buy DDR4.

[microgpt]

Basic capitalism logic is that if you think it's stupid, you put your money where your mouth is, set up a DRAM fab, and get rich.

[cassianoleal]

I'll do it. Will you give me the seed money?

[microgpt]

No, basic capitalism logic is that you already have enough money because barriers to competition are low.

[cassianoleal]

lol :D