Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Definitely expecting to see the usual yearly performance and speed upgrades. Better thermal management will be good. Think the 18 Pro will cost atleast $100 more.
On its own, I can deal with a $100 price increase for the 18 Pro/Max over the 17 Pro/Max. But if I wait for the new phones, the trade-in value on my 16 Plus is likely to drop by $200. That’s a $300 delta. Still trying to decide whether to buy now or wait.
 
Currently my 17PM stutters on the screen a lot, and it's meant to be "the most power phone apple have ever made", or something.
On my 17PM I get very noticeable stutter swiping between home screens. Haven't noticed this stutter in any of my apps, though.
 
If Apple can't secure enough RAM for the A20, these new iPhones will be severely supply-constrained. CXMT can't rescue Apple here as it doesn't make LPDDR6 (yet). Also CXMT's DRAM capacity is all bought up by domestic customers. Plus I read CXMT is only slightly cheaper than the big-3 (Samsung/SK Hynix/Micron). There's no way Apple can have its own DRAM source as the patent base/walls prevent a new competitor from emerging. There will be only Samsung/SK Hynix/Micro/CXMT.
Solution: make it very expensive. Thank me later.

- Timmy
 
  • Like
Reactions: turbineseaplane
On its own, I can deal with a $100 price increase for the 18 Pro/Max over the 17 Pro/Max. But if I wait for the new phones, the trade-in value on my 16 Plus is likely to drop by $200. That’s a $300 delta. Still trying to decide whether to buy now or wait.

That's exactly why I decided to trade in now and just be done with it. Any price increase will also be reflected on the lower trade-in value of existing devices. Also, I'd say that a 100 dollars price increase is very optimistic, but even if it is that low, it would probably convert to 200 euros so...
Trigger pulled, picking phone up tomorrow. If anything, this gives peace of mind. I was not at all comfortable with how sudden and steep Apple's price hikes were. They performed horribly IMO.
 
No one cares about performance, market doesn’t have any high/end heavy applications…

AI posers, cache our will be insane by investors
 
So with this new WMCM technology, doesn't moving the RAM to the side then increase latency compared to when it was put on top the CPU to decrease latency?
I wondered about that too, but the leaks and rumors also refer to performance increases in other areas of the RAM circuitry, so maybe those will make up for any increased latency from moving the DRAM to sit alongside the CPU. This side-by-side design is intended to improve heat dissipation, so that too will likely improve speed, or at least reduce instances of thermal-induced speed throttling.
 
  • Like
Reactions: DEMinSoCAL
So with this new WMCM technology, doesn't moving the RAM to the side then increase latency compared to when it was put on top the CPU to decrease latency?

I wondered about that too, but the leaks and rumors also refer to performance increases in other areas of the RAM circuitry, so maybe those will make up for any increased latency from moving the DRAM to sit alongside the CPU. This side-by-side design is intended to improve heat dissipation, so that too will likely improve speed, or at least reduce instances of thermal-induced speed throttling.

As I understand it the most recent A-Series SoCs have a three-level cache architecture with L1 & L2 caches local to each processor core and then a very large SLC (System Level Cache) that sits between the L2 caches and the main DRAM where the GPUs and NPUs also pull from the SLC.

If I've got that right (someone please correct me if not, ideally with a link to more accurate info) then if the DRAM is just off the edge or the main die where the SLC is located, i.e. at minimum horizontal distance between DRAM and SLC, then maybe signal path lengths wouldn't be too badly affected if at all vs the current stacked setup?
 
  • Like
Reactions: johnsawyercjs
If Apple can't secure enough RAM for the A20, these new iPhones will be severely supply-constrained. CXMT can't rescue Apple here as it doesn't make LPDDR6 (yet). Also CXMT's DRAM capacity is all bought up by domestic customers. Plus I read CXMT is only slightly cheaper than the big-3 (Samsung/SK Hynix/Micron). There's no way Apple can have its own DRAM source as the patent base/walls prevent a new competitor from emerging. There will be only Samsung/SK Hynix/Micro/CXMT.

You seem to know about this stuff. Are you able to provide any more informed insight about something I've been wondering for a while now?

With the huge demand for memory to build out AI data centres resulting in supply getting overloaded by such high demand thus pushing up prices I've been wondering whether, while you mention patent walls preventing new players like Apple building its own fabs, do some or all of the big players like Samsung/Hynix/Micron have any plans in place to very aggressively ramp up production (presumably needing very expensive new fabs to be built?) such that this AI-related price increase for memory might be a bubble and when lots of new capacity comes on line prices will drop significantly, or dare we dream even fall if the memory manufacturers misjudge future demand and lots of new capacity comes on line just as demand begins to fall from the crazy current situation?
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.