Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
LPDDR6: New Beginning of AS: 12GB Standard
- Thread starter TigeRick
- Start date
- Sort by reaction score
Let me restate my crazy paranoid, “Apple is evil”, theory:
Baseline M4 will be released with 8GB of accessible RAM but with the next major OS release you can pay a fee the will enable the full 12GB amount of already installed RAM. Where the purpose of custom micron 6GB chips is to enable restricting part of the RAM behind a paywall.
I think this is ludicrous, but sometimes it is fun putting forward a wild crazy rumor. Just for the fun of throwing gasoline's on a fire. [maybe I can get a paid job at Bloomberg too 🤪]
So how exactly is this evil?
This sort of practice has been common for decades in larger system sales- though typically with processor cores, not RAM. You buy a POWER system X from IBM, it has 16 enabled cores, and when you need more power you pay them an additional fee to enable another 8 cores that are already there. Lots of other vendors do/did the same thing.
There is no need for custom RAM to enable this. All it takes is code in the OS.
All that said, I strongly doubt Apple would do that. It does not seem like customers would appreciate it (just look at this forum).
1. I don’t understand why LPDDR6 was ever part of the discussion. It just doesn’t make sense timing wise. LPDDR5X makes much more sense.
2. I still don’t know what is the exact type of RAM that’s in the iPad Pros. Many claim it is LPDDR5X but others claim it is custom LPDDR5. LPDDR5X makes more sense to me but Geekerwan is smarter than me and claims it is custom LPDDR5.
The big problem with the MacRumors article though was that info about the Mac Studio and Mac Pro was not new. As mentioned, it was talked about it the previous month, but I guess MacRumors thought it would generate additional clicks if they rehashed the article since the M4 iPad Pro had been released in the interim.
OK, not exactly evil, but it would just p!ss everyone off. These are consumer and prosumer machines, not multimillion dollar institutional level purchases.
2. I still don’t know what is the exact type of RAM that’s in the iPad Pros. Many claim it is LPDDR5X but others claim it is custom LPDDR5. LPDDR5X makes more sense to me but Geekerwan is smarter than me and claims it is custom LPDDR5.
That is not correct. In that new newsletter, Gurman specifically referenced his April newsletter, which talked about the Mac Studio and Mac Pro.
The big problem with the MacRumors article though was that info about the Mac Studio and Mac Pro was not new. As mentioned, it was talked about it the previous month, but I guess MacRumors thought it would generate additional clicks if they rehashed the article since the M4 iPad Pro had been released in the interim.
It's evil.
OK, not exactly evil, but it would just p!ss everyone off. These are consumer and prosumer machines, not multimillion dollar institutional level purchases.
Last edited:
I think a better analogy was BMW attempt to make heated seats in a new car a subscription service. They were installed in all cars but you had to pay extra if you wanted to use them. The resulting consumer rage, forced BMW to reverse that policy. During that debate BMW was called many things including Evil.
Tesla does this with batteries and other features. On at least one of their models the regular (60 kWh) and long-range (90 kWh) models come with the same 90 kWh battery, and which you get to use depends on how much you pay. If, after purchasing the former, you want to upgrade to the latter, you need to pay Tesla to remove the software lock.
And Apple may already do this with CPU cores. It's possible some of their, say, 7-core devices are 8-core chips with one core deliberately inactivated. They would need to do this if the number of 7-core chips produced due to defects isn't enough to meet sales requirements.
https://i.redd.it/lvljg4edio9b1.png
www.reddit.com
It's evil.
I agree with both of you, customers would not likely appreciate this. I wouldn't like it either. But it's not evil. It's marketing and strategy - a bad strategy, I think, but not an evil one.
That's definitely something they're already doing. Variable defect rates over the lifetime of a chip are enough to ensure that. Beyond that, some manufacturers will bin chips by performance/power, where it's not a disabling defect that gets a core disabled, but simply poor (too high) power usage. It's entirely plausible Apple's doing that too.
As expected, there are no Mac hardware in the WWDC event. TLDR, my starting question still remains as below:
Question: Have you wonder why Apple rumored to launch M4+ series by the end of the year? This round Apple will launch M4 from top to bottom including Mac Mini within a quarter period, why?
Mark Gurman can't tell you, but I can. The key ingredient missing is LPDDR6.
According to MC and past lineup, we should be expecting Mac lineup in the following order:
- Q4-2024: MacBook Pro with M4+, M4 Pro and Max. iMac should come together.
- Q1-2025: MacBook Air 13+15 with M4+, Mac Mini with M4+ and M4 Pro.
- WWDC 2025: Mac Studio with M4 Max/Ultra, Mac Pro with M4 Ultra.
Last edited:
None of the M4 generation SoC will feature LPDDR6. Probably M5 or later.
Agreed. Probably the best one should hope for is LPDDR5X (although for whatever reason, Geekerwan claims the M4 iPad Pro uses overclocked LPDDR5).
Yes, we will have to get confirmation somehow whether the iPad Pro based on the M4 uses LPDDR5 or LPDDR5X.
The part number isn't publicly available unfortunately.
As far as I know the iPad Pros have been using the same RAM as the Mac
Therefore if we can confirm that we can also see what’s coming ahead
Indeed, and yet Mac lineup are completely MIA in WWDC. If you watched the keynote, Apple has at least two chances to mention M4 SoC, yet Apple has not spoken once about M4. Apple rather talked about iPad Air which comes with two display sizes and won't talk about iPad Pro with the most advanced SoC fabbed by most advanced process. Why?
I hope everyone should really think about it...
Last edited:
Are you for real?
WWDC's primary purpose is communicating with software developers to help them write software for Apple's platforms. It's "World Wilde Developer Conference", not "World Wide Hardware Announcement Conference".
Yes, sometimes they'll include hardware announcements, particularly when those announcements have a major impact on software devs. An example of that is when they used WWDC to announce their intent to transition the Mac to Apple Silicon. But they didn't even announce the M1 chip or specific Apple Silicon Mac models at that WWDC, despite the fact that the first M1 Macs shipped in the fall of that same year.
The flip side of this is that when Apple doesn't talk about hardware in the WWDC keynote, it doesn't mean anything about their hardware plans. What new things does M4 bring to the table for software devs? Nothing aside from SME, really, and they already have all the information they'd need to develop for SME, and the audience of people who want to use SME directly is pretty small.
So there's no reason for a big focus on M4 at WWDC. It means nothing for your silly and unfocused theories - it does not confirm or deny them.
We'll hear more about later in the year most likely in Oct/Nov/Dec..
What do people think about CAMM2? Anyone think Apple will adopt it?
As far as I understand it, it's supposed to have lower latency since it's closer to the processor.
This is perhaps already solved with Apple's current memory placement on their SoC? And since Apple has this whole.. "unified" memory, maybe external memory like CAMM2 wouldn't match.
Would be nice though since they are replaceable/upgradable.
As far as I understand it, it's supposed to have lower latency since it's closer to the processor.
This is perhaps already solved with Apple's current memory placement on their SoC? And since Apple has this whole.. "unified" memory, maybe external memory like CAMM2 wouldn't match.
Would be nice though since they are replaceable/upgradable.
Very unlikely, it does not offer anything they are interested in.
That is unfortunately a myth. Apple's memory does not have lower latency either. The main advantage of CAMM2 is very compact form factor, and of course, the fact that it brings modularity to RAM that has traditionally been non-modular.
Apple's unified memory is pretty much the same RAM as what CAMM2 uses. It is possible that Apple's solution is more energy-efficient (I do not understand it myself as I lack the background in electrical engineering, but there have been claims that Apple can drive the modules using lower voltages by utilising higher pin count).
As @leman alluded to it isn't latency that drives soldered LPDDR and (lp)CAMM2 but power, size, and bandwidth. While (lp)CAMM2 gets closer to the soldered on-package memory Apple uses, it can't (yet) offer the kind of solution Apple needs. At the moment to fill out a 128-bit bus you need 32GB of LPCAMM2 but I'm having trouble finding if smaller lpCAMM2 modules (while still being dual channel) are planned for the current iteration. Further, for things like the M3 Max you'd need multiple of them and, beyond the GB, they are quite large. However, as minimum RAM capacities offered by Apple increases and the CAMM technology progresses in future versions, it may be possible. But Apple may choose to go a different path ... and likely will. With the R1, they experimented with on-die memory whether that goes anywhere remains to be seen.
LPDDR6's JEDEC Specs Leaks
There was PPT leaks from Synopsys. There are 3 speed grades planned and I have calculated the memory bandwidth with different memory bus below:
There was PPT leaks from Synopsys. There are 3 speed grades planned and I have calculated the memory bandwidth with different memory bus below:
| LPDDR6 Speed | 24-Bit Base | 72-Bit for iPhone 16 Pro Series ? | 96-Bit for M4+ | 192-Bit for M4 Pro |
|---|---|---|---|---|
| 10.667 Gbps | 28.4 GBps | 85 GB/s | 114 GB/s | 228 GB/s |
| 12 Gbps ? | 32 GBps | 96 GB/s | 128 GB/s | 256 GB/s |
| 14.4 Gbps | 38.4 GBps | 115 GB/s | 154 GB/s | 307 GB/s |
- After non showing of any M4 in WWDC, I am pretty sure Apple is going all in with LPDDR6 in upcoming M4+ series by the end of the year; now supported by Ross Young. Clearly, @Confused-User is confused now cause he predicted Apple will launch Mac Studio before end of the year is delusional.
- M4 for iPad is outliner, as I speculated Apple will split the M4 SoC into two: M4 for iPad and M4+ for Mac. M4 is using 8GB 128-bit LPDDR5 providing 120GB/s memory bandwidth meanwhile M4+ will utilize 96-bit memory bus as I speculated before the leaks. There are 3 speed grades planned, 10.667 Gbps is too low for M4+, thus Apple will most likely employ 12 Gbps or 14.4 Gbps. I tend to believe Apple will use the fastest one, we shall see.
- Unlike MC's leaks about Mac, there are no leaks supporting my theory about 12GB LPDDR6 in upcoming iPhone 16 Pro. There are actually few reasons why I believe Apple will employ LPDDR6 in iPhone 16 Pro series. TLDR, let's see how the outcome...
N3B vs N3E: Why M4+ Series Will Use N3B Process not N3E
M3 (N3B) : 146 mm2
M4 (N3E) : 165 - 170 mm2
M4+ (N3B): ?
M4 Pro (N3B): ?
M4 Max (N3B): ?
Another speculation from mine is upcoming M4+ series will continue using N3B process. You can see how big the die size of M4 compared to M3. Sure, M4 has more cores and NPU, but can you imagine how big the die size of M4 Max with all the extra cores, NPU and ultra-fusion connectors? TLDR, I actually explained why N3B process is making more sense for Apple if you guys really want to know go check the difference between single vs multiple patterning of process. N3B still has untaped potential which I believe Apple will use to the fullness in upcoming M4 Max.
As I questioned before, why do you think Apple would launch M4 for Mac half a year later? Cause Apple is waiting for LPDDR6 with most features available by N3B process.
M3 (N3B) : 146 mm2
M4 (N3E) : 165 - 170 mm2
M4+ (N3B): ?
M4 Pro (N3B): ?
M4 Max (N3B): ?
Another speculation from mine is upcoming M4+ series will continue using N3B process. You can see how big the die size of M4 compared to M3. Sure, M4 has more cores and NPU, but can you imagine how big the die size of M4 Max with all the extra cores, NPU and ultra-fusion connectors? TLDR, I actually explained why N3B process is making more sense for Apple if you guys really want to know go check the difference between single vs multiple patterning of process. N3B still has untaped potential which I believe Apple will use to the fullness in upcoming M4 Max.
As I questioned before, why do you think Apple would launch M4 for Mac half a year later? Cause Apple is waiting for LPDDR6 with most features available by N3B process.