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.
Apple Reportedly Worked on Two 'Extreme' Chips
- Thread starter MacRumors
- Start date
- Sort by reaction score
Not a full computer, but 11 years ago the Nvidia GTX 970 had a big controversy around its memory architecture because the last 512MB of the 4GB VRAM was much slower.
"It Won't Happen Again:" NVIDIA CEO Breaks Silence on GTX 970 Controversy
In the wake of bad PR, and a potentially expensive class-action lawsuit over the GeForce GTX 970 memory controversy, NVIDIA CEO Jen-Hsun Huang wrote a candid letter addressed to everyone concerned, explaining in the simplest possible language what went wrong with designing and marketing the chip,...
But that said, it's not like we don't already have tiered memory. It's just how it's used that matters. All CPU's have multiple levels of cache, then memory, then storage. Swap is not all that different from extending the RAM with some slower memory except for the fact that it will still be much much faster than swap, and won't burn writes on the SSD. Seems like a win win. Sure there would need to be some software work done to manage it. But it's not like that's impossible.
Have you never worked at a large company in the tech industry? Lots of projects get shelved before they hit the public. Yes, for the engineers it can be psychologically demoralizing, especially when they wholeheartedly believe in what they are doing. Being moved to an undesirable project is the signal that they are being nice and letting you draw a paycheck while you look for employment elsewhere. [Read: being nice often means they don't want to have to report a layoff to the government.]
Boil
macrumors 68040
Introducing the all-new Mac Pro Cube...
- Debuting at WWDC 2028
- M8 Quadra - TSMC A14/1.4nm process
- 64-core CPU - 48P/16E
- 320-core GPU
- 128-core Neural Engine
- 3TB LPDDR6 RAM
- 64TB RAID - Eight 8TB NAND blades
- 1.4kW PSU
- 8" Cube
Happy_John
macrumors 68000
This is the most sensible interpretation so far.
That already exists, the base iPad just needs to be rebranded. 🙂
MikeTheTiger
macrumors newbie
Throw that M3 Extreme into an Apple tv like form factor and give us a dedicated macOS gaming console. 😤
vegetassj4
macrumors 68020
DrNevs
macrumors member
Yes…sensible, even though two steps down, ‘Max’ continues to loose even more of it’s literal meaning.
Or is it Max as in short Maxwell, Maximilian. Are we just giving them regular names now? How about Fred?
Apple decided to take a different approach, instead of Extreme chips, they went for Extreme pricing.
orbital~debris
macrumors 68030
Love it!
And on a similar note I'd like to see iPhone Duo for the upcoming foldable.
Kashifnawaz85
macrumors member
In my opinion, Apple should continue developing the x86 version of macOS and sell it as a separate licensed operating system so that anyone with Intel or AMD hardware can install it.
They "kicked" those people, the ones left, anyway, to Thunderbolt 5 and externals.
topmounter
macrumors 68030
The people who are fine with Thunderbolt 5 were already off the Mac Pro.
PCIe could be more than 4x the bandwidth of Thunderbolt 5.
Anyone needing more than Thunderbolt offers are being kicked off the platform.
Antony Newman
macrumors member
With Apple making considerable (25-40%) annual speed improvements in highly desirable areas - purchase of an Ultra or Extreme max not be enough of jump in speeds; Especially if they are delivered two to three quarters after the Max is available in a laptop.
If the AI SoC jump in performance is mostly in the M7, and the M8 is potentially arriving on TSMC A14, and 1TB memory prices expected to be under $4000 by the end of 2028, and Apple adopting vapour cooling for desktops - we could have a very potent M8 (chiplet) Extreme - followed by several years of SoC performance stagnation (where most gains are actually come from software iteration).
It would then be a good time to buy an M8 Extreme .. but by then, would 99.5% people think an M8 Max was already insanely powerful for their needs?
If the AI SoC jump in performance is mostly in the M7, and the M8 is potentially arriving on TSMC A14, and 1TB memory prices expected to be under $4000 by the end of 2028, and Apple adopting vapour cooling for desktops - we could have a very potent M8 (chiplet) Extreme - followed by several years of SoC performance stagnation (where most gains are actually come from software iteration).
It would then be a good time to buy an M8 Extreme .. but by then, would 99.5% people think an M8 Max was already insanely powerful for their needs?
IvyKing
macrumors 6502a
The CDC 6400 at Cal had 64K words in standard core storage and IIRC 512K words in extended core storage. I'm not sure if this was the same thing as the "tiered performance RAM" you were referring to. OTOH, the relatively large register set (compared to other computers of the 1960's) of the CDC 6000 machines was to reduce the need for memory access.
Apple needs to come up with better terminology for these chips and more broadly its products. Is 'Ultra' considered better than 'Pro'? Is 'Extreme' better than 'Ultra'? What about 'Max'--does that apply to Apple Silicon chips or only to iPhones? I literally have no idea as these terms are arbitrary and incomparable.
This report is a really interesting tidbit.
To me it shows that Apple really shot themselves in the foot with their own silicon strategy. Not really thinking it through, by the looks of it. Or making assumptions that turned out to be simply wrong.
By forcing the RAM and SSD capacities to no longer be expandable but now built-in via pre-configured options - which the user must decide at purchase time - they gained speed and unified memory capabilities - features which are especially interesting to high-end users - yet also at the same time shuts those high-end users off as these are also typically users who want RAM and SSD to be fully configurable and expandable.
Apple probably thought that this would not be a real issue.
But the fact that the "Extreme" chip and Mac Pro were both cancelled shows that it was actually a real issue.
The users who would most benefit from the "Extreme" chip and a Mac Pro were limited with RAM and SSD configurations. The latter could be addressed but the former could not. Which further narrowed the potential user base so much so that the whole platform was no longer viable.
I fault Apple for not realizing this sooner.
So now Apple is stuck in this weird place. They have an Apple silicon platform that offers amazing performance per watt, which would be ideal for professional uses - yet by pre-configuring RAM and SSD they cut exactly most of those professional users out.
The real irony is that most consumers probably are OK with an iPhone chip. They don't even need an "M" class chip.
And for everyone else the Apple silicon is overpowered - as they cannot fully make use of it due to enforced RAM and SSD limitations.
And the final irony is that this whole pre-configured RAM / SSD disaster was probably built on the assumption that memory chips will get cheaper and cheaper over time, so that pre-configurations become a non-issue eventually.
Yet Apple in its thrive to optimize their supply chains to the max, squeezed the prices for memory producers so much, that these could no longer earn enough to invest in future expansions - making the memory shortage disaster we are currently in only worse - and most likely even causing it to some degree.
Apple is actually in a really dire situation now.
Their Apple silicon strategy with everything built-in and pre-configured is the worst strategy at a time when memory prices are high. If users could just buy the bare minimum RAM and SSD configs today - and expand them later when memory prices came down - that would resolve the whole issue for most people.
And on top Apple now has this high-performing silicon that is perfect for high-end users, yet these are the users who typically really want full expandability on their hardware, often to ludicrous levels, something Apple silicon simply does not offer. So it's not actually for them.
So for whom is the current Apple silicion really?
All those consumers who are OK with iPhone chips in their Macs?
I am very curious to see what John Ternus is going to do about this!
The current Apple silicon strategy is a bit of a dead end or a corner Apple painted themselves into, in my opinion.
It's great for iPhones but not really suited to computing, let alone high-end computing.
To me it shows that Apple really shot themselves in the foot with their own silicon strategy. Not really thinking it through, by the looks of it. Or making assumptions that turned out to be simply wrong.
By forcing the RAM and SSD capacities to no longer be expandable but now built-in via pre-configured options - which the user must decide at purchase time - they gained speed and unified memory capabilities - features which are especially interesting to high-end users - yet also at the same time shuts those high-end users off as these are also typically users who want RAM and SSD to be fully configurable and expandable.
Apple probably thought that this would not be a real issue.
But the fact that the "Extreme" chip and Mac Pro were both cancelled shows that it was actually a real issue.
The users who would most benefit from the "Extreme" chip and a Mac Pro were limited with RAM and SSD configurations. The latter could be addressed but the former could not. Which further narrowed the potential user base so much so that the whole platform was no longer viable.
I fault Apple for not realizing this sooner.
So now Apple is stuck in this weird place. They have an Apple silicon platform that offers amazing performance per watt, which would be ideal for professional uses - yet by pre-configuring RAM and SSD they cut exactly most of those professional users out.
The real irony is that most consumers probably are OK with an iPhone chip. They don't even need an "M" class chip.
And for everyone else the Apple silicon is overpowered - as they cannot fully make use of it due to enforced RAM and SSD limitations.
And the final irony is that this whole pre-configured RAM / SSD disaster was probably built on the assumption that memory chips will get cheaper and cheaper over time, so that pre-configurations become a non-issue eventually.
Yet Apple in its thrive to optimize their supply chains to the max, squeezed the prices for memory producers so much, that these could no longer earn enough to invest in future expansions - making the memory shortage disaster we are currently in only worse - and most likely even causing it to some degree.
Apple is actually in a really dire situation now.
Their Apple silicon strategy with everything built-in and pre-configured is the worst strategy at a time when memory prices are high. If users could just buy the bare minimum RAM and SSD configs today - and expand them later when memory prices came down - that would resolve the whole issue for most people.
And on top Apple now has this high-performing silicon that is perfect for high-end users, yet these are the users who typically really want full expandability on their hardware, often to ludicrous levels, something Apple silicon simply does not offer. So it's not actually for them.
So for whom is the current Apple silicion really?
All those consumers who are OK with iPhone chips in their Macs?
I am very curious to see what John Ternus is going to do about this!
The current Apple silicon strategy is a bit of a dead end or a corner Apple painted themselves into, in my opinion.
It's great for iPhones but not really suited to computing, let alone high-end computing.
Last edited:
All marketing is arbitrary and mostly incomparable... That is what marketing does best.
Apple is not prioritizing "what makes sense", they're prioritizing "what will make us more money".
Could their branding be better? Yes, I personally agree. But I can't say that I could make them more money.
Does it really matter what they are called?
If one is going to spend >$1000 on an item it’s probably a good idea to do the research to figure out what you need and what is the best value for your budget
