Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
But now we're going in circles. Apple's cost isn't an additional $400. Maybe an additional $40.

They priced and configured it the way because they wanted to. They could've also priced it the same and configured it with higher specs.



I think they were aware that they were stretching that spec. By the time the M3 shipped, it had gotten quite low.



I'm not personally outraged. Overall, the ARM Macs have been good products. Industry-leading in several respects — SoC, trackpad, screen (in the case of the Pro), speakers — and decent in others. Configured a little higher, they're easy to recommend to almost anyone unless you a) play games or b) use a lot of Windows apps.
It isn’t about Apple’s component cost, it’s about the perceived added value of the product on offer. Most products are not priced purely based on material costs. Prices are balanced for maximum profitability, what customers are willing to pay for said product, etc. Even many low-end budget laptops that use Antique RAM cards charge $100 or more for RAM upgrades. Because it isn’t about the component cost, but the added value for customers…

And they could also decide to price it as a loss-leader and make zero or very little profit from it at all… Just because a business could do something (which is technically an assumption since we don’t really know what profits Apple makes on these Macs to begin with, though I’m guessing there’s reasonable margin) doesn’t mean they should do it. Several competitors are charging similar or even higher prices for their upgrades…

Again, that’s merely your opinion. And many people are still using 8GB M1 Macs to this day without issue, so I don’t think it’s “stretching” a spec just because some need more. Is Apple “stretching” 16GB because some need 32GB? Or what about those who need 128GB? The existence of some who want or need more RAM doesn’t mean the base spec is being “stretched”.

If you’re starting premise is “everyone needs 16GB” (in your case, I think you believe everyone needs at least 12GB) (while ignoring many who are still using 8GB Macs perfectly fine, and the high sales of those models), then of course when the base spec is changed to 16GB you’ll find it easier to recommend. What I don’t get though, is why the base spec being lower automatically makes the entire lineup somehow not worth recommending. I don’t know, maybe people could always just buy the RAM spec they wanted or thought they would need… And that would still be worth recommending… Like, if someone needs a 32GB or 64GB computer, the entire MacBook lineup is suddenly somehow not worth recommending anymore because the base spec isn’t 32GB or 64GB?
 
People repeat that a lot, but it's not true in any meaningful or relevant way.
And that’s just your opinion. It’s a fact that indirect bus connections introduce additional latency, and reduced efficiency. They also increase the risks of issues due to a loose or faulty connection. There are good reasons why computer manufacturers are moving to direct soldered RAM, it improves performance and efficiency, which more customers care about than cracking open their computer and risking destroying it to swap fragile parts…
 
And how much additional latency would that be?
Let’s use Intel’s Lunar Lake chips as an example here. By moving to on-package memory, the Lunar Lake chip’s memory bandwidth moved from 119.5 GB/s on the prior gen chip to 136.5 GB/s, a significant performance increase. And the Lunar Lake chip sees a 40% reduction in power consumption compared to off-package memory solutions. Both are major improvements. And in Apple Silicon, you have the added advantage of ARM vs X86, etc.

Physics exists. More distance and resistance from leads, and an indirect bus connection equals higher latency and lower efficiency. That’s just a fact.
 
Let’s use Intel’s Lunar Lake chips as an example here. By moving to on-package memory, the Lunar Lake chip’s memory bandwidth moved from 119.5 GB/s on the prior gen chip to 136.5 GB/s, a significant performance increase. And the Lunar Lake chip sees a 40% reduction in power consumption compared to off-package memory solutions. Both are major improvements. And in Apple Silicon, you have the added advantage of ARM vs X86, etc.
I was asking about your claim regarding memory latency. You answered with memory bandwidth, power consumption, and even processor architecture. Back to memory latency. I still have not seen any evidence that soldered RAM makes a meaningful or relevant difference. And I have not just been looking in this thread.
 
I was asking about your claim regarding memory latency. You answered with memory bandwidth, power consumption, and even processor architecture. Back to memory latency. I still have not seen any evidence that soldered RAM makes a meaningful or relevant difference. And I have not just been looking in this thread.
By moving RAM onto the SoC, Lunar Lake achieves higher bandwidth memory. That’s because there’s less distance, aka latency between the RAM and the CPU. Any time you work with signals, the longer the wire you transmit your signal through, the higher the latency. This is pretty basic stuff. Seemingly “insignificant” or “small” differences in this distance can absolutely have a significant impact on latency/bandwidth.

Basically every major tech publication agrees that on-package RAM is faster and more energy efficient. Even PC publications that aren’t friendly to Apple…. And this is why many PC manufacturers are moving to try to either get RAM as close to the CPU as possible with a soldered connection, or use on-package RAM. It has benefits in terms of performance and efficiency. Do you think the major tech publications and many PC manufacturers are just making these improvements up? It’s just not really a tenable position, because physics demonstrates why closer RAM will perform better…
 
But they won't do it again in Lunar Lake's successor.
And that was specifically because of the cost of production and the impact on profitability. They didn’t expect to ship as many of them as they ended up shipping…

There are obvious performance and efficiency improvements, major tech publications agree on this, and so do PC manufacturers. And physics backs it up…
 

Attachments

  • IMG_6048.jpeg
    IMG_6048.jpeg
    116.5 KB · Views: 16
  • IMG_6059.png
    IMG_6059.png
    507.7 KB · Views: 13
By moving RAM onto the SoC, Lunar Lake achieves higher bandwidth memory. That’s because there’s less distance, aka latency between the RAM and the CPU.
Lunar Lake uses 8533MHz RAM, a speed that is exceeded by the fastest pluggable DIMMs, albeit not in LP, but I have no problems believing that soldered RAM can reduce power consumption.
 
Lunar Lake uses 8533MHz RAM, a speed that is exceeded by the fastest pluggable DIMMs, albeit not in LP, but I have no problems believing that soldered RAM can reduce power consumption.
The specific RAM spec used on Lunar Lake chips may have been lower than some newer RAM chips, but that doesn’t prove that on-package RAM doesn’t provide better performance than off-package RAM… On-package RAM performs better according to major tech publications, PC manufacturers, and physics backs it up…
 
…but I have no problems believing that soldered RAM can reduce power consumption.
And even if you don’t believe multiple major tech publications (including ones not friendly to Apple), device manufacturers, etc. about the performance improvements that on-package RAM offers, even just the improved energy efficiency alone is more useful for many more consumers than the ability to crack open their computer and risk destroying it by trying to swap fragile components…
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.