Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Wow, what a straw man argument. He never said nobody knows how to make LPDDR like apple’s. He said M1 is a CPU, and nobody knows how to make the M1 other than TSMC. He knows the memory is on separate modules. He pointed out that you stated that “many people make these,” referring to the *CPU*

You ignored the gist of my comment. Now that Apple (TSMC) has made it, everyone can pull it apart and see exactly how it as made. Licensing and IP is the biggest thing in the way now, not knowledge. Samsung can't knock off a design? When was this true? Samsung can't do 3nm fabrication? What about next year, or the year after?

Continuous metal helps latency because it conducts and has a low impedance. A slot, clamp, or other type of connection has higher impedance which increases latency. This is simple electrical engineering. Not to mention work function effects. That’s advanced electrical engineering. You keep ignoring the laws of physics.

Do you know the difference between a static DC voltage - a voltage which is *required* to vary very slowly (so that, for example, parasitic capacitance is *good*), and a high frequency data signal, which required to vary very quickly (many millions of times a second)?

I guess you don’t.

Relax, it was an example of electrical connection without solder, it wasn't meant to be a blueprint for CPU manufacturing.

Yet "high frequency data" connections exist in the electronic industry that aren't soldered, corroded, current limited or whatever forum guy electonic engineering conjecture can be dropped to support the limited marketing Apple provides as to their exact specs vs. available manufacturing/design options.

I also noticed you've ingored the hard drive issue. Also you're making a massive assumption with this "unsoldered" memory latency, that it would be a significant system bottleneck, over previous types, rather than the all-in-one approach being a cheaper and easier way for Apple to save on costs, over seperate system and video memory, and the resulting additional cooling reqs.

So when the Apple silicon Mac Pros support faster graphics cards, memory and drive expandibility, while invariably outperfomring this "overclocked/cored" all-in-one, laptop in a box. which will forever be "soldered" into 2022... well 2021 actually, will you still be drinking this SoC / built-in memory kool aid?

But instead of endless arguing about this machine, that I clearly am disappointed in, and have zero interst in buying... if you guys want to keep cheerleading, looks like there's another M1 marketing fire to put out... S0C vs older "unsoldered" graphics cards... https://www.macrumors.com/2022/03/17/m1-ultra-nvidia-rtx-3090-comparison/
Better get over there quick!
 
But instead of endless arguing about this machine, that I clearly am disappointed in, and have zero interst in buying...

If you want a machine with expandable RAM, the Mac Pro does exist.

if you guys want to keep cheerleading, looks like there's another M1 marketing fire to put out... S0C vs older "unsoldered" graphics cards

"Hey everyone, I'm not interested in this product, but also dislike it. Please argue with me!"
 
Last edited:
If you want a machine with expandable RAM, the Mac Pro does exist.
Yes, what I use for my work machine. The biz has several, going back a few generations.
The recent dev history (or lack there of) has been a not un-serious concern, and a very common complaint.

But enjoy... like I said, I'm a shareholder so buy lots of Mac Studios, silver lining, lemons into lemonade and all that.
 
Is the SoC soldered on to the mainboard in the Mac Studio, or does it sit in a socket? And does the RAM sit in the same package? If it is in a socket and you can take out a defective one and replace it, that does rather mitigate the issue of non-replaceable RAM chips, though it would be much more expensive. I've had so many damn RAM failures in the last 30 years, but none ever really mattered until it meant dead laptops thanks to Apple soldering the darn stuff down!

How significant or marginal are the gains from the current RAM over what could be achieved by socketing it? Unless it's a massive improvement, I personally think it is a terrible decision, designed with the hope of killing your computer in the long run and forcing you to buy another.
 
If any government cared one iota about the consumer products their citizens used, or about the environmental impact of e-waste, they'd ban the practice of soldering RAM, CPU's/SoC's, and storage to a board, in any device as large as a modern laptop. RAM, CPU's/SoC's, and storage could and should all be on sockets, and the performance hit (if any) would be negligible. The RAM and CPU/SoC can be soldiered together if genuinely necessary, and their package can sit together on a socket. It's about saving a few pennies for Apple and preventing devices from being user survivable, and it is despicable. The marginal thickness increase, small added cost and requirement for additional RF shielding would be worth paying for society, in my opinion.

If anyone has concrete proof I'm wrong about any of this, please do share it ?
 
Last edited:
As I said, if Apple hadn't "glued these in" another manufacturer could fabricate their own memory to fit into Apples design, which as you stated contain off the shelf components.

Yes, but there are real advantages to performance to soldering the ram on-package with the SoC. Which is why Apple does it. And so does every other OEM that pairs memory with an SoC (like smartphones, tablets, etc.).

The memory modules, using off the shelf parts can't be put onto another module sized for the mechanical dimensions and to the electrical spec? What is stopping others besides Apple licensing? This SoC contains separate fabricated components and you're sure the memory part can only be fabbed by TSMC? How about next year?

Apple does not license RAM. They buy it from other companies and then solder it onto the package that contains the SoC.
 
Yes, but there are real advantages to performance to soldering the ram on-package with the SoC. Which is why Apple does it. And so does every other OEM that pairs memory with an SoC (like smartphones, tablets, etc.).



Apple does not license RAM. They buy it from other companies and then solder it onto the package that contains the SoC.
I think you're missing the other poster's point. Once the SoC and RAM are soldered to the same package, why must that package also be soldered down?
 
Last edited:
I think you're missing the other poster's point. Once the SoC and RAM are soldered to the same package, why is that package glued down??

According to all the iFixit teardowns (M1 MacBook Air, M1 iMac, M1 Pro/Max MacBook Pros) none of the SoCs are glued down. They are soldered to the systemboard and then the heatsink is screwed down on top of them with thermal paste applied between the SoC and the heatsink.
 
I think you're missing the other poster's point. Once the SoC and RAM are soldered to the same package, why is that package glued down??
It‘s not glued, it‘s soldered. It‘s soldered instead of socketed to save space, improve reliability, reduce cost, and improve I/O performance. It also helps assure compliance with FCC (and other international) rules re: radio interference, etc.
 
It‘s not glued, it‘s soldered. It‘s soldered instead of socketed to save space, improve reliability, reduce cost, and improve I/O performance. It also helps assure compliance with FCC (and other international) rules re: radio interference, etc.
Let's go through these:
1, Saving space. Yeah, a tiny fraction. This is the same lousy argument as no longer putting batteries in cases. A Mac Studio doesn't LOOK as thin as an iPhone, so I'll assume you're only taking this into consideration on extremely thin devices.
2, Performance? How much does a socket degrade performance? Over 0.1%? Server farms around the world can't be THAT wrong.
3, Reduce cost. Yes, they'll save a few pennies. Gosh, maybe even a dollar. ?
4, CPU's in sockets clash with rules on radio interference? Care to elaborate on the logic behind a socketed system causing radio interference? (A little Google suggests it makes designs simpler because there's less signal 'leak' from a soldered design, so they can do away with/reduce the shielding they would otherwise have to employ.)
 
Last edited:
According to all the iFixit teardowns (M1 MacBook Air, M1 iMac, M1 Pro/Max MacBook Pros) none of the SoCs are glued down. They are soldered to the systemboard and then the heatsink is screwed down on top of them with thermal paste applied between the SoC and the heatsink.
Ok ok, I was using the word glued because someone else was.

I should have asked: why are they soldered down to the main board rather than socketed, aside to save a few pennies and intentionally reduce item life expectancy? RAM is soldered to the CPU for apparent gains, but beyond that, is it necessary to solder the package down? Thanks.

Edit: As I added to my last post, a little Google suggests it makes designs simpler and cheaper because there's less signal 'leak' from a soldered design, so they can do away with/reduce the shielding they would otherwise have to employ, to prevent a class with radio frequencies. The device becomes fractionally thinner and lighter as a result, and a couple of dollars cheaper. Anyone think this should matter in a $1,999+ desktop? I tried to find any modern or Apple specific details on performance, but failed. Information regarding PC's from a couple of years ago suggested a CPU being socketed or soldered made no difference to performance.
 
Last edited:
I should have asked: why are they soldered down to the main board rather than socketed, aside to save a few pennies and intentionally reduce item life expectancy? RAM is soldered to the CPU for apparent gains, but beyond that, is it necessary to solder the package down? Thanks.

It is not necessary, but there are performance benefits, as noted above.

You also save on production costs not having to create a ball or pin grid array on the underside of the SoC and then a corresponding socket, each containing thousands of extremely fragile connectors that can be damaged during installation.

And honestly, the SoC is probably highly unlikely to suffer a catastrophic failure in normal use. CPUs and GPUs almost never fail on their own - it is some outside influence like a power-surge or extreme overheating that kill them. And when that happens, usually the systemboard goes with it so you'd be replacing everything, anyway.

And Apple does not sell SoCs at retail or under warranty so having it socketed would do nothing for the end-user since they would not be able to buy a replacement SoC, much less an upgrade.
 
  • Like
Reactions: ric22
It is not necessary, but there are performance benefits, as noted above.

You also save on production costs not having to create a ball or pin grid array on the underside of the SoC and then a corresponding socket, each containing thousands of extremely fragile connectors that can be damaged during installation.

And honestly, the SoC is probably highly unlikely to suffer a catastrophic failure in normal use. CPUs and GPUs almost never fail on their own - it is some outside influence like a power-surge or extreme overheating that kill them. And when that happens, usually the systemboard goes with it so you'd be replacing everything, anyway.

And Apple does not sell SoCs at retail or under warranty so having it socketed would do nothing for the end-user since they would not be able to buy a replacement SoC, much less an upgrade.
You're right, I've not heard of CPU's/SoC's frying themselves without overclocking, even after years in servers. It's the RAM element I hate to see soldered down. Either I've been obscenely unlucky with RAM, or it remains the general point of failure inside our devices.

In laptops it used to be cool to be able to upgrade the CPU in a laptop after a couple of years, but sockets change so fast that that isn't really going to be a thing, even if they weren't soldered down.
 
  • Like
Reactions: CWallace
Either I've been obscenely unlucky with RAM, or it remains the general point of failure inside our devices.

Yeah in terms of the thousands of servers I support, it is RAM and disks (HDD and SSD) that fail. We almost never lose a CPU and when we do, it is usually down to a socket failure on the systemboard that also takes out the CPU (go figure).

In laptops it used to be cool to be able to upgrade the CPU in a laptop after a couple of years, but sockets change so fast that that isn't really going to be a thing, even if they weren't soldered down.

Exactly.
 
  • Like
Reactions: ric22
Yeah in terms of the thousands of servers I support, it is RAM and disks (HDD and SSD) that fail. We almost never lose a CPU and when we do, it is usually down to a socket failure on the systemboard that also takes out the CPU (go figure).



Exactly.

as i noted above, sockets are bad for reliability, which is why it’s good to solder things down.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.