Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
You stated it as a definitely fact that those Broadwell chips do not work with DDR3L 1866 MHz.

Why do you keep fudging together what applies to Skylake and select Broadwell chips to ALL Haswell/Broadwell/Skylake ?

I gave you a response based on established testing (Intel's specifications), not based on my sampling of my own equipment - poor data set. It is a definite FACT that Intel did NOT specify those chips to run on 1866 Mhz DDR3L.

You asked for opinions on why Apple decided to use soldered RAM - to which I gave an educated guess on memory speeds and availability based on:

(1) Soldered RAM on most of Apple's lineup began with the Haswell generation (FACT).
(2) Most of Haswell/Broadwell non-desktop chips are not specified to use 1866 Mhz or above DDR3L (FACT based on Intel's published specifications) - only LPDDR3.

Both of those items are correct - which makes my opinion to your question an educated guess.

NOW, I do not know what the reason is for Apple's decision to forgo Intel's specifications for the Skylake iMacs, which appear to be an exception. To this, I replied:

(3) Skylakes appear to be an exception - I do not definitely know why but gave a few plausible ideas.

NOW, up to (2), everything is factual, which makes my original response a sound educated guess.

So you concede that your "facts" might be wrong?

You need go get a statistical grasp on what I said - as PER INTEL, there is no official support for DDR3L at 1866 Mhz or beyond on the chips you implicitly referenced to, which means NOT ALL SYSTEMS ARE GUARANTEED TO WORK THAT WAY. My testing shows that in very specific instances, it can work - BUT this is specific to my setup and usage scenario. I did not make the fatal mistake to assume that what applies to my two computers, is sufficient evidence that it should work for all, despite what Intel specified. This argument needs to be made statistically, not based on limited, anecdotal evidence. I.e., the following does not work:

(A) I test a few setups to verify that DDR3L works at higher frequencies than published in 2 computer systems.
(B) I can assume that Intel is wrong and that whatever works for me, necessarily works for all configurations.

All that we can say is:

(C) It seems to work sometimes in specific scenarios, but probably not for all configurations, that's why Intel specified the chips their way.

Somehow, I don't think you are a computer engineer working at Apple, who can verify this as fact.

Then please, by all means, ask this question directly to the Apple engineers instead of reaching out to a casual Mac forum, if you are this antagonistic toward contributors.
[doublepost=1467365997][/doublepost]For what its worth, I know you support user-repairable computers based on your postings. I do not think the issue is Tim Cook, but rather Jony Ive. The latter appears to be hell bent on making everything thinner and into fashion accessories rather than work machines. I do not personally agree with their direction toward mass planned obsolescence. I find the keyboards on the new Macbook deplorable and the proprietary SSDs a pain to upgrade, financially. I used to have an iMac before the current glued-together machines and I will not own another iMac unless they reverse course on their designs.
 
Last edited:
  • Like
Reactions: Samuelsan2001
I asked why the MacBook Pro Retina have soldered RAM. Some replied that having soldered RAM allow the MacBook Pro Retina to be thinner, never mind that there are laptops as thin as the MacBook Pro Retina that have upgradable RAM. I am letting this one slide it now.

So, why does the Mac mini have soldered RAM?

Also, why does the 21.5-inch iMac have soldered RAM?

I can't help but to reason that this was done so that Apple can charge exorbitant upgrade prices at the time of purchase.

After all, Scrooge McDuck ...I mean Tim Cook... needs more money so that he can build another solid gold swimming pool.

The simple answer is that Apple makes appliances that are glued and soldered together with planned obsolescence and control of prices for upgrades such as SSD prices and Ram additions. (profit center)

The only user upgradeable computer is the 27" iMac to increase Ram and which is a pro-consumer model. Even the Mac Pro is limited to upgrades do to no PCIe slots and Apple custom parts.
 
  • Like
Reactions: tubeexperience
The simple answer is that Apple makes appliances that are glued and soldered together with planned obsolescence .
If I hear that abomination one more time I am gonna turn green like Hulk and destroy Macrumors.....
 
The simple answer is that Apple makes appliances that are glued and soldered together with planned obsolescence and control of prices for upgrades such as SSD prices and Ram additions. (profit center)

The only user upgradeable computer is the 27" iMac to increase Ram and which is a pro-consumer model. Even the Mac Pro is limited to upgrades do to no PCIe slots and Apple custom parts.

It is very unfortunate that Apple is now hell-bent on making a profit by making appliances that are glued and soldered together so that they become obsolete faster.

Why do you keep fudging together what applies to Skylake and select Broadwell chips to ALL Haswell/Broadwell/Skylake ?

I gave you a response based on established testing (Intel's specifications), not based on my sampling of my own equipment - poor data set. It is a definite FACT that Intel did NOT specify those chips to run on 1866 Mhz DDR3L.

You asked for opinions on why Apple decided to use soldered RAM - to which I gave an educated guess on memory speeds and availability based on:

(1) Soldered RAM on most of Apple's lineup began with the Haswell generation (FACT).
(2) Most of Haswell/Broadwell non-desktop chips are not specified to use 1866 Mhz or above DDR3L (FACT based on Intel's published specifications) - only LPDDR3.

Both of those items are correct - which makes my opinion to your question an educated guess.

NOW, I do not know what the reason is for Apple's decision to forgo Intel's specifications for the Skylake iMacs, which appear to be an exception. To this, I replied:

(3) Skylakes appear to be an exception - I do not definitely know why but gave a few plausible ideas.

NOW, up to (2), everything is factual, which makes my original response a sound educated guess.



You need go get a statistical grasp on what I said - as PER INTEL, there is no official support for DDR3L at 1866 Mhz or beyond on the chips you implicitly referenced to, which means NOT ALL SYSTEMS ARE GUARANTEED TO WORK THAT WAY. My testing shows that in very specific instances, it can work - BUT this is specific to my setup and usage scenario. I did not make the fatal mistake to assume that what applies to my two computers, is sufficient evidence that it should work for all, despite what Intel specified. This argument needs to be made statistically, not based on limited, anecdotal evidence. I.e., the following does not work:

(A) I test a few setups to verify that DDR3L works at higher frequencies than published in 2 computer systems.
(B) I can assume that Intel is wrong and that whatever works for me, necessarily works for all configurations.

All that we can say is:

(C) It seems to work sometimes in specific scenarios, but probably not for all configurations, that's why Intel specified the chips their way.



Then please, by all means, ask this question directly to the Apple engineers instead of reaching out to a casual Mac forum, if you are this antagonistic toward contributors.

You said that Broadwell doesn't support DDR3L 1866, even though the Core i5-5675R and Core i5-5575R are officially listed by Intel as chips that do support DDR3L 1866.

After that, you conclude that Skylake does support DDR3L 1866, even though neither the Core i5-6500 nor the Core i5-6600 are officially listed by Intel as chips that support DDR3L 1866.

Then after that, you make the speculation that somehow Skylake unofficially support DDR3L 1866, while Broadwell does not. This makes no logical sense.

And as you correctly pointed out, soldered RAM begun in the iMac with the Haswell and the Haswell iMac does use neither DDR3L 1866 nor LPDDR3 1866, hence further pointing to the conclusion that the use of soldered RAM has nothing to do with the use of LPDDR3.

For what its worth, I know you support user-repairable computers based on your postings. I do not think the issue is Tim Cook, but rather Jony Ive. The latter appears to be hell bent on making everything thinner and into fashion accessories rather than work machines. I do not personally agree with their direction toward mass planned obsolescence. I find the keyboards on the new Macbook deplorable and the proprietary SSDs a pain to upgrade, financially. I used to have an iMac before the current glued-together machines and I will not own another iMac unless they reverse course on their designs.

This seems to be the more plausible answer.
 
It is very unfortunate that Apple is now hell-bent on making a profit by making appliances that are glued and soldered together so that they become obsolete faster.



You said that Broadwell doesn't support DDR3L 1866, even though the Core i5-5675R and Core i5-5575R are officially listed by Intel as chips that do support DDR3L 1866.

After that, you conclude that Skylake does support DDR3L 1866, even though neither the Core i5-6500 nor the Core i5-6600 are officially listed by Intel as chips that support DDR3L 1866.

Then after that, you make the speculation that somehow Skylake unofficially support DDR3L 1866, while Broadwell does not. This makes no logical sense.

And as you correctly pointed out, soldered RAM begun in the iMac with the Haswell and the Haswell iMac does use neither DDR3L 1866 nor LPDDR3 1866, hence further pointing to the conclusion that the use of soldered RAM has nothing to do with the use of LPDDR3.

I said that most models of Broadwell/Haswell does not support DDR3L 1866 Mhz. Each CPU is different in its electrical specifications despite the underlying transistors being the same due to its clock speed, voltage and other factors. For example, the integrated memory controller on modern CPUs are usually clocked synchronously with some parts of the CPU - the L3 cache, for example. But it is clocked at a ratio with the main CPU cores, which leads to different memory controller frequencies. This directly controls how fast the memory can effectively work at. The voltage of the memory controller also affects how fast you can store charges in the RAM. If you push the voltage lower to reduce power consumption, then it affects the memory reliability until you lower memory speeds. These are common themes in the PC enthusiast world since the first integrated memory controller on the Athlon 64 CPUs a decade ago.

Just because the desktop versions (of which there are only TWO) support DDR3L 1866 Mhz, doesn't mean the mobile chips do as well. In fact, due to lowered clocks and lower voltage available in mobile systems are just two factors of why you consistently see mobile chips not supporting as high memory clocks compared to desktop counterparts. In fact, none of the ultra low voltage Skylake CPUS (AFAIK) support DDR4 of any kind, only DDR3.

If you skim through Intel's specifications on their website, you'll see that what I said about MOST Broadwell and Haswell CPUs not natively supporting DDR3L >1600 Mhz is absolutely correct. The only 2 desktop CPUs you quoted are exceptions which Intel does natively support DDR3L > 1600 Mhz, which I suspect is due to the higher allowed TDP.

I never assumed that Skylake somehow support DDR3L > 1600 Mhz, what I have said is: "I would think the Skylake generation obviously paired well with the 1866 Mhz DDR3L in Apple's testing, or they were able to work something out with Intel." Which echoes what I summarized for you earlier - "(3) Skylakes appear to be an exception - I do not definitely know why but gave a few plausible ideas." With the plausible ideas being what I think may have happened. I never said I know that for a fact, but just I think it could be.

As for your last point about the use of LPDDR3 starting with Haswell - I need to remind you that the chipsets and layouts that work for Haswell works for Broadwell as well, as they are pin-compatible (in most systems). It is logical that the plan to use LPDDR3 started already at the Haswell generation, reaping performance benefits in the Broadwell generation. I also said that DDR3L at speeds greater than 1600 Mhz (especially SODIMMS) during the first half of Haswell's lifespan was rather sparse.
[doublepost=1467379309][/doublepost]
This seems to be the more plausible answer.

These ideas are not mutually exclusive. It is very possible to achieve a smaller form factor for whatever fashion reason AND still achieve better performance by, as an example, adopting LPDDR3 1866 Mhz or more. This picture of where Apple is headed isn't necessarily as black and white as what I interpret you are trying to wedge the thread toward.

In fact, my impression is that a large amount of the market in portable electronics is moving towards lower power form factors, first netbooks, then tablets, then phones, then watches and IoT. Most of the major players in the semiconductor industry - Intel, AMD, Nvidia, ARM, Qualcomm and Samsung are all heading towards lower power consumption devices. I expect to see LPDDR3/4 devices proliferate and it would make financial sense to design around these chips. It wasn't long ago that LPDDR support was completely absent from Intel CPUs. We all know what the new RX480 from AMD is touting, along with what Nvidia's Maxwell and Pascal have been about - performance and power optimizations. Intel's newest CPUs have reserved large performance gains for more power consumption reductions, something that I think many people are complaining about - the lack of pure performance increases over the last few generations.

I think you can see the larger picture here.
 
Last edited:
Cost and reliability
No. It makes repairs more costly and upgrade impossible.

Yes, it reduces the cost of manufacture, and it makes a more reliable product, which is less likely to require repair.

True, it also makes upgrade impossible, which may or may not be an issue, depending on an owner's situation.

In the case of the Mac Mini, only RAM was user upgradeable to a maximum of 16 GB. It can still be ordered with up to 16 GB RAM.

There is also a choice of HDD, Fusion or SSD. Some users have swapped out the HDD for an SSD.
 
Yes, it reduces the cost of manufacture, and it makes a more reliable product, which is less likely to require repair.

True, it also makes upgrade impossible, which may or may not be an issue, depending on an owner's situation.

In the case of the Mac Mini, only RAM was user upgradeable to a maximum of 16 GB. It can still be ordered with up to 16 GB RAM.

There is also a choice of HDD, Fusion or SSD. Some users have swapped out the HDD for an SSD.

In portable devices like the MacBook Pro, the devices are moved around and that might loosen the RAM and causes issues with reliability. I have never had issue with loosen RAM, but it theoretically could be possible.

That said, the the Mac mini is not a portable device, so this isn't be a problem in the first place.

Furthermore, soldered RAM means more expensive repair since defective RAM means that the whole logic board had to be replaced.
 
In portable devices like the MacBook Pro, the devices are moved around and that might loosen the RAM and causes issues with reliability. I have never had issue with loosen RAM, but it theoretically could be possible.

That said, the the Mac mini is not a portable device, so this isn't be a problem in the first place.

Furthermore, soldered RAM means more expensive repair since defective RAM means that the whole logic board had to be replaced.

Why ask the question? The answer is simple enough, but you seem to want to complicate things.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.