Phil Schiller Elaborates on Power Concerns with 32GB MacBook Pro

[Arri]

7 minutes ago, LAwLz said:

That is not true. Just because people buy something, doesn't mean that it actually suits their needs. That's the problem with "using your own judgement" when it's about a subject you have very little knowledge about, there is a high chance that you make the wrong judgement.

 

Argumentum ad populum

That same argument can be used for what you just did what people are doing. Using one's own judgment to decide what someone needs even though you have very little knowledge of said person.

 

Argumentum ad populum.

 

Just because many others are agreement that someone is making the wrong decision, doesn't mean that they are making the wrong decision.

[Sniperfox47]

1 hour ago, Arri said:

That same argument can be used for what you just did what people are doing. Using one's own judgment to decide what someone needs even though you have very little knowledge of said person.

 

Argumentum ad populum.

 

Just because many others are agreement that someone is making the wrong decision, doesn't mean that they are making the wrong decision.

I can empirically prove that most people I know make the wrong decisions about laptops. If I could get a large enough sample size, I'm also pretty sure I can empirically prove people suck at making product choices.

 

People are dumb.

[Dabombinable]

45 minutes ago, Sniperfox47 said:

I can empirically prove that most people I know make the wrong decisions about laptops. If I could get a large enough sample size, I'm also pretty sure I can empirically prove people suck at making product choices.

 

People are dumb.

AMD A8 "uad core" at "2.4GHz" with AMD Radeon "Graphics". Turns out that AMD went with the technically true route with their APU-they act more like dual cores in the majority of circumstances, the clock speed never goes above 1.8GHz, and even then it rapidly starts fluctuating between 900-1.5GHz due to TDP throttling that is far too aggressive. And the graphics they were talking about....are slower than the MR HD5650 in my old laptop-especially when both the CPU+iGPU are under a heavy load. That laptop had a Phenom II P920 BTW which is a true uad core, and what I expected to see inside the newer laptop.

[Misanthrope]

On 11/21/2016 at 0:53 PM, dalekphalm said:

Who the HELL doesn't charge their laptop for a WEEK?

 

This just confirms it for me. Phil Schiller doesn't know what the fuck he's talking about - or he's marketing to people that don't exist.

 

I know multiple people with Macbooks (Both Air and Pro) and not one of them will leave their computer in standby mode for more then a day.

 

Furthermore, who fucking cares if the battery dies after leaving it in standby mode for 3 days? macOS should have built in protections that will write current memory to the HDD, and safely shut down the laptop once the battery reaches critical low state. This is STANDARD technology that has been in laptops for literally over a decade.

 

They've designed the RAM and power system to solve a problem that no one has...

My work laptop is old enough that it has an i5 520m (So that's closing in on 7 years old) and it already does exactly that: I forgot to plug in today and it just hybernated when it reached 5% battery.

 

Can we get 2010 technology on a 2016 product Apple? Cause that would be great instead of your shitty excuses.

[Dabombinable]

2 minutes ago, Misanthrope said:

My work laptop is old enough that it has an i5 520m (So that's closing in on 7 years old) and it already does exactly that: I forgot to plug in today and it just hybernated when it reached 5% battery.

 

Can we get 2010 technology on a 2016 product Apple? Cause that would be great instead of your shitty excuses.

Also, a fucking touch screen-something that sub $1000 AUD laptops have had since at least 2012.

[shymi]

12 hours ago, LAwLz said:

That is not true. Just because people buy something, doesn't mean that it actually suits their needs. That's the problem with "using your own judgement" when it's about a subject you have very little knowledge about, there is a high chance that you make the wrong judgement.

 

Argumentum ad populum

^ this.
I have made the mistake(several times) to buy something without researching the product properly, using a single source of information or just get it on a whim with the reason "I NEED IT FOR... stuff" and actually not needing it afterwards.

[Keco185]

21 hours ago, Sauron said:

Memory compression is a hack you'd only use out of desperation, it's not equivalent to having more ram. If I have a 64gb swap that's not euivalent to having 64gb of ram.

It's different than swap space. Swap space is virtual memory.

[Sauron]

23 minutes ago, Keco185 said:

It's different than swap space. Swap space is virtual memory.

It's different from swap space, but it's still not the same as having more memory.

[Sniperfox47]

2 hours ago, Sauron said:

It's different from swap space, but it's still not the same as having more memory.

It actually is the same as having more memory without ram compression... Plus like a couple CPU cycles of extra latency from decompression...

 

Edit: Just like swap is the same as having more memory, just way way way slower, and higher latency memory.

[Sauron]

45 minutes ago, Sniperfox47 said:

It actually is the same as having more memory without ram compression... Plus like a couple CPU cycles of extra latency from decompression...

 

Edit: Just like swap is the same as having more memory, just way way way slower, and higher latency memory.

if you think "a couple of cpu cycles" is all it takes to effectively compress over 16gb of data you're deluding yourself.

[RedRound2]

21 hours ago, LAwLz said:

I am calling bullshit. How can I be the only one calling bullshit on this entire statement?

He is saying that LPDDR3 gives 30 days of standby time, but DDR4 would reduce that to less than 7 days? So he is saying that DDR4 uses about 4 times as much lower as LPDDR3. Or wait, less than 7 days? So 6 days maybe? All of a sudden DDR4 uses 5 times as much power as LPDDR3. Not just that actually, because other components are also on while in standby (but in very deep sleep states), so it might be more like 8 times as much.

 

Does anyone actually buy this for a second? That DDR4 uses something like 8 times as much power as LPDDR3? Have you all gone insane?

 

LPDDR can enter deep sleep mode while normal DDR can't. Sure the power gap between LPDDR3 & DDR3 is much greater than LPDDR3 & DDR4 but the gap still exists and plus power saving features like deep sleep is what enables LPDDR equipped device to have a very long standby time

 

And btw, they can't just lie. They must've done numerous tests and there may be a whole lot more variables in play here but it is what it is

[RedRound2]

18 hours ago, Misanthrope said:

My work laptop is old enough that it has an i5 520m (So that's closing in on 7 years old) and it already does exactly that: I forgot to plug in today and it just hybernated when it reached 5% battery.

 

Can we get 2010 technology on a 2016 product Apple? Cause that would be great instead of your shitty excuses.

 

You do realize hibernate kills an SSD faster than usual (and it's slower to resume too). Standby with close to zero power consumption (hence LPDDR) is the ideal way and currently MacBooks can last months on standby

[Misanthrope]

1 minute ago, RedRound2 said:

You do realize hibernate kills an SSD faster than usual (and it's slower to resume too). Standby with close to zero power consumption (hence LPDDR) is the ideal way and currently MacBooks can last months on standby

At the expense of gimped amount of ram? That's a very bad compromise: SSDs today are reliable enough that it wouldn't affect them much vs just having a "pro" product without sufficient ram.

[dalekphalm]

6 minutes ago, RedRound2 said:

You do realize hibernate kills an SSD faster than usual (and it's slower to resume too). Standby with close to zero power consumption (hence LPDDR) is the ideal way and currently MacBooks can last months on standby

Err when you say "Hibernate kills an SSD faster than usual", what you really mean, is that it adds writes to the SSD.

 

That's not a problem, really. Modern SSD's can take PB's worth of writes. Writing a few GB extra data every few days is not going to significantly impact the life of the SSD.

 

Also, yes, SSD's are slower then system RAM. But resuming hibernation is even faster then booting off an SSD. In my experience, we're looking at maybe 10-15 seconds tops, and that's if you had a shitload of stuff in RAM at the time of hibernation. In most cases, I've seen SSD enabled laptops resume from hibernation in under 5 seconds.

[LAwLz]

16 minutes ago, RedRound2 said:

LPDDR can enter deep sleep mode while normal DDR can't. Sure the power gap between LPDDR3 & DDR3 is much greater than LPDDR3 & DDR4 but the gap still exists and plus power saving features like deep sleep is what enables LPDDR equipped device to have a very long standby time

[Citation Needed]

I want a good source as well. Not some random guy just saying it is true on Macrumors or something. The reason I want a good source is because DDR4 has lots of power saving features as well, such as a low-power self refresh mode and a power-down state. You can read about the low power auto self refresh here (page 148) and you can read about DDR4's maximum power saving mode on page 158.

There are also other features like only refreshing the parts of the array that are in use. My guess is that the features you are talking about like deep sleep, is the same ones as we now have in standard DDR4 as well.

 

18 minutes ago, RedRound2 said:

And btw, they can't just lie. They must've done numerous tests and there may be a whole lot more variables in play here but it is what it is

Yeah, surely they can't just lie. No company has ever lied to anyone. /sarcasm

Surely you can not be this naive...

[RedRound2]

18 hours ago, Misanthrope said:

At the expense of gimped amount of ram? That's a very bad compromise: SSDs today are reliable enough that it wouldn't affect them much vs just having a "pro" product without sufficient ram.

 

 

You talk as if Apple's limiting the RAM to 8GB or something. Can you stop complaining about something you and most people don't need. At this point, you're complaining for the sake of complaining

17 hours ago, dalekphalm said:

Err when you say "Hibernate kills an SSD faster than usual", what you really mean, is that it adds writes to the SSD.

 

That's not a problem, really. Modern SSD's can take PB's worth of writes. Writing a few GB extra data every few days is not going to significantly impact the life of the SSD.

 

 

Yes of course that's what I meant. Beyond a certain writes SSDs reliability lowers significantly

Quote

Also, yes, SSD's are slower then system RAM. But resuming hibernation is even faster then booting off an SSD. In my experience, we're looking at maybe 10-15 seconds tops, and that's if you had a shitload of stuff in RAM at the time of hibernation. In most cases, I've seen SSD enabled laptops resume from hibernation in under 5 seconds.

 

3

Actually no. Unless my understanding is wrong, resuming from Hibernation is slower than starting up just because the SSD need to read off the entire OS plus the load the apps that were already open. "Resuming Windows"takes pretty much forever in my experience while standby offers instantaneous resumeption

17 hours ago, LAwLz said:

[Citation Needed]

I want a good source as well. Not some random guy just saying it is true on Macrumors or something. The reason I want a good source is because DDR4 has lots of power saving features as well, such as a low-power self refresh mode and a power-down state. You can read about the low power auto self refresh here (page 148) and you can read about DDR4's maximum power saving mode on page 158.

There are also other features like only refreshing the parts of the array that are in use. My guess is that the features you are talking about like deep sleep, is the same ones as we now have in standard DDR4 as well.

 

Yeah, surely they can't just lie. No company has ever lied to anyone. /sarcasm

Surely you can not be this naive...

 

 

I can't find any direct comparison of LPDDR3 & DDR4. What I'm trying to say here is your conclusions are based on some assumptions and you and I know there is a lot more than that going on. For Apple, it's pretty much as easy as prototyping while we are trying to understand something beyond our expertise through fine print. Since LPDDR4 exist there must be some features exclusive to LPDDR RAM in addition to coming up with news way to make 4th gen more efficient

Quote

Yeah, surely they can't just lie. No company has ever lied to anyone. /sarcasm

Surely you can not be this naive...

 

 

Companies usually don't lie especially in cases like this. If they do this will just be a ticking time bomb that'll blow up in their face when someone experienced and reliable comes into the picture defuncting their claims. There's absolutely no reason for Apple to not add 32GB option especially considering they could've easily made it a higher priced model. 

[Sauron]

3 minutes ago, RedRound2 said:

Actually no. Unless my understanding is wrong, resuming from Hibernation is slower than starting up just because the SSD need to read off the entire OS plus the load the apps that were already open. "Resuming Windows"takes pretty much forever in my experience while standby offers instantaneous resumeption

Hibernation copies what's in your ram to your ssd. On resume, that exact data is copied back to ram. Because of how computers work, there is no need to reload everything from scratch. Resuming will only take as long as it takes to copy whatever was in your ram back to it from your ssd, which given the high speeds of the macbook pro's ssd should take 15, 20 seconds at most (assuming 32gb of ram full of data). Your windows experience is probably limited to sata ssds with 500MB/s reads.

[Senzelian]

On 21.11.2016 at 7:53 PM, dalekphalm said:

Who the HELL doesn't charge their laptop for a WEEK?

I do. I'm not always using it and when I'm on vacation I don't take it with me. 

 

But that makes it even weirder since my notebook's battery doesnt die because if that. Seems like Apple uses shitty batteries. 

[LAwLz]

2 minutes ago, RedRound2 said:

Yes of course that's what I meant. Beyond a certain writes SSDs reliability lowers significantly

That might have been an issue in the past, but not on modern SSDs.

Can't remember which website it was (Tech Report maybe), but some website did an SSD torture test to see how much they could write to some SSDs before it became an issue. The 250GB 840 Evo only started getting uncorrectable errors after 800TB. Please bear in mind that larger drives has more writes before they die, so 800TB on a 250GB SSD is fantastic.

 

Assuming you put your computer into hibernate once every day, and you averaged about 8GB of RAM usage when you put it into hibernate (assuming no memory compression), then it would take you 274 YEARS before it started to fail. Of course, that does not include other writes but the point is that it is by no means an issue. It was an issue back in the early days of SSDs when we had 40GB drives, with lower endurance, and people were doing things like defragging them, but that's not the case today.

Your SSD will be obsolete long before it dies from too many write cycles (and even then, drives today are good at putting themselves into a read-only state before that happens).

 

 

 

21 minutes ago, RedRound2 said:

Actually no. Unless my understanding is wrong, resuming from Hibernation is slower than starting up just because the SSD need to read off the entire OS plus the load the apps that were already open. "Resuming Windows"takes pretty much forever in my experience while standby offers instantaneous resumeption

Your understanding is wrong. Hibernation is not "start windows then start all the programs that used to be open", it's "I'll save everything I got in RAM to the drive, and then I'll just transfer it back instead of starting from scratch".

 

 

23 minutes ago, RedRound2 said:

I can't find any direct comparison of LPDDR3 & DDR4. What I'm trying to say here is your conclusions are based on some assumptions and you and I know there is a lot more than that going on. For Apple, it's pretty much as easy as prototyping while we are trying to understand something beyond our expertise through fine print. Since LPDDR4 exist there must be some features exclusive to LPDDR RAM in addition to coming up with news way to make 4th gen more efficient

Excuse me but what exactly is my conclusion? I think it is funny that you are saying I am making a conclusion based on assumptions, but my post doesn't make any assumptions. Your post on the other hand, is making an assumption. You have 0 evidence to support your claim that "LPDDR3 has a deep sleep state which uses less power than DDR4, which does not have that". Your entire foundation for making that assumption is "Apple always knows best and would never lie".

 

Yes, LPDDR4 exists for a reason, and it does use less power than regular DDR4. Just because LPDDR4 uses less power than DDR4 does not mean LPDDR3 has a feature which makes it better than DDR4 for standby.

 

 

43 minutes ago, RedRound2 said:

Companies usually don't lie especially in cases like this. If they do this will just be a ticking time bomb that'll blow up in their face when someone experienced and reliable comes into the picture defuncting their claims. There's absolutely no reason for Apple to not add 32GB option especially considering they could've easily made it a higher priced model. 

Ah yes, companies don't lie when customers are upset about an aspect of their product. Sure...

There is probably some logical explanation for it, but I don't buy that DDR4 uses soooo much more than LPDDR3. I am not even convinced DDR4 will use any more power than LPDDR3.

[Misanthrope]

1 hour ago, RedRound2 said:

You talk as if Apple's limiting the RAM to 8GB or something. Can you stop complaining about something you and most people don't need. At this point, you're complaining for the sake of complaining

Sure, as soon as they stop naming it "Pro"

[Trik'Stari]

Make. The. Battery. THICKER.

 

Not every laptop needs to be retarded thin, for fucks sake. Dell Notebooks (E6440's) use a battery that just hangs off the back, and is in fact nothing but a plastic housing containing a bunch of 18650 Lithium Ion batteries.

 

Oh wait, having an external battery would mean people can replace the battery, and that's more than the average apple user can handle.

[dragosudeki]

1 hour ago, Misanthrope said:

Sure, as soon as they stop naming it "Pro"

They can't remove the "Pro" from it because the "Mac book" might be confused by the literal book they made. Not that I disagree with you, or agree with Apple's choice to make a book, but it is what it is unfortunately.

[suicidalfranco]

4 minutes ago, dragosudeki said:

They can't remove the "Pro" from it because the "Mac book" might be confused by the literal book they made. Not that I disagree with you, or agree with Apple's choice to make a book, but it is what it is unfortunately.

Macbook should become the macbook air

Macbook pro should become the macbook

Mac book should become the mac picture

 

then they release a real macbook pro during CES, not at CES, during CES, because reason, it should have 2 USB-C/TB3 ports, 2 USB-A ports, magsafe, SD card slot, 7th gen i cpu, start at 16GB on the base model upgradeable to 32, quadro dGPU, and a better ventilation solution.

[Sniperfox47]

23 hours ago, Sauron said:

if you think "a couple of cpu cycles" is all it takes to effectively compress over 16gb of data you're deluding yourself.

Sorry, you're right, I should have been clearer.

 

What I meant was a few CPU cycles per page, not per total amount written, and even then only if assuming normal page size and not huge pages or large pages.

 

My comment was more with regards to the fact that more virtual memory literally is more memory. It's not the same as more physical memory, but it *is* more memory for your system.

[RedRound2]

18 hours ago, Misanthrope said:

Sure, as soon as they stop naming it "Pro"

And still, you complain about something only 2 people will ever need. It safe to conclude that you're only here to bash Apple for not including 32GB which they do indeed have valid reasons for. 

19 hours ago, LAwLz said:

That might have been an issue in the past, but not on modern SSDs.

Can't remember which website it was (Tech Report maybe), but some website did an SSD torture test to see how much they could write to some SSDs before it became an issue. The 250GB 840 Evo only started getting uncorrectable errors after 800TB. Please bear in mind that larger drives has more writes before they die, so 800TB on a 250GB SSD is fantastic.

 

Assuming you put your computer into hibernate once every day, and you averaged about 8GB of RAM usage when you put it into hibernate (assuming no memory compression), then it would take you 274 YEARS before it started to fail. Of course, that does not include other writes but the point is that it is by no means an issue. It was an issue back in the early days of SSDs when we had 40GB drives, with lower endurance, and people were doing things like defragging them, but that's not the case today.

Your SSD will be obsolete long before it dies from too many write cycles (and even then, drives today are good at putting themselves into a read-only state before that happens).

2

I do remember the article but that doesn't share the whole story. Manufacturers still void warranty after double digit TBs get written onto SSDs as they themselves know it's not reliable. Also laptop manufacturers still disable hibernate on their laptops just because of this reason. It's not a definite point where SSD's fail rather the probability of an SSD failing is much higher beyond a certain point  and that point is something anyone can surely reach if they do use their SSDs intensively like hiberanting

 

8GB RAM is a very conservative amount. For 16 and 32GB the estimate you put above will get divided by atleast two (with compression and assuming there isn't any write amplification). Couple that with a general use of an SSD and the fact that in usage scenarios like mine I would have to put the ssd's in hibernate state atleast like 5 times a day significantly hampering the SSD in a long run

19 hours ago, LAwLz said:

Your understanding is wrong. Hibernation is not "start windows then start all the programs that used to be open", it's "I'll save everything I got in RAM to the drive, and then I'll just transfer it back instead of starting from scratch".

 

 

I stand corrected but that still doesn't address the issue that where resumption from standby is usually less than second while from hibernate it takes a good 10-15 sec which is pretty much close to a cold boot

19 hours ago, LAwLz said:

Excuse me but what exactly is my conclusion? I think it is funny that you are saying I am making a conclusion based on assumptions, but my post doesn't make any assumptions. Your post on the other hand, is making an assumption. You have 0 evidence to support your claim that "LPDDR3 has a deep sleep state which uses less power than DDR4, which does not have that". Your entire foundation for making that assumption is "Apple always knows best and would never lie".

 

Yes, LPDDR4 exists for a reason, and it does use less power than regular DDR4. Just because LPDDR4 uses less power than DDR4 does not mean LPDDR3 has a feature which makes it better than DDR4 for standby.

 

Ah yes, companies don't lie when customers are upset about an aspect of their product. Sure...

There is probably some logical explanation for it, but I don't buy that DDR4 uses soooo much more than LPDDR3. I am not even convinced DDR4 will use any more power than LPDDR3.

2

 

Here's a good Reddit article with proper documentation links where it's stated the LPDDR3 RAM consumes about 10x less power than DDR4 in standby although their power consumption in normal workload is about the same. Couple this with the fact that SODIMM takes a whole lot more space in the PCB compared to LP RAMs and we can talk about a significant decrease in battery life of a laptop