32GB RAM MacBook Pro release - when?

[polbit]

I've used practically very 15" model since 2008 and the 2016 one has the best battery life of them all. Independent reviews also show this. So unsure what you are basing this statement on. The battery life on the 2016 model at least is excellent. Didn't work with the 2017, but I can imagine that it only improved with Kaby Lake.

I think it is pretty well established that the new MBPs have decent battery life when watching optimized video and/or light web surfing. Apple has done a fantastic job to maintain the stamina for that usage despite a much smaller battery.

It has ALSO been pretty well established that when pushing the laptop, the battery life dives predictably, and is worse than prior rMBPs. I've had MBPs for many years now, and I've never seen a any laptop discharge in an hour while running Parallels, or doing long compiles, or even under R.

 

[jerryk]

I'm using the MBP to prototype the code which is then run on a supercluster in a research center.

I do the same things on my Windows machine with 2 1080 GPUs. When I get the models running OK on small batches I feed it to the larger systems in the cloud with the full dataset. The GPUs give me a 8-10 to 1 performance advantage over the i7 in the machine. Probably 14+ to 1 over my Macbook 15.

What type of things are you doing? I am doing image based models with Convolution type Nets.

 

[Hippocrates]

Why would you expect these people to be using an MacBook Pro? Most are probably using something like an iMac. Data Scientists probably use something with an NVidia GPU to offload their processing to the GPU(s) for quick tests, and then do their real work on server farms. Big Data work is probably also done on servers.

That's precisely my point. Real hardcore computing tasks are often offloaded to the more appropriate machines, or done with MBP plugged in. MBP is built to be reasonably powerful so that one could still interface with the heavy-duty work while on the move and away from power source... for example, on a train, in a conference hall, or showcasing the work to collaborators/clients. In those cases, battery life and mobility are important considerations in the design of MBP too.

 

[ggggggmmmmm]

Why do you need 32GB?

A lot of us need 32GB or more. I need it for sims. Not all of us are bloody graphic designers or video editors. I am desperately waiting for an OS X laptop that accepts 32GB of ram so I don't need to keep buying fricken iMacs that are annoying and often impossible to travel with.

 

[EugW]

So I "only" have 16GB or RAM in my 2016 MacBook Pro but I was able to open 65 Tiff images in photoshop and work on them all without any slow downs or closing out any of the images. (My memory pressure didn't get yellow until I got to the 60th image). I hope this helps those who think they need 32GB of RAM to do professional type of work .

If you're getting to yellow frequently, then you've arguably overtaxed your RAM. You're compressing like crazy and probably jumping to the swap more than you should.

IMO ideally you should try to avoid the yellow if possible. OK, it's fine if it's just occasionally, but I'd argue if that you frequently have significant amounts of compressed RAM, you should consider getting more RAM.

BTW, a have a buddy who only has a few files open at a time and 16 GB is definitely inadequate. He'll be hitting the swap. However, his files are up to 500 MB each and he frequently has stuff Photoshop and In Design and a bunch of other apps open at the same time. 24 GB is usually OK for him but on a MacBook Pro, that would mean getting 32 GB.

A lot of us need 32GB or more. I need it for sims. Not all of us are bloody graphic designers or video editors. I am desperately waiting for an OS X laptop that accepts 32GB of ram so I don't need to keep buying fricken iMacs that are annoying and often impossible to travel with.

I find it strange when 32 GB Mac laptops don't exist, people will so often just say that we don't need them.

Well, I personally don't need them, but a lot of people can benefit from more than 16 GB RAM. Sure, they may only need 24 GB or whatever, but you can't buy a 24 GB MBP.

I personally don't need 32 GB, but I want more than 8 GB. How does that statement apply here? Similarly, when MacBooks were only available with 8 GB, the default response was that MacBook users shouldn't need more than 8 GB. Then all of a sudden, Apple started selling 16 GB MacBooks. Probably next year, Apple will be selling 32 GB MacBook Pros.

Truthfully, I don't usually need more than 8 GB, but there are occasions more than 8 GB is helpful for me, and this will only become more common as the years go by and the OSes get more bloated. Even if I only need 10 GB or so, you can't get 12 GB MacBooks, so the option is to get 16 GB. If you want to keep your machines a long time, given that you can't upgrade the RAM, then it makes sense to get more up front if you need it.

Furthermore, if you ever need even just a single VM, all those numbers are shot. Add another 4 GB per VM right there. Also, if you have a second user on the same computer, realistically you should add another 2 GB for even just light usage. I have a wife, and while her needs are light, she never logs out, so I should always assume I have 2 GB less RAM than what I have installed in my computer. IOW, just adding 1 VM and a second user will kill another 6 GB of RAM. Suddenly 16 GB doesn't seem like so much.

 

[leman]

It has ALSO been pretty well established that when pushing the laptop, the battery life dives predictably, and is worse than prior rMBPs. I've had MBPs for many years now, and I've never seen a any laptop discharge in an hour while running Parallels, or doing long compiles, or even under R.

Established by whom? We have a fair bunch of new models running now and there were no complaints about battery life, except early in the year, before Apple has fixed a GPU power state switching bug. For instance, I've took the laptop (2016 15" model) down from the charger last evening, it was at 100%. I was watching some youtube videos and reading articles after that (maybe for an odd hour), and now I've spent around 3 hours prototyping code in R and C (using Xcode). My battery is now at 77%.

BTW, R GUI for macOS used to have a GPU switching bug that triggered the dGPU when running.

Again, if you manage to utilise all your CPU cores 100%, your power draw will be at least 60 Watt, all included. A MBP battery (76Wh) will give you 76 minutes of runtime. A 99.9Wh battery, which is legally a max in a laptop, gives you 23 minutes more with the same power draw. I don't see a practical difference here — both is fairly unusable.

Edit: just did a quick test. The laptop was on 80%, started an infinite loop doing some randomised chi-square test on random data in R, which results in constant 100% CPU utilisation (single core). Was using the laptop a normal during that time (writing a paper). After an hour and five minutes, the battery was showing 42%. So 100% CPU utilisation drains 38% of battery in approx 1 hour. Which gives you 2 hours and 40 minutes on full battery, while doing intensive stuf non-stop. Of course, utilising the GPU or doing multi-core processing will drain the battery faster.

What type of things are you doing? I am doing image based models with Convolution type Nets.

Very heterogenous stuff, from some custom probabilistic modelling to bayesian phylogeny. Mostly stuff that involves a lot of replications or long MCMC walks...

 

[Pablo.A.Pucci]

I think that Apple it's lying to all of us, with the battery life excuse... RAM does't consume much energy... The reason that Apple could be doing this is that they feel if they gave us the 32 GB ram, we won't buy iMac's, Mac pro's or Final cut...

 

[leman]

I think that Apple it's lying to all of us, with the battery life excuse... RAM does't consume much energy... The reason that Apple could be doing this is that they feel if they gave us the 32 GB ram, we won't buy iMac's, Mac pro's or Final cut...

http://www.tomshardware.com/reviews/intel-core-i7-5960x-haswell-e-cpu,3918-13.html

Other tests I've read show similar results. So basically, we are looking just at around 6W always on additional power consumption (which for a laptop is very significant), not counting the advantage that LPDDR3 has in low power states.

Also, quite evil of Apple to create all this conspiracy agains us. They are even willing to lose money on more expensive RAM chips to pull it though!

 

[EugW]

I don’t think many of us would be happy with Mac laptops that didn’t use low power DDR.

I read somewhere that in standby mode, LPDDR3 uses roughly as low as 10% the power of regular DDR3.

 

[Msivyparrot]

I really need something more than the 8 miserable kilobugs apple shipped in my hunk o junk mac laptop!!! I would love to buy more maybe 128 kilobugs...my money...I dont care for battery, ipads last about as long as you need from home to the hairyport...

The silver budgies have many windows on the seats...with loads of loverly frames playing at something close to 25 frames per second and 48 0000 bits of audio in the same time span....

Things break, get thrown in the pool...which would not happen if I could spend my money on apple tax and buy new things called RAM...long ago, before steven and his sinodoors...we could change ram and things inside, it was so easy, a 40 year old could do it.... now only 5 year old can with torques...or is that the screwdriver...

I just want to spend my money at the temple to fruit...

 

[darksithpro]

https://www.reddit.com/r/hardware/comments/5dimal/lpddr3_vs_ddr4_power_usage/

"I researched this a while back.

So, the voltage isn't the whole story. Just to clarify, RAM is volatile storage, which means it needs a "constant" stream of power to keep the data, unlike SSDs (non-volatile) which obviously keep all their data even after you turn the system off.

Thus, on standby, standard RAM actually consumes ~30% of a mobile device's power draw.

RAM doesn't need "constant" power; periodic refreshes (every few milliseconds) are enough to retain the data. We can modulate those refreshes (see last sheet in this Excel sheet from Micron).

DDR3 uses 1.5V, while LPDDR3 uses 1.2V. However, LPDDR3 RAM compared to DDR3 RAMuses ~70% the active power usage, but 10% the standby usage. Image sourced from this article.

A 20% voltage difference, but (because of many other things not related to voltage, i.e. modulating the refreshes) a 30% decrease in active power and a 90% decrease in standby power.

DDR4 uses about 330mW when active, but even ancient LPDDR2 uses just 200mW when active (page 11). LPDDR3 uses about 50% more than LPDDR2, though when active (page 10). So that would put LPDDR3 at about the same active power consumption as DDR4, but I believe LPDDRx still holds a sizeable advantage in standby power consumption.

Micron states the advantage of LPDDR3 over DDR4 very directly in this PDF:

• DDR4 is "suitable for Windows 8 Connected Standby" (the same rating they give to power-hungry DDR3L; additional source with mW on pg 10).
  
• LPDDR3 is "ideal for Windows 8 Connected Standby"

Windows' Connected Standby is a low-power connected state that requires certain low-power states that only certain hardware can achieve; RAM is just one part (it also deals with WiFi chipsets and other stuff that I can't remember right now, haha).

The difference between LPDDR3 and DDR3L isn't just in measured power consumption. Micron's testing shows that DDR3L will give ~11 days of standby, while LPDDR3 yields ~55freaking days of standby. That's a 5x increase in standby!

JEDEC's research also shows this: when 10% of battery life remains, LPDDR3 yields 50% longer "Connected Standby" time than DDR3L.

So, when Micron says DDR3L and DDR4 are similar...that's pretty bad, at least compared to LPDDR3."

 

[badsimian]

Aaah the 32GB chestnut. I've been waiting for 6 years or more for this. Back in 2012 I *did* actually swap to a Windows machine (Lenovo W530) with 32GB of RAM. I moved back a year or so later as I could not stand the Windows experience with driver crashes etc all the time. I used to use that much RAM for setting up multiple VMs. Server environments with domain controller, SQL, Web etc all on different machines to simulate deployments etc with memory hungry server software. It had to be mobile as I would travel a lot, work on client site etc with often poor internet connectivity.

These days with more and more infrastructure moving to the cloud, be it Office 365 or Amazon etc and better mobile connectivity my need for 32GB has basically gone away.

Maybe that is what Apple predicted all along! I do have a friend who frequently runs out of RAM in Logic Pro with largish samples though....

 

[BlackHawk00]

Is an announcement imminent?

Well, it seems like we won't be seeing a 32GB RAM MBP for quite some time...

Intel reportedly delays Cannon Lake to end of 2018, raising questions about next year’s MacBook Pro
https://9to5mac.com/2017/09/20/2018-macbook-pro-cpu/

 

[EugW]

Well, it seems like we won't be seeing a 32GB RAM MBP for quite some time...

Intel reportedly delays Cannon Lake to end of 2018, raising questions about next year’s MacBook Pro
https://9to5mac.com/2017/09/20/2018-macbook-pro-cpu/

If that turns out to be true, it is possible that refresh could be delayed into 2019. However, it's also possible they would release an interim model, still at 16 GB.

It's rather a sad state of affairs that my Core m3 MacBook has as much RAM as a top of the line Core i7 MacBook Pro 15".

 

[sparkie7]

If that turns out to be true, it is possible that refresh could be delayed into 2019. However, it's also possible they would release an interim model, still at 16 GB.

 will definite release an interim model. With a new feature like 6GB graphics card, maybe even resurrect a 17" MBP for us dreamers...

 

[polbit]

 will definite release an interim model. With a new feature like 6GB graphics card, maybe even resurrect a 17" MBP for us dreamers...

17" MBP return is about as likely as a return of the headphone jack to the iPhone, unfortunately...

 

[sparkie7]

17" MBP return is about as likely as a return of the headphone jack to the iPhone, unfortunately...

Now, now. Don't be a pessimist...

 

[The Mercurian]

Why would you expect these people to be using an MacBook Pro? Most are probably using something like an iMac. Data Scientists probably use something with an NVidia GPU to offload their processing to the GPU(s) for quick tests, and then do their real work on server farms. Big Data work is probably also done on servers.

GPU doesn't suit all problems and data scientist is not synonymous with big data. Given a specific dataset, I could simply do a linear or logistic regression, or I could do a complex hierarchical Bayesian model for example. The CPU and RAM resources needed by the different models are quite different - even if operating on the same dataset. Certain models slow down when you run out of RAM, certain models crash when you run out of RAM. Some of this is down to user choice of software and model, some of it isn't. One particular piece of third party software I had to use (had to because it is the standard tool for the particular problem) would silently crash when you hit the physical RAM limit, and you would not know for days perhaps that this had happened (ahem....... ). The support service denied the issue, so my only solution was to move to a machine with bigger RAM.

Whether some people like it or not, extra RAM does solve/make easier certain problems for some people. Apple should give professional users the option to choose between RAM vs battery. Like their competitors do. Nobody is gonna force anyone to buy 32GB if they only want 16GB, so there is no need to get super defensive because others want more RAM.

Oh ....and please fix the keyboard.

 

[_Kiki_]

you should buy pc laptop with 64GB RAM it will last longer

 

[Naalsi]

I'm a composer and my job has me traveling all the time so an iMac or MacPro is out of the question.

Orchestral libraries are huge (there's one that's 100GB+) and eat up a lot of RAM. People with 64GB of RAM on desktops are having issues with them and have to freeze tracks to lessen the load.

So yeah, I'd love a 2018 MBP with 128GB of RAM if that's possible. I'd buy that in a heartbeat!

 

[_Kiki_]

I'm a composer and my job has me traveling all the time so an iMac or MacPro is out of the question.

Orchestral libraries are huge (there's one that's 100GB+) and eat up a lot of RAM. People with 64GB of RAM on desktops are having issues with them and have to freeze tracks to lessen the load.

So yeah, I'd love a 2018 MBP with 128GB of RAM if that's possible. I'd buy that in a heartbeat!

you can have 128GB RAM on desktop pc

 

[sparkie7]

I'm a composer and my job has me traveling all the time so an iMac or MacPro is out of the question.

Orchestral libraries are huge (there's one that's 100GB+) and eat up a lot of RAM. People with 64GB of RAM on desktops are having issues with them and have to freeze tracks to lessen the load.

So yeah, I'd love a 2018 MBP with 128GB of RAM if that's possible. I'd buy that in a heartbeat!

Why don't you buy a very fast SSD RAID and use it as a scratch disk? 2TB should more than suffice..

 

[Hitrate]

I'm a composer and my job has me traveling all the time so an iMac or MacPro is out of the question.

Orchestral libraries are huge (there's one that's 100GB+) and eat up a lot of RAM. People with 64GB of RAM on desktops are having issues with them and have to freeze tracks to lessen the load.

So yeah, I'd love a 2018 MBP with 128GB of RAM if that's possible. I'd buy that in a heartbeat!

Ya, I’m kinda irritated that soo many people in here, still, in late 2017 get this childish ignorant need to spout out their lame arguments against 32+ gb ram when we musicians have had this problem for yeears! Talk about cognitive dissonance... folks, please understand that SOME PEOPLE NEED MORE RAM !
[doublepost=1506673112][/doublepost]

you can have 128GB RAM on desktop pc

You clearly are not a producing musician with that comment.

 

[Ethosik]

Hey, it's great that things are working out for you, but that's only one type of professional work. Many of us find the 16GB limit to be...well...limiting. It's not some imagined problem in our heads.

And there are professionals that need 128GB of RAM. There are professionals that need SLI NVIDIA Quadro video cards. There are those that need 18 cores. What is the point exactly?

This whole my pro needs = everyones pro needs; therefore, these MacBook Pros are not pro! need to stop. Doesn't have the right hardware for your needs? Do not buy it! That simple.
[doublepost=1506688454][/doublepost]

That means it's using 16 GB, not that it needs 16 GB. Unused RAM is wasted RAM, so MacOS memory management doesn't start clearing out what isn't needed until more is actually needed.

If you have a 32 GB Mac, and it's showing 20+ GB used, I can almost guarantee the amount would be under 16 GB if that's all your Mac has, unless you have VMs or other apps open using very large amounts of memory.

Yep, I agree. I have about 7 GB of RAM used if I ONLY have iTunes open. My iMac has 40GB of RAM. It is different on my other Mac with only 8GB. People really do not understand how RAM works these days.

 

[Hater]

No you can buy a new Mbp, if you currently have one it has no more than 16gb of ram and has been doing the job or you would have moved to something else. Better SSD speeds will help as well.

You have a few choices

1. buy an mbp and accept that you may have to close some things when working on a ram intensive project or accept that it may not be as fast as you would like. Then sell to buy a new one when the 32gb models are released.

2. Buy a windows laptop with 32gb and accept the battery hit and the change in OS, and the myriad of other little niggles.

3. Get an iMac and do any ram intensive work on that.

As you have not stated why you need 32gb of ram, it's impossible to give you any specific advice.

4) Buy one of the business spec HP Laptops's (ZBook, EliteBook, ProBook) that cater to your memory needs. The EliteBooks will do 32GB, the ZBooks will do 64GB. All will run High Sierra.