Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
I could very well see them knowingly installing a worse usability keyboard if they could not shrink without it.
It's worse in some ways, better in others. Many people prefer it, can type faster on it, have fewer errors, like the feel, etc. Many find the opposite. Based on comments here and elsewhere, I think @CaptRB is justified in his claims about most users liking it. That doesn't mean that can be proven here. You have to read the threads about it and draw your own conclusions.
 
Again, as just explained, this is pure mythology. Again, there is already room for a larger battery in the new MBPs. Again, the size has nothing to do with the reason there isn't 32 GB RAM. Again, that's about battery life.

I was talking about RAM and battery combined. So what the heck are you arguing about? Or posting (yet again) just to defend precious Apple? :)

P.S.
EOD from my part, weekend is over, and since there is no real discussion over here anywhere, I really don't want to argue with anyone, so cya :)
 
  • Like
Reactions: SteveJUAE
My assumption is that datasets (or better say workloads) requiring 32GB in 2016 were already common enough to make this kind of memory available in laptops whatever the way (hence Dells and others with desktop RAM in laptops to achieve this) but not common enough for a Pro laptop not having it to be considered outdated no matter its other characteristics.
Comparing a server to a notebook is rather silly, they are different beasts with different workloads and purposes. That's why they have different kind of hardware and why they are tuned differently. An ssd in a notebook or desktop is more tuned to random reads/writes whereas a server has one tuned for sequential reads/writes. That means a notebook/desktop is faster with random reads/writes than a server is.

Well, here are my stats:
Which are meaningless since you didn't post your uptime. The longer the machine is up, the higher these stats will be but that doesn't mean there is a memory issue or that the machine requires lots of memory. The OS will simply keep things in memory until it needs the additional memory. It then only takes what it needs. That way we have hot startup of an app which is a gain in performance.

Mine is actually using 455MB of swap even though 7GB out of 16 is in use. Memory management today is rather different than back then.

Funny to see here that none of the people claiming to need big amounts of memory are capable of mentioning the fun things one can do with memory: caching. Nobody mentions something like ccache which greatly speeds up compiling code.
 
Holy ****! :eek:
What do you run that maxes out your ram like that?

The things eating up significant RAM in that instance were a Windows VM running HP ICR software (where the Windows VM needs 8 GB of RAM to run the applications smoothly), DiskWarrior, Mac OCR ran on the host OS, Excel, Preview, SPSS, and a Windows NT 4.0 VM that would be waken intermittently when-needed from Suspend (but I'm not sure if the OS releases most of the RAM when suspended or not until I shut down the VM?)

(So certainly the computer does pretty well with what it's got already, but I look forward to when we see a little more :))
 
  • Like
Reactions: Sanpete and Meister
The things eating up significant RAM in that instance were a Windows VM running HP ICR software (where the Windows VM needs 8 GB of RAM to run the applications smoothly), DiskWarrior, Mac OCR ran on the host OS, Excel, Preview, SPSS, and a Windows NT 4.0 VM that would be waken intermittently when-needed from Suspend (but I'm not sure if the OS releases most of the RAM when suspended or not until I shut down the VM?)

(So certainly the computer does pretty well with what it's got already, but I look forward to when we see a little more :))
Windows VM, that explains it
 
Windows VM, that explains it

Yeah, neither Microsoft nor the developer of the ICR suite have quite gotten memory efficiency as well as Mac has with OS X - for whatever reason, working with uncompressed TIFFs in batches of scans seem to impact Windows RAM use more significantly than it does with Mac :(
 
  • Like
Reactions: Meister
Well of course we all understand there are datasets that take more than many TBs of RAM some servers have. It's just how common they are...

My assumption is that datasets (or better say workloads) requiring 32GB in 2016 were already common enough to make this kind of memory available in laptops whatever the way (hence Dells and others with desktop RAM in laptops to achieve this) but not common enough for a Pro laptop not having it to be considered outdated no matter its other characteristics. So MBP late 2016 had its share of critics for RAM but was still OK overall that's why it flied.

Yes, that seems likely enough.

In 2017 pretty much everyone will offer 32GB in their pro series of laptops and there will be opportunities to make this without using desktop RAM so if Apple releases next MBP without 32GB option we'll hear much more critics.

If Apple can achieve 32GB with LPDDR, I'm sure they will. (Not sure DDRL would make enough difference.) The Kaby Lake specs you link to aren't clear about whether the chip can accept 32 GB LPDDR3, but if it can, that explains the rumor that Apple intends to offer a 32 GB model this year.

JFYI, Intel mobile CPUs can address 64GB of DDR4, LPDDR3 and DDR3L, here is an example http://ark.intel.com/products/97462/Intel-Core-i7-7920HQ-Processor-8M-Cache-up-to-4_10-GHz, see Max Memory Size. But there are no LPDDR3 and DDR3L 32GB modules yet and I believe Apple doesn't want to install 4 modules no matter what for battery and size reasons. I believe there were no 16GB LPDDR3 and DDR3L modules when MBP 2016 released that's why we didn't have 32GB then. I think I've read they're available now.
I've seen news of 16GB LPDDR3 too, for phones. Looks like Kaby Lake will only accept two modules, if that's what two memory channels means.
 
Yeah, neither Microsoft nor the developer of the ICR suite have quite gotten memory efficiency as well as Mac has with OS X - for whatever reason, working with uncompressed TIFFs in batches of scans seem to impact Windows RAM use more significantly than it does with Mac :(
Another reason I never use windows.
 
I was talking about RAM and battery combined. So what the heck are you arguing about?

Huh? it's very clear what I'm talking about. You very plainly implied that if the MBP were thicker it could accommodate a bigger battery and more RAM. Again, and I don't know how to be more clear and simple about this, it doesn't need to be thicker to accomplish those things. That it does is a persistent myth. Yes, no point to arguing about that.
 
We can all dream, of course. I'm guessing you don't really want Apple to use a battery that doesn't pass their tests, though. And you seem to see the conflict between wanting better battery life and a higher-powered dGPU.
The list I made was referring to what I would have realistically liked to see from the MBP 2016, not my wish list which is quite different. As I said, increasing GPU performance would be hard given the TDP limitations. But with graphics switching, it wouldn't be as much of a problem.
 
I hope that software has some great qualities that make up for the huge drawback of not being able to accommodate swapping! Maybe it's software for some particular company?

Actually most standard statistics packages require you to load all the data into RAM to fit a regression for example. I've also run into problems where the data might not be that big but the model is big and programs can crash. Certain kinds of model you can treat each data point as a variable and they get huge in RAM quickly - this is where I first needed > 16GB - a model with 18,000 data points = 18,000 variables running in the niche gold standard proprietary software that would just crap out after 24 hrs runtime. Only solution to that was more RAM. There are other things like hadoop for working with big data of course that can deal with swap space, but its not useful for my problems.

But like I know ye were all thinking about the intricacies of evolving solutions for multi-level hierarchical bayesian mcmc models with thousands and variables when spouting your generalised opinions, right ? (Edit: by the way before someone says I'm talking about a niche. Its that niche known as data science/statistics - heard of it ?)
 
Last edited:
The list I made was referring to what I would have realistically liked to see from the MBP 2016, not my wish list which is quite different. As I said, increasing GPU performance would be hard given the TDP limitations. But with graphics switching, it wouldn't be as much of a problem.
I see. Can you be more specific about what chip you have in mind? The 460 is faster than what was offered in the XPS at the time, for example. The alternatives faster than that did seem unsuitable for an MBP, even with graphics switching.
 
These are your words, assumption and opinion, not mine :rolleyes:



Yet again I think your have misread I clearly stated:



Which you have replied with "needed by a lot of people" :rolleyes: unless you mean this, then you have contradicted yourself in the next sentence with "and a small minority needs more than 16Gb RAM" LOL

No one suggested 32GB should be mandatory just an option like other good OEM's for a minority now or maybe a majority in the future, who knows but currently your option less with Apple.

The screen is only better in some respects others eg PPI have not changed. The reason I implied that colour accuracy is more a pro feature is simply based on professionals who mandatory require this for printing (for example). For general consumption and entertainment it's generally accepted that much higher gamma etc is more pleasing. Even Apple realised this before we could easily adjust our monitors, when they swapped over to the more vibrant but less accurate windows default at the time.

I am not wrong I simply stated



You assumed and limited it to October 2016, not me :rolleyes:

They could of done the same as other OEM's for example and released just prior to the holiday season a MBP that supported 32GB with a slightly larger battery to compensate for some of the battery endurance lost whilst in nanny/mimic mode of less powerful laptops. There is no doubt many assumptions and best guesses why they did not do this, but the fact remains others managed it.

Time out now as I have had enough of these trivia ping pong matches :(

Thing is some are incapable of understanding that others may have a real need for significantly more RAM and or processing power. Apple simply do not produce a heavy lifter for those on the go, with adequate usability. Focus is the well heeled consumer, with "Pro" being little more than a differentiator in Apple's range of notebooks. Consumer products for consumers...

Like others I would like to see the MPB focused on performance, not being ever thiner :rolleyes:

Q-6
 
Last edited:
Actually most standard statistics packages require you to load all the data into RAM to fit a regression for example. I've also run into problems where the data might not be that big but the model is big and programs can crash. Certain kinds of model you can treat each data point as a variable and they get huge in RAM quickly - this is where I first needed > 16GB - a model with 18,000 data points = 18,000 variables running in the niche gold standard proprietary software that would just crap out after 24 hrs runtime. Only solution to that was more RAM. There are other things like hadoop for working with big data of course that can deal with swap space, but its not useful for my problems.

But like I know ye were all thinking about the intricacies of evolving solutions for multi-level hierarchical bayesian mcmc models with thousands and variables when spouting your generalised opinions, right ?
Is there no software available that does what you want that offers swapping? As you say, that wouldn't solve all your issues, but it would be a lot more flexible. Laptops can't realistically do all that a desktop can, so the software needs to fit to get the most out of a laptop.

If I understand, you believe a Skylake laptop could handle your 18K-variable analysis if only it had 32 GB RAM. Have you tested this on, say, an XPS? As dyn and others have mentioned, there are other limitations that might also affect the practicality of doing that on a laptop.

What generalized opinions do you refer to?
[doublepost=1491166389][/doublepost]
Thing is some are incapable of understanding that others may have a real need for significantly more RAM and or processing power.
I haven't seen anyone say or imply that.
 
  • Like
Reactions: HenryDJP
and a Windows NT 4.0 VM that would be waken intermittently when-needed from Suspend (but I'm not sure if the OS releases most of the RAM when suspended or not until I shut down the VM?)
First: good lord, someone is still using Windows NT 4.0 in 2017! Talking about getting your moneys worth :p

Second: when you suspend a vm all the memory contents get written to a file on disk. The application may still hold the memory so it can assign it to other vm's but I'm not entirely certain on that. Hypervisors tend to do their own resource (cpu, memory, etc.) management (they request a certain amount from the OS and manage whatever they are given themselves). If the hypervisor does give it back to the OS it probably is going to stay put until something else requires it.

Pausing a vm is a different thing because it only pauses use of the vm, the resources it uses are not being touched so there is no writing to disk.

Windows VM, that explains it
Don't underestimate memory management in modern Windows versions (and memory usage in Windows NT 4.0 is peanuts compared to todays standards).

I've seen news of 16GB LPDDR3 too, for phones. Looks like Kaby Lake will only accept two modules, if that's what two memory channels means.
That's another thing to note: with 32GB of memory you do not want to have only 2 memory channels available, it has a negative impact on performance. You may have lots of memory now but it will be slower than having 16GB. SSDs have the same issue which is why they equip the large SSDs with additional lanes (i.e. 4 instead of 2).

Again, one does not simply slap in twice the amount of memory ;)
 
  • Like
Reactions: ZapNZs
That's another thing to note: with 32GB of memory you do not want to have only 2 memory channels available, it has a negative impact on performance. You may have lots of memory now but it will be slower than having 16GB. SSDs have the same issue which is why they equip the large SSDs with additional lanes (i.e. 4 instead of 2).

Again, one does not simply slap in twice the amount of memory ;)
Would be interesting to see some comparisons of the 16 and 32 GB versions of the XPS to see if it helps or hurts in particular situations.
 
I have a 2016 MBP and have used it at work and home a little over two months. I would say it is the first Apple OS X device I've considered returning (I've owned two previous MBP and an iMac). It has many pro's: beautiful, light machine, with an excellent screen. It's quiet. It's still the most powerful portable device in the Mac ecosystem. I like OS X, which is probably the biggest reason I haven't sold it yet. I like the Finger print ID and find that to be one of the more helpful features.

Neutral: I don't mind the keyboard (though it is unnecessarily loud) or USB C (which was a surprise). I have one new cord that connects to my T'bolt display and a couple of others that I keep in my backpack. This feels like a temporary concern as more devices move to BT or WiFi, but I understand some of the frustration in the interim. I miss the SSD reader, but can connect my camera to it by USB... I don't much care one way or the other about the Touch Bar. It looks cool and my kids like it, but in day to day use for work, I don't find it to be much of an asset. It may become more useful over time, but it's not much use for me right now.

Cons: The track pad is a pain. It is unnecessarily large and with "tap to click" selected in user preferences, the palm rejection is nonexistent. If you turn off "tap to click," it requires an actual click to use the trackpad, but I no longer have so many palm rejection issues. I just don't see the point in the giant track pad. It seems to start up slower than my 2011 MBP with 16GB RAM a 3rd party SSD. Not sure why, but the new MBP takes a few seconds longer to start up, even from sleep. The Bluetooth is a bit temperamental. Typically, it doesn't recognize at least 1 or 2 devices until I open Bluetooth settings and force it or repair. This is odd, since I am connecting it to Apple accessories (track pads at home and work, keyboards at home and work). Connecting to VPN (as we're unfortunately required to do for some of our clients) is much, much slower than on a PC. In fact, I more often use Parallels to connect and get what I need than Mac.

Bottom line: For a very expensive machine, this just doesn't feel like it's worth the premium over similar Windows devices. It feels like a compromise - in RAM, in processing, and even in half attempts to make up for their desires to avoid a touch screen (gigantic track pad and touch bar).

As a small business owner, I frequently purchase computers for our firm and am in the process of getting two additional machines. I'm increasingly inclined to go to the Dell XPS/Precision 15 or SurfaceBook Performance. Both have touch screens and are the same (SB2) or lower cost (Dell).

Hope this helps
 
Last edited:
  • Like
Reactions: c0ppo and muratura
More random venting. What on earth is "wannabe portability"? Who knows what you mean by "nanny mode," but most people, including most pros, do use their laptops in ordinary ways for browsing, watching videos, text editing and other things for which the MBP gets excellent battery life. That's not eking out good battery life, it's just using the machine normally. Obviously that doesn't mean those people only use it for those things. They may also use it for more heavy-duty things.



A few more blows to this dead horse are just what's needed! The wider color, brighter--and higher-contrast--screen is something that can be achieved while preserving battery life and and other goals. Again, that's not true of 32 GB RAM, which would eat into battery life. Apple did, as a matter of plain fact, use the latest CPU. The dGPU was also the most advanced available that would fit with the other goals for the laptop.

The lack of repairability/upgradability is a real drawback, of course, but none of this has much to do with form factor, which is what you appear to be trying to talk about.



You can already fully use the power of these ports. Why do you suppose otherwise?



Again, a mix of myth, confusion about what was available last year, and pure nonsense about what "all" pros need.



No, not clear in the least. You believe what you want to, not what facts support.

I normally try to ignore your posts and file under over enthusiastic but as it's self evident even here, you seem to have crossed that point 100's of posts back :rolleyes:

Unless you work for Apple or have hold of statistical data that no one else has your opinions are no more worthy than anyone else and random web sampling of postings and reviews are not facts, which is a point that seems to elude you.

It's simply tiresome and as the Mods have already noted several times encourages bickering to reply to your almost endless countering of negative MBP opinions

It's almost reaching a point where your posts are counter productive to healthy debate and just brings out the worse in people and often ends up with a debate on trivia and miss quotes/interpretation.

I'm sure you have some interesting things to say and points to put across but IMO they are currently muddled by your over enthusiasm.

Lets have some quality over quantity, until then I am not playing ping pong replies with you :p :D
 
I normally try to ignore your posts and file under over enthusiastic but as it's self evident even here, you seem to have crossed that point 100's of posts back :rolleyes:

Unless you work for Apple or have hold of statistical data that no one else has your opinions are no more worthy than anyone else and random web sampling of postings and reviews are not facts, which is a point that seems to elude you.

It's simply tiresome and as the Mods have already noted several times encourages bickering to reply to your almost endless countering of negative MBP opinions

It's almost reaching a point where your posts are counter productive to healthy debate and just brings out the worse in people and often ends up with a debate on trivia and miss quotes/interpretation.

I'm sure you have some interesting things to say and points to put across but IMO they are currently muddled by your over enthusiasm.

Lets have some quality over quantity, until then I am not playing ping pong replies with you :p :D

Yet more random venting. Why bother quoting something if you're going to ignore everything it actually says and just make stuff up instead? Look inward.
 
Isn't a bit difficult to gage the overall consensus in a forum? I could be wrong but most of us in these forums don't represent the majority of people outside who but Apple products. From what I've seen in the Apple Store it seems a mixed bag, people gravitate towards the MacBook Pro's with Touch Bar (probably because it's new and looks great) but when they see the price most people hage a little (what we in Britain call) a moan/complain. That was my experience the other day at the Apple Store I went into.

But my general opinion on the new MacBook Pro with Touch Bar was better than I thought it would be, i tried out the 13" with Touch Bar and it was very responsive, I was shocked at how bright and clear the display was and I loved the Touch Bar, I think as time goes on more and more apps and developers will make it more useful. They certainly are really nice machines. In fact I'm thinking about picking one up later in the year probably after the spec update (2nd gen is it called?) and there's rumoured to be a price drop.
 
I love my 13" tbMBP but the issues with sleep/external lg displays is pretty annoying. In general its fine if you leave it plugged in to the 2 21" LG ultrafines but if you disconnect it and work on the go for a while it quite ofter has a fit when you reconnect it to the displays. The issues can be one of the following:

Bluetooth not working - requires reset
Monitors not turning on - requires reset
System freezes - requires reset
No output to monitors and when you try to disconnect and use the laptop screen that doesn't work - requires reset

I really hope these issues are resolved by software updates but its a bit of a worry that its still not resolved so long after launch.
 
I love my 13" tbMBP but the issues with sleep/external lg displays is pretty annoying. In general its fine if you leave it plugged in to the 2 21" LG ultrafines but if you disconnect it and work on the go for a while it quite ofter has a fit when you reconnect it to the displays. The issues can be one of the following:

Bluetooth not working - requires reset
Monitors not turning on - requires reset
System freezes - requires reset
No output to monitors and when you try to disconnect and use the laptop screen that doesn't work - requires reset

I really hope these issues are resolved by software updates but its a bit of a worry that its still not resolved so long after launch.

Haven't had any of these issues on my 13" TB but I'm only using 1 LG external ("Ultrafine") - have you done PRAM and SMC reset?
 
I knew someone would post something like this. First - don't tell me how to do my work when you know nothing about it. Second - I don't need a HPC, I just need 32GB of RAM - hence the desktop I have. I could easily get HPC access, it doesn't suit my needs and working in that environment would slow me down hugely. Third - the datasets I use are typically small - its the models that are complex. The model complexity is limited by the RAM and GHz - but out of those two the RAM is harder to get around. Fourth - yes you can simplify the data/model - thats the exact opposite of what i'm trying to do which is use all the available informaiton.

Here the question though: do you really need all that RAM all the time (that is, will you access random elements within 32GB of RAM scratch space), or is it the case that you will generate 32GB of temporary (or not so temporary) results? Because if first is the case, a quad core CPU is already a limiting factor. With 8MB of L3 cache and random access across 32GB of RAM you will get cache misses constantly, which will cut down your performance dramatically.

In my experience, many "large" problems only work with small portions of data at a time. Image processing is a prime example — the data is accessed basically in sequentially, in very predictable patterns. Statistical modelling, large matrix operations etc. — same thing. For these problems, it doesn't make any difference performance-wise whether you have all your data in RAM or whether you use a multiple-buffering schema (work on one portion of the data while loading the next ones). The real problem is that software developers are most of the time too lazy to write proper code. One just assumes that the entire problem will fit into RAM. Already using memory-mapped I/O would solve a lot of the RAM-related issues that people are having.
 
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.