Hacker News Reader

Comment by mrigor | Hacker News Reader

The new Macbook Pro supports max of 16GB RAM while Thinkpads support 32GB. This makes it hard to justify getting a Mac.

It's a tradeoff someone down the chain is making, LPDDR3 has a max of 16G and LPDDR4 is not supported on skylake/kabylake.

LPDDR3 being 30% less power drawing than standard DDR4.

So. More ram or more battery..

However I have a personal vendetta against Lenovo for their absolutely abhorrent practices in terms of consumer safety and their declining hardware standard.

Apple has reduced the battery from 74.9 to 54.5-watt-hour(27%). This makes the new macbook pro very thin.

https://www.ifixit.com/Teardown/MacBook+Pro+13-Inch+Function...

Phil Schiller, their marketing chief said:

> To put more than 16GB of fast RAM into a notebook design at this time would require a memory system that consumes much more power and wouldn't be efficient enough for a notebook.

ref: http://www.theverge.com/2016/10/28/13460496/apple-macbook-pr...

Nice to get a reply from the top as I was wondering why it was 16GB max too. I can't comment on the veracity.

I'm a Windows user, if that makes any difference.

It'll be interesting to see if & when this starts to influence desktop software design. Since the early 90s, it's largely been considered a waste of time to optimize client programs for speed or memory usage, because a new machine will just come out in a couple years with double the memory and double the processing power. There's a lot of low-hanging fruit in application-level performance, optimizations that could be done but aren't because the user won't perceive a difference. Now that Moore's Law has largely stalled out, I wonder if we'll start seeing innovation in software platforms (languages, frameworks) to optimize for speed & space instead of ease-of-use, to try and recover that performance that the hardware isn't giving us.

Jumping on the next-big-thing trend, but: Compilers with an AI that give you zero-cost optimisations. Plus IDE plugins for interpreted languages.

Actually, I'd guess someone is already working on it :)

Bill by milliseconds (AWS Lambda and the like) is going to push us into that direction. The 100 ms billing unit could become a benchmark. It will punish languages with a long startup time. Luckily for Java it's introducing ahead of time compilation. Many scripting languages are going to suffer. Too bad because they are usually the ones that make programming easier.

To add to that, the FFA caps the maximum battery size to 100 watt-hours, which is probably the main reason why it's limited to 16GB. I posted it here yesterday: https://news.ycombinator.com/item?id=12841293

Unless you don't need more than 16GB for the near term future. My MBP has 16gb and never seems to be near an issue for me even running kube-solo, a few vboxes, and light PS work. I would have liked the option to get 32 just to future proof since I tend to keep my mac laptops for 4-5 years but I'm willing it trade that off for the better screen, trackpad and software.

My MBP has 16gb as well and no problem whatsoever with that. But it's a 5 years old model and when I bought I had only 4gb. So 4x in 5 years. By extension I would probably need at least 2x 16gb in 3-4 years but instead I will be stuck with 16gb.

What part of your workflow are you finding increasing dramatically in memory used? I'm just a programmer so most of my stuff is pretty steady except I use a lot more virtual machines these days. But even then I can comfortably run a half dozen on my laptop with no problems and still do other work.

Factoring in the real world battery performance between those machines helps understand the Thinkpad tradeoff. One I'm not willing to take.

Thinkpads are supporting up to 64GB of ram. 4 slots, 16GB each

Yes, with the more power-hungry DDR4 modules. They take ~1/3 more power than their LPDDR4 counterparts.

Really? What about for a user who doesn't have a pressing need for 32GB?

Reply on news.ycombinator.com