Internet is not reliable. Pls, we are talking about PRO work.

> What do people need with 32 GB of RAM on a dev laptop?

Databases. I run Sql Server, PostgreSQL, MySql, Firebird. All at the same time? Yes.

Also, the thing about "Big Data" is need as much RAM as possible.

---

I could live with a iMac (what I have) or a MacPro (that I can't buy because the dolar in my country double in price!) but the main issue here is the signaling. Apple is like telling "We don't care anymore about you".

How will be trust the (maybe next?) upgrades to the rest of the line will be almost good now? And what about the next after this ones?

I work from home 2-3 days a week so good internet is key - especially since my main machines need to be domain joined so RDP is ideal.

Just trying to understand what your scenario is.

Also the latency make some things slower.

Either way - fair enough. You don't like the reliability of your internet so the power of the machine locally is important.

<sigh> Because when you demo, being tied to a box on the other side of the planet is asking for trouble.

I really can't believe the "why do you need that much RAM?" question still gets asked.

I record my demos now as I don't care for live demos but that's just my way of dealing with it.

I'm not contesting people's needs, I'm curious what they are.

It's more realizing the promise of hibernate/suspend options- if you don't have to reboot or stop what you are working on to do something else, why would you? And when it's in a portable laptop you can just take it anywhere, where a big box may be unavailable/down. Not to mention the TCO of a big box versus a laptop in electricity and everyday costs is actually much lower and comes with free battery backup, embedded keyboard and screen! I repurpose many of my old laptops to be servers/development support for that reason and I want my systems to be expandable, because RAM is what I really want to keep things speedy.

Things that need a lot of RAM:

Solidworks, Fusion360 (I can't even run it on 8bg with and an integrated gpu) And using Solidworks remotely is a bad joke.

Robot Path planning, motion planning

AI Assignments,

ROS + RVIZ + Gazebo + actual robot software

Anything written in python

[EDIT almost forgot] Matlab, VMs

Playing music or having a web browser open while any of the above things are running...

16bg would probably be fine for 2 years. But >$2000 is a lot to pay for only 2 years of use. People seem to think that just because its no longer current_year-1 that their desktop app can use previous_version's_RAM+0.5gb

There are enough people with enough reasons for having only a laptop. The dell precision m5510 is around same price as mbp (depending on setup) and yet it comes with a real graphics card, and the option for 32 gb of RAM.

[0]: https://www.macintouch.com/forums/showthread.php?tid=1046 (Spotify is killing SSDs)

My usage is obviously not like everyones: lots of docker container and vagrant vms both with dedicated memory, unhealthy amount of browser tabs, memory hogging intellij (especially with scalaz implicits and having lots of projects open at the same time). And the biggest memory killer: working for clients with very large ecosphere stack of JVM based microservices that you spin most of up locally including dbs.

Whenever I pair with devs using 8gb MBAs on these projects they continually apologises for the MBA's slowness and their need to continually shutdown part of microservices stack as they run into 8gb constraints.

16gb is a sweet spot today, but not in 2 or less years.

Remoting into private/cloud vms or integrating with staging environments is possible sometimes, but I work on the train for 2 hours a day as well as from home a lot, and in demos, meetings, coffee shops. So relying on dodgy VPNs over fragile connections is painful. So spinning up the whole stack locally has always been a must for me. Again, this is not most people's pattern.

( Though spinning up microservices not just locally is a frustration why I in anger started https://github.com/flurdy/philharmonic )

And being reliant on connectivity to a "big box" to simply get your work done is suboptimal. Also, how big of box do you need when you've got hundreds if not thousands of devs? Much easier (and cheaper, since there's no upkeep) to throw another 16GB of memory inside every developer's system.

The message I'm getting from folks is that if you do contract/client work, you want everything local for many reasons.

In my case, I work for a Large Software Company so my scenario is more constrained.

:)