If you do anything that's heavy on I/O (packaging jar files in my case), this laptop is faster than my big desktop, the I/O is crazy fast actually.
Memory pressure has been low so far - my biggest fear opting for the base model but looks like i'll be fine - i spun up bunch of dev processes and was fine.
I seem to remember watching a real time video transcoding demo, or something like that, on an iPad Pro and the dev demoing it saying it was the I/O that made it possible. Desktop machines with much faster processors would choke due to bandwidth limitations, but Apple can custom tune every aspect of the iPad architecture and aren't limited to standard, generic interfaces between components.
The original 5k iMacs were a similar case in some ways. It was only possible in an all-in-one design, because there wasn't a standard video connection at the time that was capable of the throughput.
Tt's got a rather fast NVME SSD, 2.7GB/s reads and 2.2GB/s writes (Intel version was 1.3 and 1). Don't know if the IOPS & latency improved, but it wouldn't surprise me much.
The permanently soldered in SSD that has a finite amount of writes.
Saving $200 on RAM up front could get wiped out if you kill your SSD prematurely with constant, heavy swapping.
Here I see the fast SSD speed as a negative - making it painless to do something inherently dangerous towards the lifespan of your SSD.
The SSDs from the previous gen are rated for something like 40Gb writes per day for 40 years i think.
I’d guess the battery will fail first, then the keyboard or screen hinge, then the logic board. But thats a pure guess.
I reckon the biggest risk is that all in one logic board, anything fails and you lose your HDD.
Obsolete, who defines that?
I'm typing this on a MBP 2015. Still works perfectly fine (I did have screen replaced under warranty). Battery is replaceable on this machine, its sitting at ~85% capacity with 500 cycles. Apple hardware tends to have good resale value, too.
Performance-wise, hardware tends to last longer and longer. Repairability-wise, yeah, that's a can of worms of a discussion.
I had the same on Windows + a SSD. Docker with some Sqlite applictions apparently did crazy writes. SSD went from 100% to 80% in like half a year time or so, when I noticed it. That SSD is still alive, but I had to adapt.
Every additional 'Level' in the flash decreases the overall life cycle. Mind you, oftentimes drive controllers are fancier, often doing things like treating one of the sections as a less dense flash type to improve write cycles, and keep/move longer held data into the regions treated as QLC.
But, that means that some heavy write patters will indeed wind up worse on a modern drive than the better MLC/TLC models.
By the time I left they were shipping appliances with SSDs that were predicted to have lifespans longer than the longest hardware support you could buy from the vendor (longest hardware support contract - seven years). Hundreds of thousands of writes per second, every second, for 7+ years, uninterrupted.
Surely, they won't keep using the same aging technology. At least I hope not.