Hacker News Reader

More RAM would be nice.

Would always be nice, for sure, but for Deep Learning I use a big cheap box with loads of RAM and the 16GB in my M1 MBP with the 1TB SSD seems plenty good so far. I don't have use cases yet where I notice a problem (other than yeah, overhead of having to do DL jobs on a different machine).

What causes you to hit the limits? I don't do any video editing, so I could see that being a slowdown for that use case.

Being able to locally run the integration tests for my work on my 32GB thinkpad has been a godsend during the 'rona times. If I only had 16 (or even 24) I'd have to carefully consider whether I should close Firefox to run them.

These tests spin up a bunch of containers and run certain components with instrumented runtimes that take up gobs of RAM, so having a fat laptop is very useful.

Maybe, but the RAM and disk performance is so good [1] that it might not be needed so badly. Unless you want to run a whole bunch of VMs

https://mobile.twitter.com/Sirupsen/status/13442670199449477...

I'd like to see how much gaming Apple Silicon can support. So I'm very curious to see what next year's M1X/M2 machines will bring. Of course, we're also at the dawn of widespread cloud gaming so perhaps this consideration will also be moot.

Reply on news.ycombinator.com