Hacker News Reader

Comment by blux | Hacker News Reader

Lots of comments from people complaining about the lack of 32GB of RAM. Can someone explain to me the need for 32GB of RAM as a developer? Is it the IDE or your particular application area? As for me, I do all my professional development (C++17/Linux, Vim, EDA industry) on a five year old T450s with 12GB of RAM. For me the limiting factor is CPU power for sure. I only run out of RAM running some of the larger customer tests for which I have to go to a beefy machine in our network, but that happens rarely.

View on HN

I can easily make good use of 32 GB doing normal full stack development: Browser, Visual Studio Code, a few Docker containers including PostgreSQL, Slack.

Actually this should consume a dozen GB of RAM more or less, the rest is in caches, which I think you're undervaluing a lot. Having everything you use in hot caches means any applications starts in a couple hundred of milliseconds, and everything is as fast as your RAM and CPU are, especially if you're running an OS that has a good implementation of file caching (e.g. Linux).

Having more leeway means I never have to close anything if I forget to. I might have my dev environment chugging along in a desktop workspace, while I'm in another with a few dozen browser tabs, talking on Discord with a full screen game running on the other monitor.

That's what a £150 module of 32 GB of RAM gives me. But actually I'm running 64 GB :-)

I do CAD.

Both SolidWorks and Altium are such RAM hogs that I’ve had to upgrade to 64GB.

It would be lovely to work on laptops, since it’s the default assumption for most managers, so I either stick out as a weirdo for asking for a desktop every time I switch jobs, or shut up and spend most of my day dealing with slow laptops and random crashes, and having to make design choices around my computer (unnecessary) limitations.

Chrome (~1-4Gb), Slack with quite a few orgs (~2Gb), Discord again with quite a few orgs (~1Gb), Docker for local dev work (~1-8Gb), Gopls (golang's language server) (~1Gb per open vscode).

Running out of RAM is not my main problem, continuously swapping is as has a performance hit across the whole system for some reason.

As with many things, it's not a NEED, it's more that having to care about what I have open at any given moment gets annoying.

> Chrome (~1-4Gb), Slack with quite a few orgs (~2Gb), Discord again with quite a few orgs (~1Gb)

Those ravenous apps are the problem here. Seriously, fuck Electron.

I know Android Studio will basically eat up however much RAM is on your system. Also, with containers becoming more common, lots of people use something like Docker Compose locally to mock multiple services or parts of a system, which, depending on how well optimized those containers are, can add up quickly.

Had a t470s (24Gb), now on a p1 gen 2 (32Gb). Even 24Gb can feel limited when running kubernetes, number of docker images, oracle rdbms, kafka, java development environment, a browser, office suite...

4 gb for Chrome tabs, 6 gb for your docker WM, 2gb or whatever slack requires, 2gb because your music is now streamed and the app is just a few html pages rendered through Chrome. Plus a couple of gb for your IntelliJ IDE.

If you are writing C++ in Vim for an embedded platform, you are limited by CPU power because C is so heavy to compile, but you are most likely also not likely to be the typical HNer.

I asked the same question. I get why people need more than 16GB of RAM, but honestly a lot of the answers seem to neglect that RAM (in most laptops) is not going to be your limiter, the CPU and/or GPU will be the limiter first for many of the cases people list. At that point, no amount of RAM will save you (though it may help).

Any applications with image processing are usually bottlenecked by the available RAM on a system. If you're running any ML applications, it's most efficient to run models on batches of images and batch size is determined by RAM.

I can see people needing more RAM, but I think people are forgetting that these are just the first few models Apple are bring to the market. If Apple truly believe that no one needs more, they could have dropped more of the Intel models.

I imagine that it's someone that runs either either a lot of virtual machines or a lot of Docker Containers.

As a developer, I routinely need over 128GB. 32GB is cutting is short.

why aren't you using a server?

docker, JVM VMs, IDEs that take a couple gigs. You can always use it.

Reply on news.ycombinator.com