I guess RAM is a bell curve now.
- 32GB: Enough.
- 16GB: Not enough.
- 8GB: Not enough.
- 4GB: Believe it or not, enough.
I actually audibly laughed when Raspberry Pi came out with an 8GB version because for anyone who thinks 4GB isn’t enough probably won’t be happy with 8 either.
zram to the rescue
The only time I can remember 16 GB not being sufficient for me is when I tried to run an LLM that required a tad more than 11 GB and I had just under 11 GB of memory available due to the other applications that were running.
I guess my usage is relatively lightweight. A browser with a maximum of about 100 open tabs, a terminal, a couple of other applications (some of them electron based) and sometimes a VM that I allocate maybe 4 GB to or something. And the occasional Age of Empires II DE, which even runs fine on my other laptop from 2016 with 16 GB of RAM in it. I still ordered 32 GB so I can play around with local LLMs a bit more.
Yeah, but if you’re interested in running an LLM faster than 1 token per minute, RAM won’t matter. You’ll need as much VRAM as you can get.
Sure, but I’m just playing around with small quantized models on my laptop with integrated graphics and the RAM was insanely cheap. It just interests me what LLMs are capable of that can be run on such hardware. For example, llama 3.2 3B only needs about 3.5 GB of RAM, runs at about 10 tokens per second and while it’s in no way comparable to the LLMs that I use for my day to day tasks, it doesn’t seem to be that bad. Llama 3.1 8B runs at about half that speed, which is a bit slow, but still bearable. Anything bigger than that is too slow to be useful, but still interesting to try for comparison.
I’ve got an old desktop with a pretty decent GPU in it with 24 GB of VRAM, but it’s collecting dust. It’s noisy and power hungry (older generation dual socket Intel Xeon) and still incapable of running large LLMs without additional GPUs. Even if it were capable, I wouldn’t want it to be turned on all the time due to the noise and heat in my home office, so I’ve not even tried running anything on it yet.
Apple: Enough!
But you’ll have to buy a whole new laptop when it turns out that was a lie.
💪 32GB RAM
💵 64GB RAM
128 GB here which runs out if I compile the complete project at work with -j32. And this sucks because 128 GB right now means the RAM cannot run super fast, meaning it is a bottleneck to any modern Ryzen…
256GiB here, i sometimes need to run chrome
I use separate hdd solely as swap for these cases, used 500gb laptop hdd is good enough if you turn zswap zstd compression on, with recent kernels using mglru it works splendidly
I only run out of ram (16GB) when I’m playing minecraft with 280 mods
Have you even used Linux? 16GB of RAM is enough, even with electron apps
We used to say 4GB is enough. And before that, a couple hundred MB. I’m staying ahead from now on, so I threw in 64GB. That oughtta last me for another 3/4 of a decade. I’m tired of doing the upgrade race for 30 years and want to be set for a while.
I can literally trace my current Ryzen PC’s lineage like the ship of Theseus to an Athlon system I built in 2002. A replacement GPU here. Replacement mobo there. CPU here, etc.
Not in my experience. The electron spotify app + electron discord app + games was too much. Replacing electron with dedicated FF instances worked tho.
I have a browser tab addiction problem, and I often run both LibreWolf and Firefox at the same time (reasons). I run discord all the time, signal, have a VTT going on, a game, YouTube playing… and I look at my RAM usage and wonder why did I buy so much when I can never reach 16 GB.
While I agree electron apps suck and I avoid them… Whatever you guys are running ain’t a typical use case.
About 6 months ago I upgraded my desktop from 16 to 48 gigs cause there were a few times I felt like I needed a bigger tmpfs.
Anyway, the other day I set up a simulation of this cluster I’m configuring, just kept piling up virtual machines without looking cause I knew I had all the ram I could need for them. Eventually I got curious and checked my usage, I had just only reached 16 gigs.I think basically the only time I use more that the 16 gigs I had is when I fire up my GPU passthrough windows VM that I use for games, which isn’t your typical usage.
16 MiB is enough depending on what you’re running.
It used to be that 640K oughta be enough for anyone.
deleted by creator
lightweight usage
That doesn’t mean anything. If you have tons of free RAM, programs tend to use more than strictly necessary because it speeds things up. That doesn’t mean they won’t run perfectly fine with 8GiB as well.
deleted by creator
What distro are you using? What apps are open?
What does
free -hsay?[ugjka@ugjka Music.Videos]$ free -h total used free shared buff/cache available Mem: 29Gi 17Gi 1,8Gi 529Mi 11Gi 11Gi Swap: 14Gi 2,0Gi 12GiI was wondering if your tool was displaying cache as usage, but I guess not. Not sure what you have running that’s consuming that much.
I mentioned this in another comment, but I’m currently running a simulation of a whole proxmox cluster with nodes, storage servers, switches and even a windows client machine active. I’m running that all on gnome with Firefox and discord open and this is my usage
$ free -h total used free shared buff/cache available Mem: 46Gi 16Gi 9.1Gi 168Mi 22Gi 30Gi Swap: 3.8Gi 0B 3.8GiOf course discord is inside Firefox, so that helps, but still…
about 3gigs goes to vram as i don’t have dedicated card yet, but i’m getting 16 gig dedicated gfx soon
16gb and a number less than 16gb both not being big enough numbers is making me crack up
I don’t know. I’m running 16gb with 8gb of swap just fine.
Couple dozen tabs open in librewolf (across multiple windows), android studio with an emulator and some other utils. All under KDE Plasma on nixos unstable and it’s fine. It could be better, but it’s good enough.
I was happy with 16GB until i inherited a huge Angular legacy project.
I use 8gb ram and it worked for everything except running stable diffusion yet
deleted by creator
