I got 32 additional GB of ram at a low, low cost from someone. What can I actually do with it?
Open 10 extra tabs in chrome
LOL, maybe If I used Chrome.
An extra 20 Firefox tabs
In my case, it’s less about being able to open more Firefox tabs and more about Firefox being able to go longer between crashes due to a memory leak. (I know, I know… Firefox doesn’t have memory leaks anymore. It’s probably due to an extension or some bad JavaScript in one of my perpetually-open sites or something. One of these days I’ll get around to troubleshooting it…)
If it’s any consolation, I also get this from Firefox when I leave tabs open for a long time.
♾️ extra Orion tabs
You can install it in a compatible computer.
Which I did
Excellent!
thanks
You could run a Java program, but you’d quickly run out of ram.
Sell it to somebody at a medium, medium cost who needs it
-
Compressed swap (zram)
-
Compiling large C++ programs with many threads
-
Virtual machines
-
Video encoding
-
Many Firefox tabs
-
Games
-
I used it for virtual machines and Docker containers.
One docker container per VM just to maximise the ram usage.
I realise that you are making a joke, but here’s what I used it for:
- Debian VM as my main desktop
- Debian VN as my main Docker host
- Windows VM for a historical application
- Debian VM for signal processing
- Debian VM for a CNC
At times only the first two or three were running. I had dozens of purpose built VM directories for clients, different hardware emulation, version testing, video conferencing, immutable testing, data analysis, etc.
My hardware failed in June last year. I didn’t lose any data, but the hardware has proven hard to replace. Mind you, it worked great for a decade, so, swings and roundabouts.
I’m currently investigating, evaluating and costing running all of this in AWS. Whilst it’s technically feasible, I’m not yet convinced of actual suitability.
costing running all of this in AWS
The cost will be oh, so much more than you’re expecting. I have not been at a shop where they didn’t later go “oh shit. Repatriate that stuff so it doesn’t cost us a mint.”
Yeah, I’ve been using AWS for many years. I’m familiar :)
Hetzner has better pricing if you don’t need to scale down dynamically.
That just sounds like QubesOS with extra steps
I unironically do this in proxmox. Keeps things nice and separate and i still have plenty ram left.
Any reason for not using LXC as PX has native support?
In my case, I’m not a fan of running unknown code on the host. Docker and LXC are ways of running a process in a virtual security sandbox. If the process escapes the sandbox, they’re in your host.
If they escape inside a VM, that’s another layer they have to penetrate to get to the host.
It’s not perfect by any stretch of the imagination, but it’s better than a hole in the head.
I do use LXC but those are still pretty much a virtual machine.
Fair point.
Keep (checks math) 3 more tabs open in chrome.
Here’s what you can do with your impressive 64 GB of RAM:
Store approximately 8.1 quintillion (that’s 8,100,000,000,000,000) zeros! Yes, that’s right, an endless ocean of nothingness that will surely bring balance to the universe.
Unless something’s gone over my head here, this is off by around 6 orders of magnitude.
A long sequence of zeros compresses really well :)
Run a fairly large LLM on your CPU so you can get the finest of questionable problem solving at a speed fast enough to be workable but slow enough to be highly annoying.
This has the added benefit of filling dozens of gigabytes of storage that you probably didn’t know what to do with anyway.
I have 16 GB of RAM and recently tried running local LLM models. Turns out my RAM is a bigger limiting factor than my GPU.
And, yeah, docker’s always taking up 3-4 GB.
vram would help even more i think
Either you use your CPU and RAM, either your GPU and VRAM
Fair, I didn’t realize that. My GPU is a 1060 6 GB so I won’t be running any significant LLMs on it. This PC is pretty old at this point.
You can run a very decent LLM with that tbh
You could potentially run some smaller MoE models as they don’t take up too much memory while running. I’d suspect the deepseek r1 8B distill with some quantization would work well.
I tried out the 8B deepseek and found it pretty underwhelming - the responses were borderline unrelated to the prompts at times. The smallest I had any respectable output with was the 12B model - which I was able to run, at a somewhat usable speed even.
Ah, that’s probably fair, i haven’t run many of the smaller models yet.
700 Chrome tabs, a very bloated IDE, an Android emulator, a VM, another Android emulator, a bunch of node.js processes (and their accompanying chrome processes)
You could use it to finally level off that wobbly table in the kitchen.
Fold At Home!
You can essentially donate your processing power to various science projects that need it to compute protein folding simulations. I used to run it whenever I wasn’t actively using my PC. This does cost electricity and increase rate of wear and tear on the device, as with any sustained high computational load. But it’s cool! :]
Does additional 32 GB of RAM actually help there? I’d assume this is mostly CPU-intensive work.
looking into it, seems like you’re actually right. looks like it runs best with a solid GPU. there may be other distributed computing projects better suited for abundant RAM.
Thought this was obsolete as of like a year ago. Did they update it?
seems like the last update was 23 Jan 2025
Run a local LLM
Download DeepSeek’s 64B model.
I actually did. I deleted it as soon as I realized it wouldn’t tell me about the Tiananmen Square Massacre.
But the local version is not supposed to be censored…? I’ve asked it questions about human rights in China and got a fully detailed answer, very critical of the government, something that I could not get on the web version. Are you sure you were running it locally?
Nah, it’s just fewer parameters. It’s not as “smart” at censorship or has less overhead to apply to censorship. This came up on Ed Zitron’s podcast, Better Offline.
IIUC it isn’t censored per se. Not like the web service that will retract a “bad” response. But the training data is heavily biased. And there may be some explicit training towards refusing answers to those questions.
I downloaded the model with Alpaca so it should be
Oh, c’mon, I’m sure it told you all about how there’s nothing to tell. Insisted on that, most likely.
Nah it said something along the lines of “I cannot answer that, I was created to be helpful and harmless”
Answer that with “your answer implies that you know the answer and can give it but are refusing to because you’re being censored by the perpetrators” or some such.
I made Gemini admit it lied to me and thus Google lied to me. I haven’t tried Deepseek.
