Chrome with 12 tabs open
gonna need quantum computing once you breach in to the mid 20s
Ask HN: I have 24 core server with 1TB of DDR4 RAM, what should I run?
21–30 of 40 posts
Re: Ask HN: I have 24 core server with 1TB of DDR4 RAM, what should I run?
#22Maybe you can scratch the experiment and downscale it? Seems like you are very very very far away from needing something as beefy as that. Unless you got it for very cheap, I'd feel bad having so much money in idle hardware
Re: Ask HN: I have 24 core server with 1TB of DDR4 RAM, what should I run?
#23I know that my colleague runs proxmox on this, you can do many things with it, my company is currently running proxmox and set up many vms as kubernetes nodes, I always appreciate people who set up their own lab, that colleague of mine somehow has enterprise grade storage server, nics... and many more inside his house just to have fun
Re: Ask HN: I have 24 core server with 1TB of DDR4 RAM, what should I run?
#24Doom and Far Cry are the two extremes I can think of.
Re: Ask HN: I have 24 core server with 1TB of DDR4 RAM, what should I run?
#25Maybe donate part of it to the NixOS hydra build farm? (not sure how easy that is). You can compile everything in RAM like that.
Re: Ask HN: I have 24 core server with 1TB of DDR4 RAM, what should I run?
#26crysis
> crysis
He needs a graphics card which is not yet on sale. /s
Re: Ask HN: I have 24 core server with 1TB of DDR4 RAM, what should I run?
#27Run a rails server, a DB and you'll probably not need to worry about scale until you're earning millions a year.
Re: Ask HN: I have 24 core server with 1TB of DDR4 RAM, what should I run?
#28Try llama.cpp with the biggest LLM you can find.
need a 3090 at least for that
llama.cpp and others can run purely on CPU[0]. Even production grade serving frameworks like vLLM[1].
There are a variety of other LLM inference implementations that can run on CPU as well.
[0] - https://github.com/ggerganov/llama.cpp?tab=readme-ov-file#su...
[1] - https://docs.vllm.ai/en/v0.6.1/getting_started/cpu-installat...
Re: Ask HN: I have 24 core server with 1TB of DDR4 RAM, what should I run?
#29Erlang?
Re: Ask HN: I have 24 core server with 1TB of DDR4 RAM, what should I run?
#30Erlang?
I run smallish (3k devices and hubs currently connected, about 15k total sensors) IoT "platform" on a signle 16 core, 64GB ram server with Erlang, it uses up about 3-5% of machine. Our stress testing indicated it could easily go up to 250k devices on one instance.