6,867 karma · joined October 19, 2014
* https://danuker.go.ro/ - My blog about software, economy, and politics
* https://github.com/danuker - My GitHub
* https://www.sftpplus.com/ - My employer: build bridges between your clouds with SFTPPlus!
I approve of this, but in your place I'd wait for hardware to become cheaper when the bubble blows over. I have a i9-10900, and bought an M.2 SSD and 64GB of RAM in july for it, and get useful results with Qwen3-30B-A3B (some 4-bit quant from unsloth running on llama.cpp).
It's much slower than an online service (~5-10 t/s), and lower quality, but it still offers me value for my use cases (many small prototypes and tests).
In the mean time, check out LLM service prices on https://artificialanalysis.ai/ Open source ones are cheap! Lower on the homepage there's a Cost Efficiency section with a Cost vs Intelligence chart.
https://blog.jetbrains.com/appcode/2022/12/appcode-2022-3-re...
Don't even go close:
https://www.theguardian.com/world/2021/may/23/belarus-divert...
What projects need manual memory management? Those where the hardware costs are comparable to development/maintenance costs. That is much rarer than people think.
RAM is cheap, and few applications really need bespoke allocation. And it's not just a question of skill; even the most disciplined make mistakes. It's one of how much brainpower you want to allocate to... memory allocation.
Thunderfoot showed how misleading it was: https://www.youtube.com/watch?v=SsdLDFtbdrA
Why would I buy a separate piece of hardware taking up more space in my home and life? More money down the drain.
Sure, it's nice that it "just works", but so does `pacman -Sy xonotic`.
Still useful for taking advantage of employees not in the know.
https://totalrealreturns.com/n/VFINX,VBMFX,USDOLLAR?start=20...