Not a bad idea. What's your budget? That might be a large factor.

It's just me that uses this one, I have a separate 220J that my Mum should be using to store photos on but doesn't.

I have to say nothing beats the convenience and reliability of Synology yet... the complain is weak CPU and Synology seems to deprecating a lot of features lately.

Kinda looking for a storage vps in the APAC region for my own storage use replacing gdrive as it's getting expensive if store over 200gb.

As others have said, inference on a CPU will be really, really slow. So there's not really much point in doing it without a GPU. The best solution where you don't have to buy a GPU is just renting one

(Quote)

(Quote)

(Quote)

LM Studio is the easiest to deploy platform imo, try it and let me know the real results, I am curious.

(Quote)

The inference speed is mainly dependend on memory bandwidth, Dual rtx 3090 runs 70B model around 15-20 tokens/second. There is some speed loss due to dual setup, yet DDR 2400 bandwidth is about 50xtim

(Quote)

(Quote)

You have severe packet loss on your local network which correlates the performance. Check it outside of your own network.

(Quote)

For a self hosted LLM you need a lot of VRAM especially for coding. For home PC probably the best model you can run is currently Qwen2.5-32B-Coder . For coding, unlike chat you need at least half(8 bi

I need to setup virtual hosts files for docker containers and proxmox VMs sharing the same public IP address and the standard public ports 80 and 443.

Thank you for your understood.

Agree with this, but the moment the business decides to offer transfers as a service (with a fee), they’ve accepted the administrative burden that such transfers might entail.

The provider does not permitted pushing at first, but later they open the push function for 5 dollar once. And provider agreed the refund to me (more details I have said above). The headache thing for