I have a 5080 and 128gb of ram running on a AMD 9950X.

Depending on the task I can get over 170-200t/s when the MOE only calls a few agents and can fit inside the VRAM or as low as 5-10ts when it calls more agents and has to hit the system memory. But for grunt work that doesn’t need professor level tasks, it’s more than capable and if you have the time, it’s super worth it because it’s basically free tokens.

I only use this for overnight work to save on tokens during the day. When I’m pulling analytics for my work and it just needs basic analysis that doesn’t touch multiple tooks.

During work hours I’m using GLM5.2 for web development, Kimi k2.7 for complicated data analysis and Minimax m3 if I need the context window to be bigger than what kimik2.7 can give me.

I’m running dwarfstar which is a 2 bit deepseek v4 flash. It’s quite capable even at 2 bit.

Ya, I think your industry is safe.

Agreed. The entire industry engineer/developer industry is fucked.

While I can afford a newer car, I can’t bring myself to spend 700usd just to drive a new car when my old car is still running. Even if my entire engine explodes, a new engine replacement is like 5k, it’s still worth it. Maybe one day my engine won’t be available but for now, I’m happy with not paying a car payment.

Not mine. My car was built in 1999. Going to drive it until they stopped offering fuel at gas stations and then just transplant a electric drivetrain.

My car has a cassette tape and no Bluetooth.

I had the same issue. I didn’t want to depend on meds my entire life. But honestly. After taking it regularly, I’m a better husband and a even better father. I am truly not following my father’s footsteps.

I will say that I loved my Windows Phone. Way ahead of it’s time.

I’m a dev at work with 1 dev Jr dev under me.

We are in desperate need of another jr dev. But my work rather spend tokens.

So I setup agentic workflows which works 90% of the time. The 10% requires a me, a senior developer to babysit the workflow. My hourly cost are significantly higher.

After calculating the token usage and my hourly rate there are still some savings using the agentic workflows. But not much.

But we end up sacrificing is a pool of junior developers that I can train to be a senior. If I ever leave the company the entire company is screwed.

Fable is so expensive. I made a Blinko notes desktop app on Opencode last night using Minimax M3 and it would have cost me 3000usd if I was using Fable.

I actually switched it to a LXC and it significantly faster than running it on a VM.

I’m running immich with tailscale.

I would love to do this. I’m great at changing out motorcycle tires and changing out brake pads, but I would love to learn how to solder or even diagnose bad diodes on a board.

It would be a great exchange of knowledge.

I have too much data to have it on the proxmox. The source of truth is on the Synology and I have the proxmox running the immich database and thumbnails.

Xeon E3-1245 v5

32gb ddr4

1tb NVME gen 3

I did this. I kept both.

I have Synology photos and immich.

Synology photo is faster to load photos when you scroll to a random date. Also on mobile, the map view is fast and reliable. Also the Android widget can show images of a specific person instead of a random photo.

Immich has better face detection and significantly better search tool. But is super laggy when you’re looking for a specific date and time and location.

My Synology is a DS 1821+ with dual nvme cache and 32gb mem My immich is running on a proxmox using a late model Xeon processor and NVME gen 3 and 32gb mem. Nothing else is running on the proxmox.

I still have my edu email and it does a verification for class enrollment.

This is literally a few cars on the road. How about we don’t stop investing into electrification and we won’t have this problem in the future.

I manage a web team for a pretty big company. It’s just me and a Jr dev. Even with AI, we still can’t keep up.

The reason for their labor shortages is that they make it so hard to immigrate into their country.