https://rosmine.ai/2026/05/13/was-my-48k-gpu-worth-it/
-
https://rosmine.ai/2026/05/13/was-my-48k-gpu-worth-it/
"help I quit my FAANG job and then spent 50k USD on GPUs someone help me with finances my family is starving"
-
https://rosmine.ai/2026/05/13/was-my-48k-gpu-worth-it/
"help I quit my FAANG job and then spent 50k USD on GPUs someone help me with finances my family is starving"
> Because of this I got a motherboard with slow GPU interconnect. It’s good for running many small experiments in parallel (which is my main use case) but horrible for any models split across gpus.
You spent $50k and your thing can't even actually make proper use of the GPUs you spent an approximate average US person's salary on?
-
https://rosmine.ai/2026/05/13/was-my-48k-gpu-worth-it/
"help I quit my FAANG job and then spent 50k USD on GPUs someone help me with finances my family is starving"
@cwebber Would you send a DM?
-
https://rosmine.ai/2026/05/13/was-my-48k-gpu-worth-it/
"help I quit my FAANG job and then spent 50k USD on GPUs someone help me with finances my family is starving"
@cwebber The last GPU I bought was a GTX 1060 back in 2017. How many AIs can I run on that?
-
> Because of this I got a motherboard with slow GPU interconnect. It’s good for running many small experiments in parallel (which is my main use case) but horrible for any models split across gpus.
You spent $50k and your thing can't even actually make proper use of the GPUs you spent an approximate average US person's salary on?
@cwebber “Ironically, after designing the entire build around apartment power constraints, I ended up moving grumbl to my parents’ basement, where I could upgrade the circuits anyway.”
Between this dumbass and the one that sunk his cyber truck in a lake in purpose to test its wade mode, I’m convinced God gives too much money to utter jackasses and I don’t know why.
-
> Because of this I got a motherboard with slow GPU interconnect. It’s good for running many small experiments in parallel (which is my main use case) but horrible for any models split across gpus.
You spent $50k and your thing can't even actually make proper use of the GPUs you spent an approximate average US person's salary on?
@cwebber lol: “… I ended up moving grumbl to my parents’ basement, where I could upgrade the circuits anyway.”
-
> Because of this I got a motherboard with slow GPU interconnect. It’s good for running many small experiments in parallel (which is my main use case) but horrible for any models split across gpus.
You spent $50k and your thing can't even actually make proper use of the GPUs you spent an approximate average US person's salary on?
@cwebber@social.coop the motherboard isn't really the expensive part is it?
Source: person who buys the cheapest motherboard possible for his builds
-
R relay@relay.infosec.exchange shared this topic
