Running AI Models Locally

[Elefant&Castle]

Spec says 170W ... where did you get 1400 from?

https://www.nvidia.com/en-us/product...ons/dgx-spark/

That's the litte one - still 3x more than a mac mini.

The desktop system with the GB300 is 1400W

https://www.techpowerup.com/330154/n...ce=chatgpt.com

[shri]

This could get interesting if you have a right gear.

https://youtu.be/zQi-8mCTNf8?si=7g8PIiuHmOchAeFr

[Elefant&Castle]

This could get interesting if you have a right gear.

What gear is that? Youtuber Alex Ziskind often does videos testing gear -did one recently compareing a Mac studio with a lot of RAM to a NVIDIA GPU. Did quite well - much cheaper, draws less power.

[shri]

What gear is that? Youtuber Alex Ziskind often does videos testing gear -did one recently compareing a Mac studio with a lot of RAM to a NVIDIA GPU. Did quite well - much cheaper, draws less power.

I have one of the new Mac mini with 64Gb ram.. runs a lot of the lager models which are optimized for apple / quantized appropriately well enough.

[shri]

Larger ... not lager models.