3 ms·

I understand Nvidia is no angel. And I'm totally OK with them doing the right thing for the wrong reason. Also, my concern is not for data centers running cree

I understand Nvidia is no angel. And I'm totally OK with them doing the right thing for the wrong reason. Also, my concern is not for data centers running creepy ML models that invade users' privacy, but for the average home user who wants to get started with ML and at the same time get decent frame rates when they play a game.

Tenoke 6y ago
>I understand Nvidia is no angel. And I'm totally OK with them doing the right thing for the wrong reason. My argument is they should and could have scaled up the production rather than just shuffling around which group they hurt or help at the expense of which other (be it miners, ML folks or gamers). This isn't the 'right thing', it just helps the group you prefer on this round.
- tomatotomato37 6y ago
  You do understand Nvidia is a fabless company, correct? Scaling up demand isn't just calling up some factory manager and telling them to work harder, it's getting into a massive cagefight with AMD, Apple, and every other company out there that needs silicon all for the amusement of Samsung/TSMC
- belltaco 6y ago
  https://www.networkworld.com/article/3602456/youre-not-imagining-things-there-is-a-serious-chip-shortage.html https://www.networkworld.com/article/3602456/youre-not-imagi...
- PixelOfDeath 6y ago
  Ask AMD why scaling up to mining demand is a bad idea. A sudden implosion of demand combined with a flooding of the used GPU marked. Better not plan on bringing out any new GPU architectures when your warehouses are still full of the previous generation.
  - Tenoke 6y ago
    Except with NVIDIA there's already extra demand if they need it - they can allow GeForce into data centers which are currently artificially disallowed.
PragmaticPulp 6y ago
> but for the average home user who wants to get started with ML and at the same time get decent frame rates when they play a game. Getting started with ML doesn't require a GPU at all. Realistically, any video card or even a modern CPU is more than sufficient to get started with ML learning. It's a myth that $1000 gaming video cards are necessary or even helpful for learning basic ML.

Tenoke 6y ago

>I understand Nvidia is no angel. And I'm totally OK with them doing the right thing for the wrong reason. My argument is they should and could have scaled up the production rather than just shuffling around which group they hurt or help at the expense of which other (be it miners, ML folks or gamers). This isn't the 'right thing', it just helps the group you prefer on this round.

tomatotomato37 6y ago
You do understand Nvidia is a fabless company, correct? Scaling up demand isn't just calling up some factory manager and telling them to work harder, it's getting into a massive cagefight with AMD, Apple, and every other company out there that needs silicon all for the amusement of Samsung/TSMC
belltaco 6y ago
https://www.networkworld.com/article/3602456/youre-not-imagining-things-there-is-a-serious-chip-shortage.html https://www.networkworld.com/article/3602456/youre-not-imagi...
PixelOfDeath 6y ago
Ask AMD why scaling up to mining demand is a bad idea. A sudden implosion of demand combined with a flooding of the used GPU marked. Better not plan on bringing out any new GPU architectures when your warehouses are still full of the previous generation.
- Tenoke 6y ago
  Except with NVIDIA there's already extra demand if they need it - they can allow GeForce into data centers which are currently artificially disallowed.

PragmaticPulp 6y ago

> but for the average home user who wants to get started with ML and at the same time get decent frame rates when they play a game. Getting started with ML doesn't require a GPU at all. Realistically, any video card or even a modern CPU is more than sufficient to get started with ML learning. It's a myth that $1000 gaming video cards are necessary or even helpful for learning basic ML.