• 0 Posts
  • 198 Comments
Joined 2 years ago
cake
Cake day: June 22nd, 2023

help-circle







  • I think it’s different. The fundamental operation of all these models is multiplying big matrices of numbers together. GPUs are already optimised for this. Crypto was trying to make the algorithm fit the GPU rather than it being a natural fit.

    With FPGAs you take a 10x loss in clock speed but can have precisely the algorithm you want. ASICs then give you the clock speed back.

    GPUs are already ASICS that implement the ideal operation for ML/AI, so FPGAs would be a backwards step.






  • They’re different, and I think this one has the capability of being more devastating.

    The dot-com bubble was really broad. Hundreds or thousands of companies, all without vowels in their names trying to break new ground. A wild west style gold rush. When it popped a lot of small companies went bankrupt.

    This is a handful of companies with billions of capital buying GPUs from NVidia to be make the largest hungriest machine they can. All in the pursuit of being first to create “AGI”. If one of them succeeds, the others are toast and multiple 500+B dollar companies will collapse in on themselves. If none of it works, the same thing happens and it takes a large chunk out of $4T Nvidia too.