Breaking News
RFK Jr. advances to full Senate vote to be HHS secretary
I've Never Seen Anything Like It! (Spent The Day Reading The 'USAID' Payments Log - WTF
Sitting down? The US Mint's new 'J6 coin' will make your blood boil…
Turns out the 'Deep State' used USAID to OUST Bolsonaro from Brazil…
Top Tech News
Retro Spaceplane aces test for space station cargo missions
Old civilizations weren't destroyed by accident.
Helion has $1 billion and 3 years to figure out fusion-powered energy
Electric spacecraft propulsion may soon take a leap, thanks to new supercomputer
'Son of Concorde' supersonic jet breaks sound barrier... here's how long it'll take
Self-balancing, omnidirectional bike with balls for wheels
$120 Raspberry Pi5 Can Run 14 Billion Parameter LLM Models … Slowly
Super Sub thrills with high speed, sharp turns and steep climbs
23 airports controlled from one locale as small airfields meet the future
Affordable housing project was 3D printed in just 12 days
Tesla Dojo, Supercomputers and AI
There does not seem to be a limit for neural nets to utilize more resources to get better and faster results.
Tesla is motivated to develop bigger, faster computers that are precisely suited to their needs.
The Google TPU architecture has not evolved as much over the last 5 years. The Google TPU chip is designed for the problems that Google runs. They are not optimized for training AI.
Tesla has rethought the problem of AI training and designed the Dojo AI supercomputer to optimally solve their problems.
If Tesla commercializes the AI supercomputer that will help to get to lower costs and greater power with more economies of scale.
One of the reasons that TSMC overtook Intel was that TSMC was making most of the ARM chips for cellphones. TSMC having more volume let them learn faster and drive down costs and accelerate technology.
99% of what neural network nodes do are 8 by 8 matrix multiply and 1% that is more like a general computer. Tesla created a superscalar GPU to optimize for this compute load.