Breaking News
Russia warns NATO of a possible nuclear response if Kaliningrad exclave is cut off | DW News
Now They're REALLY Mad
More Americans are noticing the taste of their favorite snacks keep changing
Eva: France is on fire. Hundreds of schools all over France are being besieged by second...
Top Tech News
Med Beds Just Took a MASSIVE Step Forward, Healing People in Days!
Palmer Luckey: Autonomous Weapons Are Ancient and Why Anduril Won't Build Humanoids | EP #295
A laser just photographed objects through six feet of concrete
Elon Musk's Next-Gen Motor Destroy Entire EV Industry
China's disputed satellite refueling heralds new space war era
BYD Will Put Solid-State Batteries In An EV Next Year: Executive
FDA-cleared exoskeleton puts spinal-cord patients back on their feet
Your Robotic Vacuum Is Watching You -- Could It Someday Testify Against You in Court?
Why Unigrid's Sodium-Ion Batteries Are the Game-Changer for Off-Grid Energy Storage
This Battery On Wheels Makes Any Diesel Truck Electric In 5 Minutes
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.