Flying cars
Nov. 28th, 2023 16:00![[personal profile]](https://www.dreamwidth.org/img/silk/identity/user.png)
I really do not like Elon Musk. BUT it is like saying I don't like the wind or nuclear power or gravity. He is a force and a visionary and I'll never admire him but I seem to be always seeing something he has done and saying WOW. Here is a little geeky one.
Tesla just fired up a Nvidia H100 GPU cluster to push its self-driving car development. The new cluster of 10,000 GPUs is aimed at processing the massive data from Tesla's fleet to speed up the creation of fully autonomous vehicles.
OK. They are using graphic processing units that, at a lower end, run gaming graphics instead of central processing units that run PCs and MACs for a number of reasons most of which are over my head. But the reasons mostly have to do with parallel processing, solving multiple linear algebraic problems at once and so are more capable than the general purpose cpus that computers run on.
Here is what is amazing that I CAN understand. The access to artificial intelligence we already take for granted (ChatGPT, for instance) didn't exist last October. And this month Musk is throwing a billion dollars or so at an array of processing units intending to figure out how autonomous vehicles will work next year using the access to that AI. Something that has never been done and will be a model for future civilization development.
1 - It is an avalanche of progress. For good or ill, it is already taking down the mountain.
2 - Talk about being in the right place at the right time. NVIDIA (AKA, the company that makes all those GPUs) stock is doing pretty well.
3 - Autonomous cars will be commonplace way sooner than I thought. As will a lot of other things I haven't thought of yet.
4 - Gracie does not care but would like her ear scratched, please.

The above information (not having to do with Gracie) is mostly gleaned from Daniel Miessler's 'Unsupervised Learning' email. He is way smarter than me and does understand it.
Tesla just fired up a Nvidia H100 GPU cluster to push its self-driving car development. The new cluster of 10,000 GPUs is aimed at processing the massive data from Tesla's fleet to speed up the creation of fully autonomous vehicles.
OK. They are using graphic processing units that, at a lower end, run gaming graphics instead of central processing units that run PCs and MACs for a number of reasons most of which are over my head. But the reasons mostly have to do with parallel processing, solving multiple linear algebraic problems at once and so are more capable than the general purpose cpus that computers run on.
Here is what is amazing that I CAN understand. The access to artificial intelligence we already take for granted (ChatGPT, for instance) didn't exist last October. And this month Musk is throwing a billion dollars or so at an array of processing units intending to figure out how autonomous vehicles will work next year using the access to that AI. Something that has never been done and will be a model for future civilization development.
1 - It is an avalanche of progress. For good or ill, it is already taking down the mountain.
2 - Talk about being in the right place at the right time. NVIDIA (AKA, the company that makes all those GPUs) stock is doing pretty well.
3 - Autonomous cars will be commonplace way sooner than I thought. As will a lot of other things I haven't thought of yet.
4 - Gracie does not care but would like her ear scratched, please.

The above information (not having to do with Gracie) is mostly gleaned from Daniel Miessler's 'Unsupervised Learning' email. He is way smarter than me and does understand it.