George Hotz | Exploring | sniffing NVIDIA's ioctls | open-gpu-kernel-modules | DEBUG | PTX | CUDA 

Creation of chapters is in progress. Lots of streams one after another so it's hard to keep up. Takes a while to create all the chapters. github.com/NVIDIA/open-gpu-kernel-modules | github.com/geohot/cuda_ioctl_sniffer Bounties for tiny corp / tinygrad -> docs.google.com/spreadsheets/d/1WKHbT-7KOgjEawq5h5Ic1qUWzpfAzuD_J06N1JwOCGs/ ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-lnVQsJJFcdg.html Hiring entire stack for tiny corp join if you are interested | ru-vid.com/video/%D0%B2%D0%B8%D0%B4%D0%B5%D0%BE-lnVQsJJFcdg.html work major source of value in your life Pre-order tinybox buy.stripe.com/5kAaGL6lk9uX9nW144 more info on -> tinygrad.org | github.com/tinygrad/tinygrad comma 3X comma.ai/shop/comma-3x | best ADAS system in the world openpilot.comma.ai | from $999 comma.ai/shop/body the future of people Support George by subscribing twitch.tv/subs/georgehotz | Follow George on twitter.com/realGeorgeHotz to be up to date | Read George's geohot.github.io/blog/ Chapters: - in progress

2:34:23 - LMAAOOO - "You don't understand AMD! I'm Not Locked in Here With You! You're Locked in Here With Me!"

There is no way a robot can program better than this guy.

Tony let me tell you something I ain't NEVA gonna stop watching George Hotz streams.

the answer to your question @ 9:15, imagine you're a soviet party member in 1986. A scientist, well respected in nuclear tech, approaches you about an issue regarding one of your reactors...

Run George, Run

arrived just in time ;d

Idea: Instead of using chiplets for higher yield (and dealing with that assembly and interconnects), use a larger chip that has built-in design considerations that make it possible to work around fabrication flaws.

31:59 "Do you really think [AMD] knows they're choosing not to be great?" In a Moore's Law is Dead episode last year (citation to be inserted), an AMD insider allegedly said that AMD does not want to compete directly with Nvidia and that AMD explicitly does not want to be number 1. The reason: being the top GPU company is a vastly more materially expensive position to be in than being distant 2nd or an also-ran, and yet, a tremendous amount of money can still be made simply by being in that market without the expense of scaling up to being and remaining the best.

10:47 1.Chip Architecture Eagle r3 2.Building intuition: a synthetic kernel HPC node System Modeler (XXXXXXXXXX.request modeler)MEMS 3.CFD simulation work-flow for model and run finite-temperature lattice dynamics simulations 4.Hypergraphs vs Framework for the Harmonic Balance Method : GPU SOLVER fro CFD DMS 5.Microfabricated Surface-Electrode Ion node 6. Testing 5.5E PolyHex Sp for free license simulation software Benchmark (up -90CPU) 7.Manufacturing Site InFO, CoWoS, Advanced Testing

Sorry haven't seen your other recent stream, what made you decide Nvidia over Intel?

are the AMD issues related to ROCm work, or is it a wholly different avenue or research?

Is there a link to the video when he built the HIP hook?

More people in the market is good, even when they suck. Monopoly/monoculture is the sharecropping of the digital world.

What stream explains why George didn't just use vulkan and slang?

Where are the timestamps?😢

How many adderall’s does it take to make one petaflop

Dealing with AMD really does sound like that scene from Jerry MaGuire "Help me to help you"...

George, did Devin replace us? Are we programmers all obsolete now?

Taalas raised 50m to build their ai chip!

Totally Rad. Love your work. Big Fan. Find me in the chat

AMD not being 'great' works great for my low/medium choice of video cards

...being rich and free are not synonymous, especially if you love what you do...

is this gonna be NVidia EULA compliant or nawt? xD

Hey George..

I can barely see the code. Can you please stream in better resolution?

"We're gonna use AMD!" .. "We're switching the tinybox from AMD to Nvidia." Why is AMD not hiring someone to help George with this crap? It doesn't make sense, they'll waste money on plenty of pointless stuff but not on AI?

Kind of a cop-out. "Money" is easy, the terms of accepting it are hard. Don't gloss over it.

I think I will never buy anymore an AMD GPU 😘

Waah waah waah AMD give be more open source friendly... Next sucking Nvidia balls no toddler tantrums about open source.

Get more money, i mean get elon involved and buy AMD 😂😂😂

George major flaw is to do and try to understand everything by himself. Elon musk doesn't code anymore when you are trying to build a 100b dollar company. Hire a team smarter than yourself in building that product and u deal with funding and expanding your market. I guess u should know this by now

ei eu tava pensando side to side pá - eu entendi oque voce fez deboa. comenta com ele - da pra gastar mto sem fazer nada - más fitaram no role de volta

. Nvidia chat. RTX. Supports Arabic language