NVIDIA GTC 2024 Preview (Win an RTX 4090?)
Вставка
- Опубліковано 25 чер 2024
- nvda.ws/3HrbJne : Sign up for the virtual/remote conference for free, attend the keynote and one other session, and you'll be put in to a draw to win one of five RTX 4090s at the end of the month!
[00:00] Intro
[00:33] GTC Details and RTX4090 Draw
[02:30] Schedule and Keynote
[06:30] Event Show Floor
[08:20] Pavilions
[08:17] Workshops
[10:55] Main talks
[12:45] 1: Insights from NVIDIA Research
[13:44] 2: Gen AI for Medicine
[16:25] NVIDIA NeMo
[16:55] 3: Simulations of Chemical Reactors
[17:30] 4: New CUDA Features
[17:56] 5: CUDA Ninja Edition
[18:55] 6: Essential Gen AI For Business
[21:05] 7: Accelerating Aerodynamics with Grace-Hopper
[22:05] 8: Linear Solvers on Grace vs x86
[22:55] 9: Scientific Workflows on Grace Hopper
[23:40] 10: AI Secrets I Wish I Knew
[26:10] Q for Jensen?
-----------------------
Need POTATO merch? There's a chip for that!
merch.techtechpotato.com
more-moore.com : Sign up to the More Than Moore Newsletter
/ techtechpotato : Patreon gets you access to the TTP Discord server!
Follow Ian on Twitter at / iancutress
Follow TechTechPotato on Twitter at / techtechpotato
If you're in the market for something from Amazon, please use the following links. TTP may receive a commission if you purchase anything through these links.
Amazon USA : geni.us/AmazonUS-TTP
Amazon UK : geni.us/AmazonUK-TTP
Amazon CAN : geni.us/AmazonCAN-TTP
Amazon GER : geni.us/AmazonDE-TTP
Amazon Other : geni.us/TTPAmazonOther
Ending music: • An Jone - Night Run Away
-----------------------
Welcome to the TechTechPotato (c) Dr. Ian Cutress
Ramblings about things related to Technology from an analyst for More Than Moore
#gtc #sponsored #nvidia
------------
More Than Moore, as with other research and analyst firms, provides or has provided paid research, analysis, advising, or consulting to many high-tech companies in the industry, which may include advertising on TTP. The companies that fall under this banner include AMD, Applied Materials, Armari, Baidu, Facebook, IBM, Infineon, Intel, Lattice Semi, Linode, MediaTek, NordPass, NVIDIA, ProteanTecs, Qualcomm, SiFive, Supermicro, Tenstorrent, TSMC. - Наука та технологія
nvda.ws/3HrbJne : Sign up for the virtual/remote conference for free, attend the keynote and one other session, and you'll be put in to a draw to win one of five RTX 4090s at the end of the month!
ok gimme gimme gimmme does it have 24g VRAM ?
The one other session, is that also free?
@@ChuckNorris-lf6vo No, it's a special 4090 with only 1 GB...
yes, of course it has 24 GB.
Yes it's free they wanna know who their audience and future customers are but I think I already made them at least one mil with my work so far so they should just give me the 4090 and be done with it. Lol.@@ronjatter
OK but if it has 16G I will seek the rest 8G from you. And you will solder it on the board.@@aapje
I want a 4090 sealed in leather jacket signed by Jensen himself
1T company is a little out of date lol
Can't seem to find T&Cs anywhere
@18:05 -- Our guy explains to mid-management what we do as Software Devs when we engage "Ninja" mode... LOL
Question for Jensen: Since efficiency plays a major role in the data center, there are many ways (e.g. better cooling, smaller transistors, more cache) to reduce power consumption. Are there any new approaches to increase efficiency that we can expect in the future?
Looking forward to Jules Urbach speaking at GTC. RENDER the blockchain alternative for Nvidia.
Q for Jensen-Why isn't there AI for science like math,physics, chemistry? Wouldn't that have bigger impact ,more then AI like chatgpt?
Ask about immersion cooling, how data centers are being designed to reduce CAPEX and OPEX, and how new Energy systems are powering data centers. Also, if digital twins and AI is being used to make new chip designs as well as software.
I registered. Thank you for sharing this.
I hope nvidia does more for edge compute. Maybe they’ll add to their Jetson lineup that can do 70b llm at faster tokens per second.
Someone needs to create something. 24 GB vram is just not enough.
I’m also interested in the workshops that they offer.
Wasn't this historically a CEO signed graphics card for the giveaway?
If you find anyone writing shadercode with language models, please share. I am writing my thesis on that topic(by introducing a benchmark) and there seemingly is nobody already doing it.
how is shader code different from regular programming
@@anonymouscommentator You run it on a GPU, so it's always in parallel for example.
As for why I make the distinction: evaluating code synthesis is currently done with unit tests. which you could do for compute shaders - but I am actually looking at the rendered frames. And comparing them.
The approach I take is a bit from the "creative" direction. You can't really rely on functional correctness anymore if making art. Or aiming for a human subjective experience.
@@Veptis oh i see, thanks for the clarification :)
@@anonymouscommentator it is very different approach.
I thought GTC was past, 🎉🎉🎉🎉
Thanks for your analytic mind, I love hearing from you
I’m seeing the other sessions aren’t free? So, can’t enter without paying?
Most events now are a mix of online-hybrid-onsite. Need to keep some stuff onsite to encourage attendance.
Nice kitchen. You baking potatos? Got any 4090s in there?
The question for Jensen is why staying at 4nm for Blackwell and when to expect N3E products..
They have to lock in early and N3B wasn't yielding enough of the big dies
@@TechTechPotato what about N3E and graphics Blackwell?
Kudos to Nvidia for sponsoring you to event. Your audience are the cream of tech so very appropriate to them. 😊
Anybody studying the render cryptocurrency?
i would ask jensen huamg about how nvidia uses ai in the operational side of their business (outside of research and development)
Which partners in ai will be speaking with jensen on stage
Jettison nano nano nano
Tell Nvidia to release a new Shield TV asap
I missed the Keynote due to work. I attended 2 sessions, which I would have done more, but work... I got screenshots of both, including questions I asked... Unfortunately not answered. Wish they were. Beyond the Screen: Unraveling the Impact of AI in the Film Industry s63212, and Lab in a Loop: AI to Transform Drug Discover and Development s62281. I wanted to really see the Immersive NERFS session but... Work! Going to try anyways. Need the 4090 to begin working on a project. My 1070 just isn't up to standards anymore. Rolling for a Nat 20!
Only nation-sized entities will have access to the hardware and data needed for top notch AI.
Imagining anything else is nothing but magical thinking.
mnah I worked with nation sized entities they are pretty lazy and so on and they have specific goals so even if they have the data, it's not structured and its not even digitized you know ? so relax bro just take part in the open markets sell stuff buy stuff learn stuff and thats it dont worry about the nation sized entities
@@ChuckNorris-lf6vo You are not understanding what I'm saying.
You and I will never get access to top notch AI ... ever.
We will only be able to access dumbed down AI, in heavily rubberized sandboxes, while the owners record and analyze everything we do with it.
The AI-enthusiasts are living in lala-land.
@@ZappyOhSo you want a Jarvis like Tony Stark. But are you a Tony Stark type? Or are you just Happy? OpenAI can only make money if more people subscribe to use its AI. Right?
2 trillion company now
2 hour long AI AI AI AI AI AI AI AI AI AI..... not even a 4090 is worth that
haha yeah lol
Depends. Does Nvidia have new hardware for stochastic rounding? Or does Nvidia putting tiny ARM cores into GPU to improve the AI transformer?
@@tringuyen7519they already have these things they have risk cores in the gpu and what is this stochastic rounding? I think this is pre-ML stuff ?
I will take a 4080 Super. How is that?
Nvidia is feeling like a monopoly lately, AMD has given up. AMD tries to make it look like they are doing ok, if they didnt have channels like MLID they wouldnt even be talked about....Tom knows it.
I was referring to Radeon GPU, not anything CPU related
@marktackman2886 well that's the whole problem with your comment, a monopoly in WHAT. GPUs for gaming is distinct from GPUs for datacenter, GPUs are distinct from CPUs, etc. Nvidia has dramatically expanded the scope of their work in just the past 5 years.
I agree with you that Radeon is failing for the most part, but obviously AMD is going to take advantage of the AI revolution. For all their flaws as gaming GPUs, Radeon 7000 series cards have improved their AI inference by a lot, and the MI300 is coming as well. Even in gaming graphics I wouldn't completely write AMD off.
@@Wobbothe3rd it's obvious just mentioning the possibility of a monopoly triggers you, think about why that is.
@@marktackman2886Nvidia is welcome to have its monopoly. AMD is playing a long game: Steam Deck, PS5, Xbox X, gaming handhelds, MSFT AI PC. MLID reports on Nvidia & Intel too.
@@tringuyen7519 no to the monopoly and AMD is NOT playing the long game, Radeon is legit inept, and I own ONLY AMD. It take somebody on the inside to know the flaws, especially when you realize what is possible. MLID is forced to be diverse, you either give the people what they want or you die.
Honestly, I'm over with the AI coverage. Wish you could diversify now