NVIDIA discussion [2024]

[Newguy]

Is there no updated CPU with Blackwell? Grace Next was initially scheduled for 2025 IIRC. Vera will be a 2026 part.

No it's still using Grace which uses V2/X3 derived cores (V2 announced 2022). V3 was announced this year so I presume Vera will use V4 in 2026 which should be X7 or X935 derived cores, however they decide to name them.

 

[Samwell]

I don’t think it’s possible to tell from that keynote whether Blackwell-Ultra is “just another H200” (which was also more bandwidth, and possibly a respin, not just more capacity) or if it is indeed a process shrink or more. Besides the insistence on pushing the limits of everything, the only other hint is H100 and H200 have the same “mini die shot” on the slide while Blackwell/Blackwell-Ultra are different, which is hardly conclusive.

However if Blackwell-Ultra is a H200-level change or less, then let me say bluntly that the “yearly cadence” would be complete BS. No one inside a GPU company would consider that a yearly cadence, especially when there have been companies with true yearly cadences in the mobile market with architectural changes nearly every year.

Honestly NVIDIA has too much money *not* to be making a new flagship chip every year. They need to increase their R&D as much as possible as quickly as possible to maintain their competitive advantage, but it’s hard to greatly increase team sizes in a way that doesn’t result in diminishing returns and other inefficiencies (never mind just replacing the people leaving because they made more than enough money to retire on their stock grants!)

Creating a process shrink of the flagship on a tick-tock cadence is the probably the easiest way to spend more money by creating a new independent team without affecting the rest of the roadmap. That doesn’t mean they are actually doing it, but I could not reasonably defend their failure to do so, it would highlight a catastrophic failure of execution to still only have one new flagship every two years while soon making nearly $100B revenue a year from that single chip…

There is also the naming as a hint. H200 is described as Hopper+ in the same chart. Why change nomenclature if Blackwell Ultra is the same?

All these insistence on pushing limits, while your competitor is launching a 3nm product in 2025 and will beat you with it? You can't give your competitor a better opportunity marketing wise and Chip-wise to show everyone they have good competition. Business wise this just seems stupid, especially as AMD is on a tick tock cadence with Mi350 in 3nm 2025 and Mi400 new architecture in 2026.

 

[pharma]

Speculation ....

 

[pharma]

GeForce At Computex 2024: Project G-Assist AI Assistant, NVIDIA ACE, Star Wars™ Outlaws Adds DLSS 3.5, SFF-Ready Unveiled, ComfyUI Acceleration, And Much More Announced

Get the complete rundown on all of our Computex 2024 GeForce announcements.

www.nvidia.com

 

[Malo]

It's almost an April Fools video, but totally serious.

 

[ILikeFeet]

It's almost an April Fools video, but totally serious.

it was actually an April Fools video many years ago

 

[DavidGraham]

TechPowerUp actually tried it and it works as advertised.

NVIDIA Project G-Assist Hands On and Under the Hood

On Sunday, NVIDIA announced Project G-Assist, the AI chatbot for gamers that can be pulled up in the middle of the gameplay, and sought help from. You could just pause your game and Google for help, but G-Assist can be activated in game, and is situationally aware of your game (e.g.: it knows...

www.techpowerup.com

 

[Erinyes]

No it's still using Grace which uses V2/X3 derived cores (V2 announced 2022). V3 was announced this year so I presume Vera will use V4 in 2026 which should be X7 or X935 derived cores, however they decide to name them.

Yea even I expect it to use V4, which should be X6 derived you mean. X7 will not ship until early 2027. But this means a roadmap shift, Grace will be ~2 years old by the time Blackwell ships and will be decidedly last gen.

 

[xpea]

Historic moment.
Nvidia is right now the second most valuable company in the world, just behind Microsoft.
Green team beats Apple, despite the fact that the rotten fruit doesn't buy anything from Jensen. Leather jacket man must feel real good about that

 

[LordEC911]

TechPowerUp actually tried it and it works as advertised.

NVIDIA Project G-Assist Hands On and Under the Hood

On Sunday, NVIDIA announced Project G-Assist, the AI chatbot for gamers that can be pulled up in the middle of the gameplay, and sought help from. You could just pause your game and Google for help, but G-Assist can be activated in game, and is situationally aware of your game (e.g.: it knows...

www.techpowerup.com

They asked it unique questions? All the video demonstrations I've seen asked the same questions.
The demo even showed that a pen was built in the game to keep a specific dino there.
Hopefully someone gets a real hands on to showcase it.

 

[pharma]

They asked it unique questions? All the video demonstrations I've seen asked the same questions.
The demo even showed that a pen was built in the game to keep a specific dino there.
Hopefully someone gets a real hands on to showcase it.

Asking unique questions is more about using ACE. G-Assist is more about getting a response game developers have built an LLM for, or optimizing computer related stuff.

00:00 Nvidia G-assist, how it can serve two functions to assist PC gamers.
03:07 Interacting with Nvidia ACE
04:02 ComfyUI

 

[troyan]

nVidia has joined the $3 Trillion club.

 

[trinibwoy]

The blood bath at the end of this story will be epic.

 

[Cyan]

Historic moment.
Nvidia is right now the second most valuable company in the world, just behind Microsoft.
Green team beats Apple, despite the fact that the rotten fruit doesn't buy anything from Jensen. Leather jacket man must feel real good about that

nVidia has joined the $3 Trillion club.

nVidia has beaten Apple! Yuhu. I am very happy since I like nVidia, but I don't like apple

 

[Bludd]

its still obscene and insane though

 

[doob]

More obscene and insane is the 10 to 1 stock split writing on the wall... truly a silly season.

 

[xpea]

More obscene and insane is the 10 to 1 stock split writing on the wall... truly a silly season.

Worst. Some analysts already give a $500 target after the split. Totally out of control

 

[del42sa]

 

[Seanspeed]

Worst. Some analysts already give a $500 target after the split. Totally out of control

NVDA is becoming a near memestock at this point. There's at least fundamentals underneath, so not like a true memestock, but if it can jump to like 4-5x its current value after a 10:1 split, then clearly the main driving force is simply irrational FOMO.

Would be way worse than even the Tesla situation.

 

[trinibwoy]

NVDA is becoming a near memestock at this point. There's at least fundamentals underneath, so not like a true memestock, but if it can jump to like 4-5x its current value after a 10:1 split, then clearly the main driving force is simply irrational FOMO.

Would be way worse than even the Tesla situation.

It’s not going to $500 anytime soon unless they find some killer app for AI that Nvidia sells directly to consumers.