• Hey, guest user. Hope you're enjoying NeoGAF! Have you considered registering for an account? Come join us and add your take to the daily discourse.

Digital Foundry: Inside DLSS 4 & Nvidia Machine Learning: The Bryan Catanzaro Interview

adamsapple

Or is it just one of Phil's balls in my throat?




The RTX 50-series cards are coming and at the forefront of Nvidia innovation is a full revamp of DLSS, with revised transformer model Super Resolution and Ray Reconstruction, along with up to 4x multi frame generation. At CES 2025, Alex had the opportunity to sit down and talk about the new technologies with Nvidia's Bryan Catanzaro, VP of Applied Deep Learning Research.

00:00 Introduction
00:48 Why switch from CNNs to transformers?
02:08 What are some image characteristics that are improved with DLSS 4 Super Resolution?
03:17 Is there headroom to continue to improve on Super Resolution?
04:12 How much more expensive is DLSS 4 Super Resolution to run?
05:25 How does the transformer model improve Ray Reconstruction?
09:43 Why is frame gen no longer using hardware optical flow?
13:06 Could the new Frame Generation run on RTX 3000?
13:44 What has changed for frame pacing with DLSS 4 Frame Generation?
15:37 Will Frame Generation ever support standard v-sync?
17:18 Could you explain how Reflex 2 works?
21:11 What is the lowest acceptable input frame-rate for DLSS 4 Frame Generation?
22:13 What does the future of real-time graphics look like?
 
Last edited:

Kilau

Gold Member
These frames

Star Trek Hoax GIF
 

llien

Banned

Fess

Member
I don’t understand half of what they’re saying but the little I do understand sounds promising even for old gen cards.
 

yamaci17

Member
these news heavily imply that nvidia's frame generation will land on 3000 series at some point

the big reason they had for no 3000 support was slow optical flow accelerator on that series. and supposedly they completely ditched using it on the 4000 and now will use tensor cores for generation.

there's lots of 3000 cards that have much better tensor performance than a RTX 4060 for example. and if a weak laptop RTX 4050 with subpar tensor performance is going to be able to run it, a minimum of 3060/3060ti should easily run the same workload without slowdowns
 
Last edited:

Buggy Loop

Gold Member
I am not trusting a man with those glasses. Doesn't strike me as a nerd

Excuse Me Reaction GIF by One Chicago


dafuq

This guy's parkour from a PhD to VP of applied deep learning research at Nvidia is nothing but amazing. He's in almost all papers on the subject, has plenty of publications and has been cited more times than you've played games in your life Rofif



Nvidia doesn't hire on stupid criterias like yours, clearly.
 
Last edited:

adamsapple

Or is it just one of Phil's balls in my throat?

Magic Carpet

Gold Member
I started the video way too early in the morning, my brain can't keep up with the concepts. I need breakfast to fuel my understanding. I can't get to the waffle house, too much snow.
I'll just go with the usual joke of needing a new GPU to melt the snow, I needs to eat.
 

kruis

Exposing the sinister cartel of retailers who allow companies to pay for advertising space.
AMD should send hire someone to go after Bryan Catanzaro with a pair of scissors. There's a rumor that if someone were to cut off his hair, he'd lose all his intellectual powers.
 

MacReady13

Member
Clicked on the YouTube video link. Saw that fuck knuckle Alex’s face. Turned it off. I have as much interest in small graphical upgrades as I do needing to see or hear that retard.
 
Top Bottom