Nvidia Ran Supercomputer With Thousands of GPUs 24/7 For Six Years To Train DLSS

Expert Verified By

Ran Non-Stop To Improve DLSS Over 6 Years!

Story Highlight
  • Nvidia has a supercomputer running non-stop to improve DLSS.
  • This supercomputer is made up of thousands of GPUs and attempts to find problems like ghosting, flickering, and blurriness.
  • The company then retrains the algorithm for the best possible results.

Nvidia has proven its might in the graphics card market over the last few years, and a major component of its supremacy comes from DLSS. This AI-based upscaling technology is considered the best option for PC gaming today, evident by the fact that 80% of RTX users enable DLSS when given the opportunity.

To Nvidia’s credit, the company has made rapid improvements to the technology since its inception, introducing new elements like ray reconstruction, frame generation, and the new multi-frame generation. As per the company, these improvements have been made possible by training data on a supercomputer 24/7.

Why it matters: With hardware improvements hitting their limitations, the focus has shifted to efficient software that can complement the hardware. This is why AI-based upscaling is such a major asset.

Nvidia Trained DLSS For 6 Years
Image Showcasing DLSS Learning via Nvidia

As per Nvidia, a specific supercomputer, composed of thousands of graphics cards, is dedicated to the task of improving DLSS.

The company states that this computer has been running 24/7 for the past six years without any interruptions. This rigorous process has allowed Nvidia to meticulously dissect the flaws in its upscaling technology.

Brian Catanzaro, VP of applied deep learning and research, explained that problems like ghosting and flickering are analyzed by this process. The company then attempts to augment its data set and retrain the AI model.

Our training data sets are always growing. We’re compiling examples of what good graphics look like and what difficult problems DLSS needs to solve.

-Brian Catanzaro

Nvidia DLSS 4.0 RTX 40 Series
Frame Generation Is Also Being Improved Constantly

Ultimately, this process leads to the near-perfect image quality that fans have come to love and expect from the company. This is largely why CEO Jensen Huang seems unconcerned about the RTX 5090 pricing.

What do you think about this process? Do you also use AI-based upscaling in your games regularly? Let us know in the comments and on the Tech4Gamers Forums.

Was our article helpful? 👨‍💻

Thank you! Please share your positive feedback. 🔋

How could we improve this post? Please Help us. 😔

Gear Up For Latest News

Get exclusive gaming & tech news before it drops. Sign up today!

Join Our Community

Still having issues? Join the Tech4Gamers Forum for expert help and community support!

Latest News

Join Our Community

104,000FansLike
32,122FollowersFollow

Trending

Epic Games Accuses Valve of “Illegal” Methods Over Steam Microtransactions and Fees

Epic Game's CEO accuess Valve, stating that the company still employs excessive commissions and pricing structures that he claims are illegal.

Bethesda Veteran Says Starfield Would’ve Been A Hit If It Was Developed By Some Other Studio

Skyrim lead says if a new, smaller studio developed Starfield, it would've been a hit, because the expectations from Bethesda were too high.

Modders Unlock Multi-Frame Generation Alongside FSR 4, Results Shown On The RX 7900

Modders have now managed to enable FSR 4 and Multi Frame Gen on the 4-year-old RX 7900 and showcased it by running Cyberpunk 2077.

Ex-Rockstar Dev Praises Crimson Desert: “I’m Impressed With The Game”

A former Rockstar developer has reacted to Crimson Desert's latest gameplay, sharing nothing but positive words for the RPG.

Resident Evil Requiem Will Be Less Scary Than RE7, But Still An Improvement Over Village

Resident Evil Requiem director Koshi Nakanishi says that Capcom has toned down the horror in the game since players found RE7 too scary.