Did Nvidia Just Do Us A Favor? - DLDSR Explained
0 up · 0 down · 0 ratings
Promos
Try FreshBooks free, for 30 days, no credit card required at freshbooks.com How can Nvidia's Deep Learning Dynamic Super Resolution, or DLDSR, improve the look of your games without much of a performance hit? Leave a reply with your requests for future episodes, or tweet them here: twitter.com ► GET MERCH: lttstore.com ► AFFILIATES, SPONSORS & REFERRALS: lmg.gg ► PODCAST GEAR: lmg.gg ► SUPPORT US ON FLOATPLANE: floatplane.com FOLLOW US ELSEWHERE --------------------------------------------------- Twitter: twitter.com Facebook: @LinusTech Instagram: @linustech TikTok: @linustech Twitch: twitch.tv
Did Nvidia Just Do Us A Favor? DLDSR Explained digs into how Nvidia's DLDSR technology aims to sharpen game images without demanding new hardware or hurting performance too much. The video begins by framing the problem for modern gamers: GPUs are expensive and a new card often costs more than a small fortune on the used market, so manufacturers push techniques to improve quality without fully upgrading. It recaps prior approaches like DLSS and FSR, then introduces DLDSR as Nvidia's latest enhancement that leverages tensor cores designed for machine learning. The core explanation compares traditional dynamic super resolution (DSR) and SSAA, where rendering at a higher-than-native resolution is downscaled, to the DLDSR approach that uses a neural network to infer the final pixel data and surrounding context. A key example is given from the game Prey, where rendering at 1620p and downscaling to 1080p produced nearly flat frame rate loss and a quality boost compared to older methods, with no per-game integration required. The host notes that combining DLDSR with games that support DLSS can yield additional improvements, though early reviews suggest the quality gains over regular DSR are modest and the performance impact can still be noticeable if you push the scaling factor. The video also touches on AMD's upcoming AI-based Radeon Super Resolution and contrasts the lack of dedicated ML cores on AMD cards with Nvidia's RTX tensor cores, inviting viewers to watch how both technologies fare in real-world gaming. In closing, the host injects light humor about super resolution and encourages engagement through likes, subscriptions, and future topic suggestions.
Topics · technology · graphics · hardware · ai
Questions answered
- What is DLDSR and how does it work?
- DLDSR is Nvidia's deep learning based dynamic super resolution that uses tensor cores to sample fewer pixels than traditional supersampling and then predicts the final image, potentially offering higher quality with less performance cost compared to regular DSR.
- How is DLDSR different from regular DSR and SSAA?
- DLDSR uses a neural network to determine how the image should look using surrounding pixels, rather than simply upscaling and then downscaling as in regular DSR or SSAA, and it can be combined with DLSS for extra quality, though the gains may be modest.
- Do I need an RTX GPU to use DLDSR?
- Yes, DLDSR requires an RTX graphics card to leverage the tensor cores; AMD is expected to offer a competing AI based solution but without the same hardware foundation.