Nvidia Shifts Focus to Software Amid Missing GeForce Super GPUs
▼ Summary
– Nvidia did not announce new GeForce graphics cards at CES, instead focusing its keynote on AI and releasing gaming news separately.
– The main software announcement was DLSS 4.5, which introduces new features for upscaling and frame generation.
– A key improvement is a new AI model for DLSS upscaling, designed to enhance image quality, especially in Performance and Ultra Performance modes.
– DLSS Multi-Frame Generation is being upgraded to generate more AI frames and can now dynamically adjust the number based on scene demands.
– Multi-Frame Generation still requires an RTX 50-series GPU and a high base frame rate, making it unsuitable for turning unplayable frame rates into playable ones.
Nvidia’s recent absence of new GeForce GPU announcements at CES marks a notable shift in the company’s public strategy, placing a far greater emphasis on software advancements and its booming artificial intelligence sector. The spotlight moved decisively from hardware to enhancing the capabilities of existing graphics cards through sophisticated AI-driven features. This pivot underscores a mature market where extracting more performance from current silicon through intelligent software is becoming as critical as launching new chips.
The centerpiece of this software-focused approach is DLSS 4.5, a significant update to Nvidia’s suite of upscaling and frame generation technologies. The core upscaling technology receives a boost from what Nvidia describes as a new “second-generation transformer model.” This model has been trained on a much larger dataset, which the company claims significantly improves its predictive accuracy when creating new pixels. This enhancement is said to be particularly impactful for the Performance and Ultra Performance modes, where the system must infer more visual information from a lower-resolution source image, leading to a noticeable jump in final image quality.
Another major leap comes to DLSS Multi-Frame Generation. The technology now increases the number of AI-generated frames per rendered frame from three to five, a new tier called 6x mode. This is paired with an intelligent feature dubbed Dynamic Multi-Frame Generation. This system allows the number of generated frames to adjust on the fly, ramping up during graphically intense scenes for smoother motion and scaling back during simpler sequences to conserve resources. The principle is to compute only what is necessary, optimizing both performance and efficiency.
It is important to note that the existing limitations for Multi-Frame Generation remain firmly in place. The feature continues to require an RTX 50-series GPU; RTX 40-series cards are capped at generating one AI frame per rendered frame, and older architectures cannot use the feature at all. Furthermore, the technology is not a magic fix for poor performance. Games still need to run at a reasonably high base frame rate to minimize input lag and visual artifacts. Its primary function is to make already smooth games feel even more fluid, not to rescue an unplayably low frame rate.
This strategic emphasis on software over new consumer GPU hardware reflects Nvidia’s broader business realities and the evolving demands of the market. By deepening the value of its current and upcoming architectures through AI software, the company is building a more integrated ecosystem. For gamers, it means that the potential of their existing RTX cards continues to grow, even in the absence of a new “Super” series refresh on the immediate horizon.
(Source: Ars Technica)
