NVIDIA Teases Potential DLSS 4 Frame Generation Support for RTX 30 GPUs

NVIDIA is exploring the possibility of enabling DLSS 4 Frame Generation on older RTX 30 series graphics cards, offering a ray of hope for gamers using previous-generation hardware. This was hinted at by Bryan Catanzaro, NVIDIA’s Vice President of Applied Deep Learning Research, during a discussion at CES 2025. He elaborated on the enhancements in DLSS 4, including Super Resolution, Ray Reconstruction, and a revamped Frame Generation system.

DLSS 4 Introduces Transformer Models for Better Performance

The most significant improvement in DLSS 4 is the shift from CNN models to advanced transformer models. These new models handle more complex datasets, significantly reducing issues like ghosting and shimmering. According to Catanzaro, this change boosts the computing power for Super Resolution fourfold compared to DLSS 3. These enhancements are optimized for the upcoming GeForce RTX 50 GPUs, built on the Blackwell architecture, set to debut later this month.

One of the standout features of DLSS 4 is its AI-based Multi Frame Generation, which replaces the reliance on the Optical Flow hardware accelerator. Instead, it leverages Tensor Cores to deliver superior image quality while consuming less VRAM.

Potential for Frame Generation Support on RTX 30 GPUs

When asked about support for Frame Generation on RTX 30 GPUs, Catanzaro expressed cautious optimism but acknowledged technical hurdles. The key challenge lies in the lesser Tensor Core capabilities of RTX 30 cards compared to the more powerful RTX 50 series. However, by removing the dependency on the Optical Flow hardware accelerator—which restricted Frame Generation to RTX 40 cards—the RTX 30 series might gain limited support for this feature.

Additionally, DLSS 4 introduces improvements like optimized flip metering, which reduces frame time variability for smoother pacing, and Reflex 2, an AI-driven latency-reduction technology aimed at enhancing responsiveness, particularly in competitive gaming scenarios.