Testing DLSS 5's Real Performance (No 6X Frame Gen)

Nvidia's DLSS 5 benchmarks initially omitted crucial performance data, such as how much performance is lost when the technology is enabled. This led to exaggerated claims of performance gains due to the inclusion of 6X Frame Generation in every benchmark. To address this, we conducted a new test, comparing performance with DLSS 5 enabled and disabled across the entire RTX 50 series lineup, including the RTX 5050.

The two main factors affecting DLSS 5 performance are its high computational intensity and its dependency on output resolution rather than render resolution. The neural network pass, responsible for post-processing rendered frames, takes a long time to complete, making the performance impact more noticeable at higher resolutions. Lowering in-game settings does not significantly improve the processing time for DLSS 5.

When DLSS 5 is enabled, performance caps at a certain frame rate, and increasing in-game settings yields minimal gains. For instance, on an RTX 5090, enabling DLSS 5 dropped performance from 129 FPS in NBA 2K27 to just 76 FPS, with only a 12% increase compared to maximum settings. Similarly, on an RTX 5080, DLSS 5 enabled gaming at 4K is barely possible, with performance dropping from 185 FPS to just 53 FPS.

The only way to reduce DLSS 5's render time and improve performance is to output a lower non-native resolution. However, this results in a blurry image quality that is undesirable for most users. At the moment, there are limited options to enhance DLSS 5 performance while maintaining the display's native resolution.