NVIDIA Reflex 2 vs NVIDIA Reflex

NVIDIA announced the second generation of their Reflex technology alongside the 50 Series of GPUs. Let's see how it differs from the original version.

First of all, let's look back at the first generation of Reflex, which was launched in 2020 at the same time as the RTX 30 series, with the aim of improving latency in competitive gaming.

It was a set of APIs that developers could integrate directly into their games. Low Latency Mode allows the game engine and rendering to be synchronised to reduce latency. This is achieved by optimising the pressure on the CPU, particularly during very heavy scenes, and by removing the queue for GPU rendering.

This solution significantly improves the results obtained with previous techniques based on driver optimisation. To date, just over one hundred games have been integrated into the programme.

In a classic case, even with Reflex Low Latency enabled, when the GPU renders an image, the CPU calculates the camera position of the next image in the pipeline based on the last mouse or joystick input. There is therefore a latency equal to the rendering time between your mouse movement and its display on the screen.

With Frame Warp technology, when the GPU rendering is complete, the API analyses the most recent position of your camera from the CPU, which calculates its position for the next render, and deforms the image just completed by the GPU according to this new camera position.

The distortion is performed as late as possible, so that the image on the screen and your movements are virtually in sync.

The deformation creates 'holes' in the generated images, but thanks to a predictive rendering algorithm, you won't see these holes on the screen.

Reflex Low Latency mode is particularly effective when a PC is saturated by the GPU. But Reflex 2 with Frame Warp delivers significant performance in scenarios where the CPU and GPU are saturated. In VALORANT, for example, Nvidia claims to be able to achieve an average latency of 3ms with the new GeForce RTX 5090, all at +800 FPS.

Frame Warp technology is the second step after Reflex Low Latency mode in further reducing latency, with up to a 75% reduction in the best situations.

At launch, only RTX 50 series GPUs, Valorant and The Finals will support the API, but as with Reflex Low Latency, many games should support it in the future, as well as possibly Nvidia GPUs from earlier series.