TLDR: Better performance, better frame rate, better image quality, zero developer effort needed. For Turing GPU.
I just copy from Nvidia Blog because Nvidia explained it very good.
Variable Rate Supersampling (VRSS) expands on Turing’s Variable Rate Shading (VRS) feature to deliver image quality improvements by performing selective supersampling. This can also be selectively engaged only if idle GPU cycles are available.
VRSS accomplishes this through fixed foveated super sampling, increasing the shading rate in the center mask region of the screen, while keeping the sampling rate unchanged in the peripheral region. The center mask region can be super-sampled up to 8x to optimize image quality. VRSS is completely handled from within the NVIDIA display driver without application developer integration.
Developers do not have to write any code to integrate VRSS they simply need to submit their VR game or application to NVIDIA for VRSS testing.
NVIDIA VRSS, a Zero-Effort Way to Improve Your VR Image Quality