An Nvidia supercomputer has been improving DLSS non-stop for the past six years

Nvidia's supercomputer improves DLSS for six years, launching DLSS 4 with RTX 50 GPUs.

: Nvidia has been using a supercomputer to enhance DLSS for over six years. The continuous process involves analyzing and retraining models to fix errors. DLSS 4, paired with RTX 50 GPUs, claims significant performance boosts.

For over six years, Nvidia has leveraged a supercomputer loaded with thousands of GPUs to continuously improve its DLSS technology. This robust infrastructure operates year-round with its primary goal focused on enhancing Deep Learning Super Sampling, or DLSS.

The training process involves analyzing errors such as ghosting and blurriness in games when DLSS models fail. Bryan Catanzaro explained that identifying these failures aids in refining the training data, which is then used to retrain models tested across hundreds of games.

Recently, Nvidia launched DLSS 4, alongside the RTX 50 Blackwell GPUs at CES. The new RTX 5070 is marketed to offer performance akin to the 4090 when using DLSS 4, with notable performance increases across other models as well, according to Nvidia's benchmarks.

NvidiaDLSSSupercomputer