The NVIDIA RTX 5070 is the objective winner of this comparison, securing victory in ten out of fifteen measured specifications. Its most significant technical advantage over the competition is its FP32 performance, which is 177% better. While the NVIDIA RTX 2080 SUPER maintains a notable lead in its Tensor Core count, which is 100% better than that of the newer model, the overall hardware superiority clearly favors the more modern architecture across the majority of the evaluated categories.

Users prioritizing raw computational throughput and modern hardware capabilities should choose the NVIDIA RTX 5070 for its substantial lead in FP32 performance. Conversely, the NVIDIA RTX 2080 SUPER remains the appropriate selection for specialized workflows that specifically require the higher density of Tensor Cores provided by that architecture. Selecting between these two components depends entirely on whether your specific tasks demand the superior general processing power of the newer card or the specific core configuration of the older model.