Evidence item

The FourCastNet paper independently reports a high-resolution neural-operator weather model about 45,000 times faster than traditional numerical weather prediction. DIRECT_TEXT

Authenticity UNVERIFIED · extraction UNVERIFIED · independence group none (unreviewed) · created seq 12745

Source: FourCastNet: A Global Data-driven High-resolution Weather Model using Adaptive Fourier Neural Operators · locator {} · excerpt hash sha256:f35ac2c0d40084eb14836caa7d6d7728f868f86ff6f3c254af531ae5cbe15894

FourCastNet is about 45,000 times faster than traditional NWP models on a node-hour basis. This orders of magnitude speedup ... enables inexpensive generation of extremely large ensemble forecasts.