GPU Computing for AI

cuDNN

GPU-accelerated library of primitives for deep neural networks, developed by NVIDIA, providing highly optimized implementations for convolution, pooling, and normalization routines.

← Indietro