This folder demonstrates cuSPARSELt Generic APIs usage.
-
Structured Matrix-Matrix Multiplication - Basic Concepts
The sample demonstrates how to exploit NVIDIA Ampere architecture Sparse Tensor Cores for performing Structured Matrix-Matrix Multiplication
-
Batched GEMM, Activation Function, and Bias
The sample extends the previous code to demonstrate how to perform batched GEMM computation, Split-K, and how to set up the activation function and bias