[CUDA] Matmul utils initial commit (#2441)

This commit is contained in:
Angelos Katharopoulos
2025-08-01 14:22:25 -07:00
committed by GitHub
parent 86258f292f
commit be9bc96da4
32 changed files with 856 additions and 14 deletions

View File

@@ -171,6 +171,7 @@ void ArgReduce::eval_gpu(const std::vector<array>& inputs, array& out) {
kernel,
num_blocks,
block_dim(),
0,
in.data<T>(),
out.data<uint32_t>(),
out.size(),