🏠 Home › llama.cpp › ggml › src › ggml-cuda › fattn.cuh

#include "common.cuh"

void ggml_cuda_flash_attn_ext(ggml_backend_cuda_context & ctx, ggml_tensor * dst);

bool ggml_cuda_flash_attn_ext_supported(int device, const ggml_tensor * dst);