iris / llama.cpp /ggml /src /ggml-cuda /count-equal.cuh
Mat17892's picture
llamacpp
b664585 verified
raw
history blame contribute delete
144 Bytes
#include "common.cuh"
#define CUDA_COUNT_EQUAL_CHUNK_SIZE 128
void ggml_cuda_count_equal(ggml_backend_cuda_context & ctx, ggml_tensor * dst);