Mathis/CUDANet
1
0
Fork 0
mirror of https://github.com/lordmathis/CUDANet.git synced 2025-12-22 22:34:22 +00:00
Code Issues Packages Projects Releases Wiki Activity

Refactor Backend and Layer interfaces

Browse Source
This commit is contained in:
Matúš Námešný
2025-11-18 18:27:57 +01:00
parent 25670f90c4
commit 6340b27055
23 changed files with 154 additions and 201 deletions
Download Patch File Download Diff File Expand all files Collapse all files

6
src/backends/cuda/layer_ops.cu
View File

@@ -5,21 +5,21 @@
using namespace CUDANet::Backend;
void CUDABackend::relu(Tensor &tensor) {
void CUDA::relu(Tensor &tensor) {
int gridSize = (tensor.numel() + BLOCK_SIZE - 1) / BLOCK_SIZE;
Kernels::relu<<<gridSize, BLOCK_SIZE>>>(tensor.data<float>(), tensor.data<float>(), tensor.numel());
CUDA_CHECK(cudaGetLastError());
CUDA_CHECK(cudaDeviceSynchronize());
}
void CUDABackend::sigmoid(Tensor &tensor) {
void CUDA::sigmoid(Tensor &tensor) {
int gridSize = (tensor.numel() + BLOCK_SIZE - 1) / BLOCK_SIZE;
Kernels::sigmoid<<<gridSize, BLOCK_SIZE>>>(tensor.data<float>(), tensor.data<float>(), tensor.numel());
CUDA_CHECK(cudaGetLastError());
CUDA_CHECK(cudaDeviceSynchronize());
}
void CUDABackend::softmax(Tensor &tensor, Tensor &temp_max, Tensor &temp_sum) {
void CUDA::softmax(Tensor &tensor, Tensor &temp_max, Tensor &temp_sum) {
int gridSize = (tensor.numel() + BLOCK_SIZE - 1) / BLOCK_SIZE;
// Find max value