mirror of
https://github.com/lordmathis/CUDANet.git
synced 2025-11-06 01:34:22 +00:00
Implement Add layer
This commit is contained in:
29
src/layers/add.cu
Normal file
29
src/layers/add.cu
Normal file
@@ -0,0 +1,29 @@
|
||||
#include "add.cuh"
|
||||
#include "matmul.cuh"
|
||||
#include "cuda_helper.cuh"
|
||||
|
||||
using namespace CUDANet;
|
||||
|
||||
|
||||
Layers::Add::Add(int inputSize)
|
||||
: inputSize(inputSize) {
|
||||
|
||||
d_output = nullptr;
|
||||
CUDA_CHECK(cudaMalloc((void**)&d_output, sizeof(float) * inputSize));
|
||||
|
||||
gridSize = (inputSize + BLOCK_SIZE - 1) / BLOCK_SIZE;
|
||||
}
|
||||
|
||||
|
||||
Layers::Add::~Add() {
|
||||
cudaFree(d_output);
|
||||
}
|
||||
|
||||
|
||||
float* Layers::Add::forward(const float* d_inputA, const float* d_inputB) {
|
||||
|
||||
Kernels::vec_vec_add<<<gridSize, BLOCK_SIZE>>>(
|
||||
d_inputA, d_inputB, d_output, inputSize
|
||||
);
|
||||
|
||||
}
|
||||
Reference in New Issue
Block a user