#ifndef ENCODER_FORWARD_KERNEL_CUH
#define ENCODER_FORWARD_KERNEL_CUH

#include <cuda_runtime.h>
#include "../utils/cuda_utils.cuh"

__global__ void encoder_forward_kernel(float* out, const int* inp, const float* wte, const float* wpe,
                                       int B, int T, int C) {
    // Implement this
    int x = blockDim.x * blockIdx.x + threadIdx.x;
    int y = blockDim.y * blockIdx.y + threadIdx.y;

    if(x < B && y < T){
        int token = inp[x*T + y];
        for(int i = 0; i < C; i ++){
            out[x*T*C + y*C + i] = wte[token*C+i] + wpe[y*C+i];
        }
    }
}

// Launch kernel here
void encoder_forward(float* out, const int* inp, const float* wte, const float* wpe, int B, int T, int C) {
    // Implement this
    dim3 block(16,16,1);
    dim3 grid((B + block.x -1)/ block.x,(T+block.y-1)/block.y);

    encoder_forward_kernel<<<grid, block>>>(out, inp, wte, wpe,B,T,C);
}


#endif // ENCODER_FORWARD_KERNEL_CUH