Mathis/CUDANet
1
0
Fork 0
mirror of https://github.com/lordmathis/CUDANet.git synced 2025-11-06 01:34:22 +00:00
Code Issues Packages Projects Releases Wiki Activity

Implement max pooling test

Browse Source
This commit is contained in:
Matúš Námešný
2024-03-20 21:44:04 +01:00
parent c062e89972
commit dfff0360d9
9 changed files with 134 additions and 32 deletions
Download Patch File Download Diff File Expand all files Collapse all files

8
src/kernels/pooling.cu
View File

@@ -7,6 +7,7 @@ __global__ void Kernels::max_pooling(
const float* __restrict__ d_input,
float* __restrict__ d_output,
const int inputSize,
const int outputSize,
const int nChannels,
const int poolingSize,
const int stride
@@ -15,7 +16,7 @@ __global__ void Kernels::max_pooling(
int i = blockDim.y * blockIdx.y + threadIdx.y;
int c = blockDim.z * blockIdx.z + threadIdx.z;
if (i >= inputSize || j >= inputSize || c >= nChannels) {
if (i >= outputSize || j >= outputSize || c >= nChannels) {
return;
}
@@ -32,13 +33,14 @@ __global__ void Kernels::max_pooling(
}
}
d_output[c * inputSize * inputSize + i * inputSize + j] = max;
d_output[c * outputSize * outputSize + i * outputSize + j] = max;
}
__global__ void Kernels::avg_pooling(
const float* __restrict__ d_input,
float* __restrict__ d_output,
const int inputSize,
const int outputSize,
const int nChannels,
const int poolingSize,
const int stride
@@ -62,6 +64,6 @@ __global__ void Kernels::avg_pooling(
}
}
d_output[c * inputSize * inputSize + i * inputSize + j] =
d_output[c * outputSize * outputSize + i * outputSize + j] =
sum / (poolingSize * poolingSize);
}