mmr/tkDNN
Archived
1
0
Fork 0
Code Issues 46 Pull Requests 12 Releases 4 Activity

Optimize deformable kernel

Browse Source 
Signed-off.by: Ignacio Sañudo Olmedo<ignacio.sanudoolmedo@unimore.it>
Signed-off-by: Davide Sapienza <sapienza.dav@gmail.com>
This commit is contained in:
Davide Sapienza
2020-02-11 10:06:08 +01:00
parent 34ecb6a1bb
commit 2b06836a7f
3 changed files with 180 additions and 60 deletions
Download Patch File Download Diff File Expand all files Collapse all files
include/tkDNN/kernels.h
+9 -3
View File
@@ -30,14 +30,20 @@ void upsampleForward(dnnType* srcData, dnnType* dstData,
void float2half(float* srcData, __half* dstData, int size, const cudaStream_t stream = cudaStream_t(0));
// void modulated_deformable_im2col_cuda(cudaStream_t stream,
// const float *data_im, const float *data_offset, const float *data_mask,
// const int batch_size, const int channels, const int height_im, const int width_im,
// const int height_col, const int width_col, const int kernel_h, const int kenerl_w,
// const int pad_h, const int pad_w, const int stride_h, const int stride_w,
// const int dilation_h, const int dilation_w,
// const int deformable_group, float *data_col);
void modulated_deformable_im2col_cuda(cudaStream_t stream,
const float *data_im, const float *data_offset, const float *data_mask,
const int batch_size, const int channels, const int height_im, const int width_im,
const int height_col, const int width_col, const int kernel_h, const int kenerl_w,
const int pad_h, const int pad_w, const int stride_h, const int stride_w,
const int dilation_h, const int dilation_w,
const int height_col, const int width_col,
const int deformable_group, float *data_col);
void dcn_v2_cuda_forward(cublasStatus_t stat, cublasHandle_t handle,
float *input, float *weight,
float *bias, float *ones,