mirror
/
darknet
mirror of https://github.com/pjreddie/darknet.git
1
0
Fork 0
Code Issues Projects Releases Wiki Activity

Merge d512dd7391 into f6afaabcdf

This commit is contained in:
Anup Parikh 2022-10-08 04:05:23 -04:00 committed by GitHub
parent f6afaabcdf d512dd7391
commit 02976ae6be
No known key found for this signature in database
GPG Key ID: 4AEE18F83AFDEB23
2 changed files with 24 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

19
src/batchnorm_layer.c
View File

@ -228,9 +228,28 @@ void forward_batchnorm_layer_gpu(layer l, network net)
add_bias_gpu(l.output_gpu, l.biases_gpu, l.batch, l.out_c, l.out_w*l.out_h);
#endif
} else {
#ifdef CUDNN
float one = 1;
float zero = 0;
cudnnBatchNormalizationForwardInference(cudnn_handle(),
CUDNN_BATCHNORM_SPATIAL,
&one,
&zero,
l.dstTensorDesc,
l.x_gpu,
l.dstTensorDesc,
l.output_gpu,
l.normTensorDesc,
l.scales_gpu,
l.biases_gpu,
l.rolling_mean_gpu,
l.rolling_variance_gpu,
.00001);
#else
normalize_gpu(l.output_gpu, l.rolling_mean_gpu, l.rolling_variance_gpu, l.batch, l.out_c, l.out_h*l.out_w);
scale_bias_gpu(l.output_gpu, l.scales_gpu, l.batch, l.out_c, l.out_h*l.out_w);
add_bias_gpu(l.output_gpu, l.biases_gpu, l.batch, l.out_c, l.out_w*l.out_h);
#endif
}
}

5
src/convolutional_layer.c
View File

@ -122,6 +122,11 @@ static size_t get_workspace_size(layer l){
#ifdef CUDNN
void cudnn_convolutional_setup(layer *l)
{
#if(CUDNN_MAJOR >= 7)
// Note: The library falls back to the default math mode CUDNN_DEFAULT_MATH when Tensor Core operations are not supported or not permitted.
cudnnSetConvolutionMathType(l->convDesc, CUDNN_TENSOR_OP_MATH);
#endif
cudnnSetTensor4dDescriptor(l->dsrcTensorDesc, CUDNN_TENSOR_NCHW, CUDNN_DATA_FLOAT, l->batch, l->c, l->h, l->w);
cudnnSetTensor4dDescriptor(l->ddstTensorDesc, CUDNN_TENSOR_NCHW, CUDNN_DATA_FLOAT, l->batch, l->out_c, l->out_h, l->out_w);