Fixed cuDNN 8.x usage

3dEYE · Sep 6, 2020 · eb0272f · eb0272f
1 parent ecad770
commit eb0272f
Show file tree

Hide file tree

Showing 2 changed files with 28 additions and 3 deletions.
diff --git a/scripts/get_coco2017.sh b/scripts/get_coco2017.sh
@@ -0,0 +1,25 @@
+#!/bin/bash
+# Zip coco folder
+# zip -r coco.zip coco
+# tar -czvf coco.tar.gz coco
+
+# Download labels from Google Drive, accepting presented query
+filename="coco2017labels.zip"
+fileid="1cXZR_ckHki6nddOmcysCuuJFM--T-Q6L"
+curl -c ./cookie -s -L "https://drive.google.com/uc?export=download&id=${fileid}" > /dev/null
+curl -Lb ./cookie "https://drive.google.com/uc?export=download&confirm=`awk '/download/ {print $NF}' ./cookie`&id=${fileid}" -o ${filename}
+rm ./cookie
+
+# Unzip labels
+unzip -q ${filename}  # for coco.zip
+# tar -xzf ${filename}  # for coco.tar.gz
+rm ${filename}
+
+# Download and unzip images
+cd coco/images
+f="train2017.zip" && curl http://images.cocodataset.org/zips/$f -o $f && unzip -q $f && rm $f  # 19G, 118k images
+f="val2017.zip" && curl http://images.cocodataset.org/zips/$f -o $f && unzip -q $f && rm $f  # 1G, 5k images
+# f="test2017.zip" && curl http://images.cocodataset.org/zips/$f -o $f && unzip -q $f && rm $f  # 7G,  41k images
+
+# cd out
+cd ../..
diff --git a/src/convolutional_layer.c b/src/convolutional_layer.c
@@ -319,7 +319,7 @@ void cudnn_convolutional_setup(layer *l, int cudnn_preference, size_t workspace_
         if (conv_fwd_results[i].status == CUDNN_STATUS_SUCCESS &&
             conv_fwd_results[i].algo != CUDNN_CONVOLUTION_FWD_ALGO_WINOGRAD_NONFUSED &&
             conv_fwd_results[i].memory < free_memory &&
-            conv_fwd_results[i].memory <= workspace_size_specify &&
+            (conv_fwd_results[i].memory <= workspace_size_specify || cudnn_preference == cudnn_fastest) &&
             conv_fwd_results[i].time < min_time)
         {
             found_conv_algorithm = 1;
@@ -357,7 +357,7 @@ void cudnn_convolutional_setup(layer *l, int cudnn_preference, size_t workspace_
     {
         if (conv_bwd_data_results[i].status == CUDNN_STATUS_SUCCESS &&
             conv_bwd_data_results[i].memory < free_memory &&
-            conv_bwd_data_results[i].memory <= workspace_size_specify &&
+            (conv_bwd_data_results[i].memory <= workspace_size_specify || cudnn_preference == cudnn_fastest) &&
             conv_bwd_data_results[i].time < min_time)
         {
             found_conv_algorithm = 1;
@@ -394,7 +394,7 @@ void cudnn_convolutional_setup(layer *l, int cudnn_preference, size_t workspace_
     {
         if (conv_bwd_filter_results[i].status == CUDNN_STATUS_SUCCESS &&
             conv_bwd_filter_results[i].memory < free_memory &&
-            conv_bwd_filter_results[i].memory <= workspace_size_specify &&
+            (conv_bwd_filter_results[i].memory <= workspace_size_specify || cudnn_preference == cudnn_fastest) &&
             conv_bwd_filter_results[i].time < min_time)
         {
             found_conv_algorithm = 1;