OSSForks/tensorflow
1
0
Fork 0
mirror of https://github.com/zebrajr/tensorflow.git synced 2025-12-07 00:20:20 +01:00
Code Issues Actions 5 Packages Projects Releases Wiki Activity

Fixes issue #10258 (#10366)

Browse Source 
On CUDA versions previous to 8.0, only __shared__ variables could be declared as static in the device code.
This commit is contained in:
Joan Puigcerver 2017-06-07 07:03:27 +02:00 committed by gunan
parent cd56a638da
commit 1b6453bec5
1 changed files with 6 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

6
tensorflow/contrib/image/kernels/image_ops.h
View File

@ -66,7 +66,13 @@ class ProjectiveGenerator {
projection;
// TODO(ringwalt): Add a fill value input.
#if (defined __CUDA_ARCH__) && (CUDART_VERSION < 8000)
// On CUDA versions previous to 8.0, only __shared__ variables
// could be declared as static in the device code.
const T fill_value = T(0);
#else
static const T fill_value = T(0);
#endif
switch (interpolation_) {
case INTERPOLATION_NEAREST:
// Switch the order of x and y again for indexing into the image.