diff options
author | 2017-12-03 06:30:40 -0800 | |
---|---|---|
committer | 2017-12-03 06:30:40 -0800 | |
commit | 48892001b48d628a00a535d4cc19d9d7b5fcac11 (patch) | |
tree | 98768985b3df81508d0b8417aa0f558a7f72d41c /tensorflow/core/kernels/cwise_op_gpu_bitwise_and.cu.cc | |
parent | 0fee74eadea48baec80a979763eb19096d148026 (diff) |
Add uint32 and uint64 support for `bitwise_and/or/xor` (#14883)
* Add uint32 and uint64 support for `bitwise_and/or/xor`
In `tensorflow/core/ops/bitwise_ops.cc`, uint32 and uint64
have been enabled for bitwise operations `and/or/xor/left_shift/right_shift`.
However, the kernels of `and/or/xor` have no support of uint32
and uint64. This is in comparision to `left_shift/right_shift` which
have the uint32/uint64 support, and, is tested in `bitwise_ops_test.py`.
This fix adds uint32 and uint64 to bitwise `and/or/xor` kernels and
adds relevant test cases in `bitwise_ops_test.py`, to bring `and/or/xor`
as `left_shift/right_shift`.
Signed-off-by: Yong Tang <yong.tang.github@outlook.com>
* Add uint32 and uint64 support for bitwise_and
Signed-off-by: Yong Tang <yong.tang.github@outlook.com>
* Add uint32 and uint64 support for bitwise_or
Signed-off-by: Yong Tang <yong.tang.github@outlook.com>
* Add uint32 and uint64 support for bitwise_xor
Signed-off-by: Yong Tang <yong.tang.github@outlook.com>
* Register GPU functor for bitwise_and, bitwise_or, bitwise_xor
Signed-off-by: Yong Tang <yong.tang.github@outlook.com>
* Sanitize with clang-format -i --style=Google
Signed-off-by: Yong Tang <yong.tang.github@outlook.com>
Diffstat (limited to 'tensorflow/core/kernels/cwise_op_gpu_bitwise_and.cu.cc')
-rw-r--r-- | tensorflow/core/kernels/cwise_op_gpu_bitwise_and.cu.cc | 3 |
1 files changed, 2 insertions, 1 deletions
diff --git a/tensorflow/core/kernels/cwise_op_gpu_bitwise_and.cu.cc b/tensorflow/core/kernels/cwise_op_gpu_bitwise_and.cu.cc index 27f973c90d..3fbf69c114 100644 --- a/tensorflow/core/kernels/cwise_op_gpu_bitwise_and.cu.cc +++ b/tensorflow/core/kernels/cwise_op_gpu_bitwise_and.cu.cc @@ -19,7 +19,8 @@ limitations under the License. namespace tensorflow { namespace functor { -DEFINE_BINARY6(bitwise_and, int8, int16, int32, int64, uint8, uint16); +DEFINE_BINARY8(bitwise_and, int8, int16, int32, int64, uint8, uint16, uint32, + uint64); } // namespace functor } // namespace tensorflow |