experiments/STT-tensorflow
1
0
Fork 0
Code Issues Pull Requests Projects Releases Wiki Activity

Enable the generated versions of invert, logical_not and neg.

Browse Source 
PiperOrigin-RevId: 354995894
Change-Id: I563774705838bd04e03f17f878fcef8dfcc981c6
This commit is contained in:
Stephan Herhut 2021-02-01 12:08:15 -08:00 committed by TensorFlower Gardener
parent b5d266600a
commit 21928d2d02
8 changed files with 19 additions and 16 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
tensorflow/core/kernels/cwise_op_gpu_invert.cu.cc
View File

@ -19,7 +19,11 @@ limitations under the License.
namespace tensorflow {
namespace functor {
#if !defined(MLIR_GENERATED_GPU_KERNELS_ENABLED)
DEFINE_UNARY8(invert, int8, int16, int32, int64, uint8, uint16, uint32, uint64);
#else
DEFINE_UNARY4(invert, uint8, uint16, uint32, uint64);
#endif
} // namespace functor
} // namespace tensorflow

2
tensorflow/core/kernels/cwise_op_gpu_logical_not.cu.cc
View File

@ -19,7 +19,9 @@ limitations under the License.
namespace tensorflow {
namespace functor {
#if !defined(MLIR_GENERATED_GPU_KERNELS_ENABLED)
template struct UnaryFunctor<GPUDevice, logical_not>;
#endif
} // namespace functor
} // namespace tensorflow

5
tensorflow/core/kernels/cwise_op_gpu_neg.cu.cc
View File

@ -19,8 +19,11 @@ limitations under the License.
namespace tensorflow {
namespace functor {
#if !defined(MLIR_GENERATED_GPU_KERNELS_ENABLED)
DEFINE_UNARY4(neg, int8, int16, int32, int64);
DEFINE_UNARY6(neg, Eigen::half, float, double, bfloat16, complex64, complex128);
DEFINE_UNARY3(neg, Eigen::half, float, double);
#endif
DEFINE_UNARY3(neg, bfloat16, complex64, complex128);
} // namespace functor
} // namespace tensorflow

3
tensorflow/core/kernels/cwise_op_invert.cc
View File

@ -21,8 +21,7 @@ REGISTER8(UnaryOp, CPU, "Invert", functor::invert, int8, int16, int32, int64,
#if GOOGLE_CUDA || TENSORFLOW_USE_ROCM
#if !defined(MLIR_GENERATED_GPU_KERNELS_ENABLED) || \
!defined(MLIR_GENERATED_EXPERIMENTAL_GPU_KERNELS_ENABLED)
#if !defined(MLIR_GENERATED_GPU_KERNELS_ENABLED)
REGISTER8(UnaryOp, GPU, "Invert", functor::invert, int8, int16, int32, int64,
uint8, uint16, uint32, uint64);
#else

3
tensorflow/core/kernels/cwise_op_logical_not.cc
View File

@ -19,8 +19,7 @@ namespace tensorflow {
REGISTER_KERNEL_BUILDER(Name("LogicalNot").Device(DEVICE_CPU),
UnaryOp<CPUDevice, functor::logical_not>);
#if GOOGLE_CUDA || TENSORFLOW_USE_ROCM
#if !defined(MLIR_GENERATED_GPU_KERNELS_ENABLED) || \
!defined(MLIR_GENERATED_EXPERIMENTAL_GPU_KERNELS_ENABLED)
#if !defined(MLIR_GENERATED_GPU_KERNELS_ENABLED)
REGISTER_KERNEL_BUILDER(Name("LogicalNot").Device(DEVICE_GPU),
UnaryOp<GPUDevice, functor::logical_not>);
#endif

3
tensorflow/core/kernels/cwise_op_neg_1.cc
View File

@ -20,8 +20,7 @@ REGISTER4(UnaryOp, CPU, "Neg", functor::neg, int8, int16, int32, int64);
#if GOOGLE_CUDA || TENSORFLOW_USE_ROCM
#if !defined(MLIR_GENERATED_GPU_KERNELS_ENABLED) || \
!defined(MLIR_GENERATED_EXPERIMENTAL_GPU_KERNELS_ENABLED)
#if !defined(MLIR_GENERATED_GPU_KERNELS_ENABLED)
REGISTER3(UnaryOp, GPU, "Neg", functor::neg, int8, int16, int64);
#endif

9
tensorflow/core/kernels/cwise_op_neg_2.cc
View File

@ -20,12 +20,9 @@ REGISTER6(UnaryOp, CPU, "Neg", functor::neg, Eigen::half, float, double,
bfloat16, complex64, complex128);
#if GOOGLE_CUDA || TENSORFLOW_USE_ROCM
#if !defined(MLIR_GENERATED_GPU_KERNELS_ENABLED) || \
!defined(MLIR_GENERATED_EXPERIMENTAL_GPU_KERNELS_ENABLED)
REGISTER6(UnaryOp, GPU, "Neg", functor::neg, Eigen::half, float, double,
bfloat16, complex64, complex128);
#else
#if !defined(MLIR_GENERATED_GPU_KERNELS_ENABLED)
REGISTER3(UnaryOp, GPU, "Neg", functor::neg, Eigen::half, float, double);
#endif
REGISTER3(UnaryOp, GPU, "Neg", functor::neg, bfloat16, complex64, complex128);
#endif
#endif
} // namespace tensorflow

6
tensorflow/core/kernels/mlir_generated/BUILD
View File

@ -51,6 +51,9 @@ filegroup(
"gpu_op_erfc.cc",
"gpu_op_floor.cc",
"gpu_op_imag.cc",
"gpu_op_invert.cc",
"gpu_op_logical_not.cc",
"gpu_op_neg.cc",
"gpu_op_real.cc",
"gpu_op_sqrt.cc",
"gpu_op_square.cc",
@ -69,15 +72,12 @@ filegroup(
"gpu_op_asinh.cc",
"gpu_op_exp.cc",
"gpu_op_expm1.cc",
"gpu_op_invert.cc",
"gpu_op_is_finite.cc",
"gpu_op_is_inf.cc",
"gpu_op_is_nan.cc",
"gpu_op_lgamma.cc",
"gpu_op_log.cc",
"gpu_op_log1p.cc",
"gpu_op_logical_not.cc",
"gpu_op_neg.cc",
"gpu_op_rsqrt.cc",
"gpu_op_sign.cc",
"gpu_op_sin.cc",