This file contains the porting status of GPU kernels in tensorflow/core/kernels directory on ROCm platform.

Legend:

• O : ROCm ready
• T : tuned with FIXME
• X : NOT ROCm ready
• P : in progress
• • : Issue opened

Kernels under tensorflow/core/kernels

status	memo	files
O		aggregate_ops_gpu.cu.cc
O		argmax_op.cc
O		argmax_op_gpu.cu.cc
O		avgpooling_op.cc
O		avgpooling_op_gpu.cu.cc
O		batch_kernels.cc
O		batch_norm_op.cc
O		batch_norm_op_gpu.cu.cc
O		batching_util/concat_split_util.h
O		batchtospace_op.cc
O		betainc_op.cc
O		bias_op.cc
O		bias_op_gpu.cu.cc
O		bincount_op.cc
O		bincount_op_gpu.cu.cc
O		broadcast_to_op.cc
O		bucketize_op.cc
O		bucketize_op_gpu.cu.cc
O		cast_op.cc
O		cast_op_gpu.cu.cc
O		check_numerics_op.cc
O		check_numerics_op_gpu.cu.cc
O		collective_nccl_test.cc
O		collective_ops.cc
O		concat_lib_gpu_impl.cu.cc
O		concat_op.cc
O		constant_op_test.cc
O		control_flow_ops.cc
O		conv_2d.h
O		conv_2d_gpu.h
O		conv_2d_gpu_double.cu.cc
O		conv_2d_gpu_float.cu.cc
O		conv_2d_gpu_half.cu.cc
O		conv_2d_gpu_int.cu.cc
O		conv_2d_gpu_int_spatial_convolution.cu.cc
O		conv_2d_gpu_int_spatial_convolution_backward.cu.cc
O		conv_2d_gpu_uint16.cu.cc
O		conv_2d_gpu_uint32.cu.cc
O		conv_2d_gpu_uint64.cu.cc
O		conv_2d_gpu_uint8.cu.cc
O		conv_grad_filter_ops.cc
O		conv_grad_input_ops.cc
O		conv_grad_ops_3d.cc
O		conv_ops.cc
O		conv_ops_3d.cc
O		conv_ops_fused_impl.h
O		conv_ops_test.cc
O		cross_op.cc
O		cross_op_gpu.cu.cc
O		ctc_loss_op.cc
O		cudnn_rnn_ops.cc
O		cwise_op_abs.cc
O		cwise_op_add_1.cc
O		cwise_op_clip.cc
O		cwise_op_clip_gpu.cu.cc
O		cwise_op_conj.cc
O		cwise_op_div.cc
O		cwise_op_equal_to_1.cc
O		cwise_op_floor_div.cc
O		cwise_op_floor_mod.cc
O		cwise_op_fma.cc
O		cwise_op_gpu_abs.cu.cc
O		cwise_op_gpu_acos.cu.cc
O		cwise_op_gpu_acosh.cu.cc
O		cwise_op_gpu_add.cu.cc
O		cwise_op_gpu_arg.cu.cc
O		cwise_op_gpu_asin.cu.cc
O		cwise_op_gpu_asinh.cu.cc
O		cwise_op_gpu_atan.cu.cc
O		cwise_op_gpu_atan2.cu.cc
O		cwise_op_gpu_atanh.cu.cc
O		cwise_op_gpu_bitwise_and.cu.cc
O		cwise_op_gpu_bitwise_or.cu.cc
O		cwise_op_gpu_bitwise_xor.cu.cc
O		cwise_op_gpu_ceil.cu.cc
O		cwise_op_gpu_complex.cu.cc
O		cwise_op_gpu_conj.cu.cc
O		cwise_op_gpu_cos.cu.cc
O		cwise_op_gpu_cosh.cu.cc
O		cwise_op_gpu_digamma.cu.cc
O		cwise_op_gpu_div.cu.cc
O		cwise_op_gpu_equal_to.cu.cc
O		cwise_op_gpu_erf.cu.cc
O		cwise_op_gpu_erfc.cu.cc
O		cwise_op_gpu_exp.cu.cc
O		cwise_op_gpu_expm1.cu.cc
O		cwise_op_gpu_floor.cu.cc
O		cwise_op_gpu_floor_div.cu.cc
O		cwise_op_gpu_floor_mod.cu.cc
O		cwise_op_gpu_fma.cu.cc
O		cwise_op_gpu_greater.cu.cc
O		cwise_op_gpu_greater_equal.cu.cc
O		cwise_op_gpu_igammas_double.cu.cc
O		cwise_op_gpu_igammas_float.cu.cc
O		cwise_op_gpu_imag.cu.cc
O		cwise_op_gpu_inverse.cu.cc
O		cwise_op_gpu_invert.cu.cc
O		cwise_op_gpu_isfinite.cu.cc
O		cwise_op_gpu_isinf.cu.cc
O		cwise_op_gpu_isnan.cu.cc
O		cwise_op_gpu_left_shift.cu.cc
O		cwise_op_gpu_less.cu.cc
O		cwise_op_gpu_less_equal.cu.cc
O		cwise_op_gpu_lgamma.cu.cc
O		cwise_op_gpu_log.cu.cc
O		cwise_op_gpu_log1p.cu.cc
O		cwise_op_gpu_logical_and.cu.cc
O		cwise_op_gpu_logical_not.cu.cc
O		cwise_op_gpu_logical_or.cu.cc
O		cwise_op_gpu_maximum.cu.cc
O		cwise_op_gpu_minimum.cu.cc
O		cwise_op_gpu_mod.cu.cc
O		cwise_op_gpu_mul.cu.cc
O		cwise_op_gpu_neg.cu.cc
O		cwise_op_gpu_not_equal_to.cu.cc
O		cwise_op_gpu_pow.cu.cc
O		cwise_op_gpu_random_grad.cu.cc
O		cwise_op_gpu_real.cu.cc
O		cwise_op_gpu_right_shift.cu.cc
O		cwise_op_gpu_rint.cu.cc
O		cwise_op_gpu_round.cu.cc
O		cwise_op_gpu_rsqrt.cu.cc
O		cwise_op_gpu_select.cu.cc
O		cwise_op_gpu_sigmoid.cu.cc
O		cwise_op_gpu_sign.cu.cc
O		cwise_op_gpu_sin.cu.cc
O		cwise_op_gpu_sinh.cu.cc
O		cwise_op_gpu_sqrt.cu.cc
O		cwise_op_gpu_square.cu.cc
O		cwise_op_gpu_squared_difference.cu.cc
O		cwise_op_gpu_sub.cu.cc
O		cwise_op_gpu_tan.cu.cc
O		cwise_op_gpu_tanh.cu.cc
O		cwise_op_gpu_xdivy.cu.cc
O		cwise_op_gpu_xlog1py.cu.cc
O		cwise_op_gpu_xlogy.cu.cc
O		cwise_op_gpu_zeta.cu.cc
O		cwise_op_greater.cc
O		cwise_op_greater_equal.cc
O		cwise_op_less.cc
O		cwise_op_less_equal.cc
O		cwise_op_logical_and.cc
O		cwise_op_logical_not.cc
O		cwise_op_logical_or.cc
O		cwise_op_maximum.cc
O		cwise_op_minimum.cc
O		cwise_op_mod.cc
O		cwise_op_mul_1.cc
O		cwise_op_ndtri_double.cu.cc
O		cwise_op_ndtri_float.cu.cc
O		cwise_op_neg_1.cc
O		cwise_op_not_equal_to_1.cc
O		cwise_op_select.cc
O		cwise_op_sign.cc
O		cwise_op_square.cc
O		cwise_op_squared_difference.cc
O		cwise_op_sub.cc
O		data/experimental/sleep_dataset_op.cc
O		data/finalize_dataset_op.cc
O		data/generator_dataset_op.cc
O		data/get_options_op.cc
O		data/iterator_ops.cc
O		data/map_dataset_op.cc
O		data/optional_ops.cc
O		data/optional_ops.cu.cc
O		data/options_dataset_op.cc
O		data/prefetch_dataset_op.cc
O		data_format_ops.cc
O		data_format_ops_gpu.cu.cc
O		debug_ops.cc
O		debug_ops.h
O		debug_ops_gpu.cu.cc
O		dense_update_functor_gpu.cu.cc
O		dense_update_ops.cc
O		depthtospace_op.cc
O		depthtospace_op_gpu.cu.cc
O		depthwise_conv_grad_op.cc
O		depthwise_conv_op.cc
O		depthwise_conv_op_gpu_double.cu.cc
O		depthwise_conv_op_gpu_float.cu.cc
O		depthwise_conv_ops_test.cc
O		diag_op.cc
O		diag_op_gpu.cu.cc
O		dilation_ops.cc
O		dilation_ops_gpu.cu.cc
O		dropout_op.cc
O		dropout_op_gpu.cu.cc
O		dynamic_partition_op_gpu.cu.cc
O		dynamic_partition_op_gpu.cu.cc
O		dynamic_stitch_op.cc
O		dynamic_stitch_op_gpu.cu.cc
O		fact_op.cc
O		fake_quant_ops.cc
O		fake_quant_ops_gpu.cu.cc
O		fake_quant_ops_test.cc
O		fft_ops.cc
O		fill_functor.cu.cc
O		fused_batch_norm_ex_op_test.cc
O		fused_batch_norm_op.cc
O		fused_batch_norm_op.cu.cc
O		gather_functor_batched_gpu.cu.cc
O		gather_functor_gpu.cu.cc
O		gather_nd_op_gpu.cu.cc
O		gpu_fusion_ops_batchnormactv.cc
O		gpu_fusion_ops_convbiasactv.cc
O		gpu_fusion_ops_custom.cc
O		gpu_fusion_ops_custom.cu.cc
O		gpu_prim_helpers_test.cu.cc
O		gpu_prim_helpers_test.cu.cc
O		histogram_op.cc
O		histogram_op_gpu.cu.cc
O		identity_op.cc
O		image/adjust_contrast_op.cc
O		image/adjust_contrast_op_gpu.cu.cc
O		image/adjust_hsv_gpu.cu.h
O		image/adjust_hue_op.cc
O		image/adjust_hue_op_gpu.cu.cc
O		image/adjust_saturation_op.cc
O		image/colorspace_op.cc
O		image/colorspace_op_gpu.cu.cc
O		image/crop_and_resize_op.cc
O		image/crop_and_resize_op_gpu.cu.cc
O		image/extract_image_patches_op.cc
O		image/extract_volume_patches_op.cc
O		image/extract_volume_patches_op_gpu.cu.cc
O		image/generate_box_proposals_op.cu.cc
O		image/image_ops.cc:
O		image/mirror_pad_op.cc
O		image/mirror_pad_op_gpu.cu.cc
O		image/non_max_suppression_op.cu.cc
O		image/non_max_suppression_op_gpu_test.cc
O		image/resize_bilinear_op.cc
O		image/resize_bilinear_op_gpu.cu.cc
O		image/resize_bilinear_op_test.cc
O		image/resize_nearest_neighbor_op.cc
O		image/resize_nearest_neighbor_op_gpu.cu.cc
O		image/resize_nearest_neighbor_op_test.cc
O		in_topk_op.cc
O		in_topk_op_gpu.cu.cc
O		inplace_ops.cc
O		inplace_ops_functor_gpu.cu.cc
O		l2loss_op_gpu.cu.cc
O		linalg/cholesky_op_gpu.cu.cc
X		linalg/determinant_op.cc
X		linalg/determinant_op_gpu.cu.cc
O		linalg/einsum_op_gpu.cu.cc
O		linalg/einsum_op_impl_complex128.cc
O		linalg/einsum_op_impl_complex64.cc
O		linalg/einsum_op_impl_double.cc
O		linalg/einsum_op_impl_float.cc
O		linalg/einsum_op_impl_half.cc
O		linalg/eye_functor_gpu.cu.cc
O		linalg/linalg_ops_common.h
O		linalg/lu_op_gpu.cu.cc
O		linalg/matrix_band_part_op.cc
O		linalg/matrix_band_part_op_gpu.cu.cc
O		linalg/matrix_diag_op.cc
O		linalg/matrix_diag_op_gpu.cu.cc
O		linalg/matrix_inverse_op.cc
O		linalg/matrix_set_diag_op.cc
O		linalg/matrix_solve_op.cc
O		linalg/matrix_triangular_solve_op_complex.cc
O		linalg/matrix_triangular_solve_op_impl.h
O		linalg/matrix_triangular_solve_op_real.cc
*		linalg/qr_op_complex128.cc
*		linalg/qr_op_complex64.cc
*		linalg/qr_op_double.cc
*		linalg/qr_op_float.cc
*		linalg/qr_op_half.cc
O		linalg/self_adjoint_eig_v2_op_gpu.cc
*		linalg/svd_op_gpu.cu.cc
*		linalg/svd_op_gpu.cu.cc
O		linalg/tridiagonal_matmul_op_gpu.cu.cc
O		linalg/tridiagonal_solve_op_gpu.cu.cc
O		list_kernels.cc
O		list_kernels.cu.cc
O		lrn_op.cc
O		matmul_op_fused.cc
O		matmul_op_impl.h
O		matmul_op_test.cc
O		maxpooling_op.cc
O		maxpooling_op_gpu.cu.cc
O		mlir_generated/base_gpu_op.h
O		mlir_generated/base_op.cc
O		mlir_generated/base_op.h
O		mlir_generated/gpu_op_abs.cc
O		mlir_generated/gpu_op_acos.cc
O		mlir_generated/gpu_op_acosh.cc
O		mlir_generated/gpu_op_add.cc
O		mlir_generated/gpu_op_angle.cc
O		mlir_generated/gpu_op_asin.cc
O		mlir_generated/gpu_op_asinh.cc
O		mlir_generated/gpu_op_atan.cc
O		mlir_generated/gpu_op_atan2.cc
O		mlir_generated/gpu_op_atanh.cc
O		mlir_generated/gpu_op_bitwise_and.cc
O		mlir_generated/gpu_op_bitwise_or.cc
O		mlir_generated/gpu_op_bitwise_xor.cc
O		mlir_generated/gpu_op_cast.cc
O		mlir_generated/gpu_op_ceil.cc
O		mlir_generated/gpu_op_complex.cc
O		mlir_generated/gpu_op_complex_abs.cc
O		mlir_generated/gpu_op_conj.cc
O		mlir_generated/gpu_op_cos.cc
O		mlir_generated/gpu_op_cosh.cc
O		mlir_generated/gpu_op_digamma.cc
O		mlir_generated/gpu_op_div.cc
O		mlir_generated/gpu_op_div_no_nan.cc
O		mlir_generated/gpu_op_elu.cc
O		mlir_generated/gpu_op_equal.cc
O		mlir_generated/gpu_op_erf.cc
O		mlir_generated/gpu_op_erfc.cc
O		mlir_generated/gpu_op_exp.cc
O		mlir_generated/gpu_op_expm1.cc
O		mlir_generated/gpu_op_floor.cc
O		mlir_generated/gpu_op_floor_div.cc
O		mlir_generated/gpu_op_floor_mod.cc
O		mlir_generated/gpu_op_greater.cc
O		mlir_generated/gpu_op_greater_equal.cc
O		mlir_generated/gpu_op_imag.cc
O		mlir_generated/gpu_op_invert.cc
O		mlir_generated/gpu_op_is_finite.cc
O		mlir_generated/gpu_op_is_inf.cc
O		mlir_generated/gpu_op_is_nan.cc
O		mlir_generated/gpu_op_left_shift.cc
O		mlir_generated/gpu_op_less.cc
O		mlir_generated/gpu_op_less_equal.cc
O		mlir_generated/gpu_op_lgamma.cc
O		mlir_generated/gpu_op_log.cc
O		mlir_generated/gpu_op_log1p.cc
O		mlir_generated/gpu_op_logical_and.cc
O		mlir_generated/gpu_op_logical_not.cc
O		mlir_generated/gpu_op_logical_or.cc
O		mlir_generated/gpu_op_maximum.cc
O		mlir_generated/gpu_op_minimum.cc
O		mlir_generated/gpu_op_mul.cc
O		mlir_generated/gpu_op_mul_no_nan.cc
O		mlir_generated/gpu_op_neg.cc
O		mlir_generated/gpu_op_next_after.cc
O		mlir_generated/gpu_op_not_equal.cc
O		mlir_generated/gpu_op_ones_like.cc
O		mlir_generated/gpu_op_polygamma.cc
O		mlir_generated/gpu_op_pow.cc
O		mlir_generated/gpu_op_real.cc
O		mlir_generated/gpu_op_reciprocal.cc
O		mlir_generated/gpu_op_relu.cc
O		mlir_generated/gpu_op_right_shift.cc
O		mlir_generated/gpu_op_rint.cc
O		mlir_generated/gpu_op_round.cc
O		mlir_generated/gpu_op_rsqrt.cc
O		mlir_generated/gpu_op_select.cc
O		mlir_generated/gpu_op_selu.cc
O		mlir_generated/gpu_op_sigmoid.cc
O		mlir_generated/gpu_op_sign.cc
O		mlir_generated/gpu_op_sin.cc
O		mlir_generated/gpu_op_sinh.cc
O		mlir_generated/gpu_op_softplus.cc
O		mlir_generated/gpu_op_softsign.cc
O		mlir_generated/gpu_op_sqrt.cc
O		mlir_generated/gpu_op_square.cc
O		mlir_generated/gpu_op_squared_difference.cc
O		mlir_generated/gpu_op_sub.cc
O		mlir_generated/gpu_op_tan.cc
O		mlir_generated/gpu_op_tanh.cc
O		mlir_generated/gpu_op_truncate_div.cc
O		mlir_generated/gpu_op_xdivy.cc
O		mlir_generated/gpu_op_xlog1py.cc
O		mlir_generated/gpu_op_xlogy.cc
O		mlir_generated/gpu_op_zeros_like.cc
O		mlir_generated/gpu_op_zeta.cc
O		multinomial_op.cc
O		multinomial_op_gpu.cu.cc
O		nccl_ops.cc
O		nextafter_op_gpu.cu.cc
O		one_hot_op.cc
O		one_hot_op_gpu.cu.cc
O		ops_testutil.cc
O		pack_op.cc
O		pad_op.cc
O		pad_op_gpu.cu.cc
O		parameterized_truncated_normal_op.cc
O		parameterized_truncated_normal_op_gpu.cu.cc
O		pooling_ops_3d_gpu.cu.cc
O		population_count_op.cc
O		population_count_op_gpu.cu.cc
O		quantize_and_dequantize_op.cc
O		quantize_and_dequantize_op_gpu.cu.cc
O		random_op.cc
O		random_op_gpu.cu.cc
O		reduction_ops_all.cc
O		reduction_ops_any.cc
O		reduction_ops_euclidean.cc
O		reduction_ops_gpu_bool.cu.cc
O		reduction_ops_gpu_complex128.cu.cc
O		reduction_ops_gpu_complex64.cu.cc
O		reduction_ops_gpu_double.cu.cc
O		reduction_ops_gpu_float.cu.cc
O		reduction_ops_gpu_int.cu.cc
O		reduction_ops_half_mean_sum.cu.cc
O		reduction_ops_half_prod_max_min.cu.cc
O		reduction_ops_max.cc
O		reduction_ops_mean.cc
O		reduction_ops_min.cc
O		reduction_ops_prod.cc
O		reduction_ops_sum.cc
O		relu_op.cc
O		relu_op_gpu.cu.cc
O		reshape_op.cc
O		reshape_util_gpu.cu.cc
O		resource_variable_ops.cc
O		reverse_op.cc
O		reverse_op_gpu.cu.cc
O		reverse_sequence_op.cc
O		reverse_sequence_op_gpu.cu.cc
O		rnn/gru_ops.cc
O		rnn/gru_ops_gpu.cu.cc
O		rnn/lstm_ops.cc
O		rnn/lstm_ops_gpu.cc
O		roll_op.cc
O		roll_op_gpu.cu.cc
O		scan_ops.cc
O		scan_ops_gpu_double.cu.cc
O		scan_ops_gpu_float.cu.cc
O		scan_ops_gpu_half.cu.cc
O		scan_ops_gpu_int.cu.cc
O		scatter_functor_gpu.cu.cc
O		scatter_nd_op_gpu.cu.cc
O		scatter_op_gpu.cu.cc
O		scoped_allocator_ops.cc
O		searchsorted_op.cc
O		searchsorted_op_gpu.cu.cc
O		segment_reduction_ops_gpu_0.cu.cc
O		segment_reduction_ops_gpu_1.cu.cc
O		segment_reduction_ops_gpu_2.cu.cc
O		segment_reduction_ops_impl_1.cc
O		segment_reduction_ops_impl_2.cc
O		segment_reduction_ops_impl_3.cc
O		segment_reduction_ops_impl_4.cc
O		segment_reduction_ops_impl_5.cc
O		sequence_ops.cc
O		sequence_ops_gpu.cu.cc
O		shape_ops.cc
O		slice_op.cc:
O		slice_op_gpu.cu.cc
O		snapshot_op.cc
O		snapshot_op_gpu.cu.cc
O		softmax_op_gpu.cu.cc
O		softmax_op_gpu.cu.cc
O		softplus_op.cc
O		softplus_op_gpu.cu.cc
O		softsign_op.cc
O		softsign_op_gpu.cu.cc
O		spacetobatch_functor_gpu.cu.cc
O		spacetobatch_op.cc
O		spacetodepth_op.cc
O		spacetodepth_op_gpu.cu.cc
O		sparse/add_op.cc
O		sparse/conj_op.cc
O		sparse/csr_sparse_matrix_to_dense_op.cc
O		sparse/csr_sparse_matrix_to_sparse_tensor_op.cc
O		sparse/dense_to_csr_sparse_matrix_op.cc
O		sparse/kernels_gpu.cu.cc
O		sparse/mat_mul_op.cc
O		sparse/mul_op.cc
O		sparse/nnz_op.cc
O		sparse/softmax_op.cc
O		sparse/sparse_mat_mul_op.cc
O		sparse/sparse_matrix.cc
O		sparse/sparse_matrix_components_op.cc
O		sparse/sparse_tensor_to_csr_sparse_matrix_op.cc
O		sparse/transpose_op.cc
O		sparse/zeros_op.cc
O		sparse/zeros_op.cc
O		sparse/zeros_op.h
O		sparse_concat_op.cc
O		sparse_concat_op_gpu.cu.cc
O		sparse_fill_empty_rows_op.cc
O		sparse_fill_empty_rows_op_gpu.cu.cc
O		sparse_reorder_op.cc
O		sparse_reorder_op_gpu.cu.cc
O		sparse_reshape_op.cc
O		sparse_slice_grad_op.cc
O		sparse_slice_grad_op_gpu.cu.cc
O		sparse_slice_op.cc
O		sparse_slice_op_gpu.cu.cc
O		sparse_split_op.cc
O		sparse_split_op_gpu.cu.cc
O		sparse_tensor_dense_matmul_op.cc
O		sparse_tensor_dense_matmul_op_gpu.cu.cc
O		sparse_to_dense_op.cc
O		sparse_to_dense_op_gpu.cu.cc
O		sparse_xent_op_gpu.cu.cc
O		specialmath/special_math_op_bessel.cc
O		specialmath/special_math_op_dawsn.cc
O		specialmath/special_math_op_expint.cc
O		specialmath/special_math_op_fresnel_cos.cc
O		specialmath/special_math_op_fresnel_sin.cc
O		specialmath/special_math_op_gpu_bessel.cu.cc
O		specialmath/special_math_op_gpu_dawsn.cu.cc
O		specialmath/special_math_op_gpu_fresnel.cu.cc
O		specialmath/special_math_op_gpu_spence.cu.cc
O		specialmath/special_math_op_spence.cc
O		specialmeth/special_math_op_gpu_expint.cu.cc
O		split_lib_gpu.cu.cc
O		split_op.cc
O		split_v_op.cc
O		stateful_random_ops_gpu.cu.cc
O		stateless_random_gamma_op.cc
O		stateless_random_gamma_op_gpu.cu.cc
O		strided_slice_op.cc
O		strided_slice_op_gpu_bool.cu.cc
O		strided_slice_op_gpu_complex.cu.cc
O		strided_slice_op_gpu_int.cu.cc
O		strided_slice_op_gpu_number_types.cu.cc
O		tensor_array_ops.cc
O		tensor_to_hash_bucket_op.cc
O		tensor_to_hash_bucket_op_gpu.cu.cc
O		tile_functor_gpu_bool.cu.cc
O		tile_functor_gpu_complex128.cu.cc
O		tile_functor_gpu_complex64.cu.cc
O		tile_functor_gpu_double.cu.cc
O		tile_functor_gpu_float.cu.cc
O		tile_functor_gpu_half.cu.cc
O		tile_functor_gpu_int16.cu.cc
O		tile_functor_gpu_int32.cu.cc
O		tile_functor_gpu_int64.cu.cc
O		tile_ops.cc
O		tile_ops_gpu_impl_1.cu.cc
O		tile_ops_gpu_impl_2.cu.cc
O		tile_ops_gpu_impl_3.cu.cc
O		tile_ops_gpu_impl_4.cu.cc
O		tile_ops_gpu_impl_5.cu.cc
O		tile_ops_gpu_impl_6.cu.cc
O		tile_ops_gpu_impl_7.cu.cc
O		tile_ops_gpu_impl_8.cu.cc
O		topk_op.cc
O		topk_op_gpu_double.cu.cc
O		topk_op_gpu_float.cu.cc
O		topk_op_gpu_half.cu.cc
O		topk_op_gpu_int16.cu.cc
O		topk_op_gpu_int32.cu.cc
O		topk_op_gpu_int64.cu.cc
O		topk_op_gpu_int8.cu.cc
O		topk_op_gpu_uint16.cu.cc
O		topk_op_gpu_uint32.cu.cc
O		topk_op_gpu_uint64.cu.cc
O		topk_op_gpu_uint8.cu.cc
O		training_ops_gpu.cu.cc
O		transpose_functor_gpu.cu.cc
O		transpose_op.cc
O		unique_op_gpu_0.cu.cc
O		unique_op_gpu_1.cu.cc
O		unique_op_gpu_2.cu.cc
O		unique_op_gpu_3.cu.cc
O		unpack_op.cc
O		variable_ops.cc
O		where_op.cc
O		where_op_gpu_impl_1.cu.cc
O		where_op_gpu_impl_2.cu.cc
O		where_op_gpu_impl_3.cu.cc
O		where_op_gpu_impl_4.cu.cc
O		where_op_gpu_impl_5.cu.cc
O		where_op_gpu_impl_6.cu.cc
O		where_op_gpu_impl_7.cu.cc
O		where_op_gpu_impl_8.cu.cc
O		xent_op.cc
O		xent_op_gpu.cu.cc`