From a49eb8085d60eb6ef2e8dd6082a8409c9732b92b Mon Sep 17 00:00:00 2001 From: root Date: Tue, 3 Jun 2025 12:00:03 +0800 Subject: [PATCH 1/2] add embedding op and softmax op --- .gitignore | 3 + serving/.bazelrc | 4 + tensorflow/.bazelrc | 5 + .../ci/official/code_check_changed_files.sh | 0 .../builder.devtoolset/build_devtoolset.sh | 0 .../builder.devtoolset/fixlinks_aarch64.sh | 0 .../builder.devtoolset/rpm-patch.sh | 0 .../builder.patchelf/build_patchelf.sh | 0 .../devel.usertools/get_test_list.sh | 0 .../devel.usertools/repack_libtensorflow.sh | 0 .../devel.usertools/squash_testlogs.py | 0 .../containers/linux_arm64/setup.packages.sh | 0 .../ci/official/utilities/cleanup_docker.sh | 0 .../official/utilities/generate_index_html.sh | 0 tensorflow/configure | 0 .../c/experimental/ops/update_cpp_ops.sh | 0 tensorflow/tensorflow/c/generate-pc.sh | 0 .../half_plus_two/00000123/saved_model.pb | Bin .../variables/variables.data-00000-of-00001 | Bin .../00000123/variables/variables.index | Bin .../00000123/saved_model.pbtxt | 0 .../variables/variables.data-00000-of-00001 | Bin .../00000123/variables/variables.index | Bin .../tests/reducer/unsupported-op-test.sh | 0 .../tf2tensorrt/utils/trt_engine_utils.cc | 0 tensorflow/tensorflow/core/BUILD | 7 + .../tensorflow/core/api_def/update_api_def.sh | 0 tensorflow/tensorflow/core/kernels/BUILD | 4 + .../tensorflow/core/kernels/ktfop/BUILD | 36 +++++ .../core/kernels/ktfop/embedding_lookup_op.cc | 149 ++++++++++++++++++ .../tensorflow/core/kernels/ktfop/softmax.cc | 37 +++++ .../core/kernels/mlir_generated/build_test.sh | 0 tensorflow/tensorflow/core/ops/BUILD | 8 +- tensorflow/tensorflow/core/ops/ktfop_ops.cc | 66 ++++++++ .../multibox_detector/data/surfers.jpg | Bin tensorflow/tensorflow/go/test.sh | 0 .../prev_is_different_than_current_test.sh | 0 .../lite/delegates/gpu/cl/run_tests.sh | 0 .../gpu/cl/testing/run_delegate_testing.sh | 0 .../cl/testing/run_internal_api_samples.sh | 0 .../cl/testing/run_memory_sharing_sample.sh | 0 .../dummy_delegate/external_delegate_test.sh | 0 .../lite/examples/ios/download_models.sh | 0 .../models/blazeface_mlkit_v1.tfl | Bin .../tensorflow/lite/ios/build_frameworks.sh | 0 .../lite/ios/hide_symbols_with_allowlist.sh | 0 ...hide_xcframework_symbols_with_allowlist.sh | 0 .../lite/java/build_aar_for_release.sh | 0 tensorflow/tensorflow/lite/java/demo/gradlew | 0 .../tensorflow/lite/java/ovic/demo/gradlew | 0 .../lite/lib_package/concat_licenses.sh | 0 .../python/convert_file_to_c_source_test.sh | 0 .../scripts/download_frameworks.sh | 0 tensorflow/tensorflow/lite/testdata/sub.bin | Bin .../ios/build_benchmark_framework.sh | 0 .../lite/tools/cmake/download_toolchains.sh | 0 .../test_invalid_settings.json | 0 .../test_missing_delegate_path_settings.json | 0 ...test_missing_stable_delegate_settings.json | 0 .../test_sample_stable_delegate_settings.json | 0 .../test_stable_xnnpack_settings.json | 0 .../lite/tools/pip_package/debian/rules | 0 .../lite/tools/pip_package/update_sources.sh | 0 .../lite/tools/pip_package/with_the_same_user | 0 .../autograph/pyct/inspect_utils_test.sh | 0 .../testdata/gen_tf_readvariableop_model.py | 0 .../v1/examples_v1_debug_errors_test.sh | 0 .../v1/examples_v1_debug_fibonacci_test.sh | 0 .../v1/examples_v1_debug_keras_test.sh | 0 .../v1/examples_v1_debug_mnist_test.sh | 0 .../v1/examples_v1_offline_analyzer_test.sh | 0 .../debug/examples/v1/examples_v1_test.sh | 0 .../debug/examples/v2/examples_v2_test.sh | 0 .../tools/no_xla_multithread_symbols_test.sh | 0 .../tensorflow/python/tools/skip_test.sh | 0 .../tools/xla_multithread_symbols_test.sh | 0 .../client/pip_package/build_pip_package.sh | 0 .../profiler/pip_package/build_pip_package.sh | 0 .../tensorflow/tools/android/test/gradlew | 0 .../test/jni/object_tracking/gl_utils.h | 0 .../test/jni/object_tracking/utils_neon.cc | 0 .../tools/android/test/jni/rgb2yuv.cc | 0 .../tools/android/test/jni/rgb2yuv.h | 0 .../tensorflow/tools/ci_build/build_rbe.sh | 0 .../tools/ci_build/builds/android.sh | 0 .../tools/ci_build/builds/android_full.sh | 0 .../tools/ci_build/builds/benchmark.sh | 0 .../ci_build/builds/check_system_libs.py | 0 .../tensorflow/tools/ci_build/builds/cmake.sh | 0 .../tools/ci_build/builds/configured | 0 .../tools/ci_build/builds/docker_cpu_pip.sh | 0 .../tools/ci_build/builds/docker_test.sh | 0 .../ci_build/builds/integration_tests.sh | 0 .../tools/ci_build/builds/libtensorflow.sh | 0 .../tools/ci_build/builds/pip_new.sh | 0 .../tools/ci_build/builds/print_build_info.sh | 0 .../tools/ci_build/builds/py_test_delegate.sh | 0 .../tools/ci_build/builds/run_pip_tests.sh | 0 .../tools/ci_build/builds/test_tutorials.sh | 0 .../tools/ci_build/builds/test_user_ops.sh | 0 .../tools/ci_build/builds/with_the_same_user | 0 .../tensorflow/tools/ci_build/ci_build.sh | 0 .../tools/ci_build/code_link_check.sh | 0 .../tensorflow/tools/ci_build/copy_binary.py | 0 .../ci_build/devtoolset/build_devtoolset.sh | 0 .../tools/ci_build/devtoolset/fixlinks.sh | 0 .../tools/ci_build/devtoolset/rpm-patch.sh | 0 .../gpu_build/parallel_gpu_execute.sh | 0 .../install/build_and_install_python.sh | 0 .../ci_build/install/install_auditwheel.sh | 0 .../install/install_bootstrap_deb_packages.sh | 0 .../ci_build/install/install_buildifier.sh | 0 .../install/install_buildifier_from_source.sh | 0 .../install/install_centos_python36.sh | 0 .../tools/ci_build/install/install_clang.sh | 0 .../tools/ci_build/install/install_cmake.sh | 0 .../tools/ci_build/install/install_gcc6.sh | 0 .../tools/ci_build/install/install_golang.sh | 0 .../ci_build/install/install_golang_centos.sh | 0 .../install/install_golang_ppc64le.sh | 0 .../ci_build/install/install_latest_clang.sh | 0 .../tools/ci_build/install/install_mpi.sh | 0 .../install/install_openblas_ppc64le.sh | 0 .../install/install_pi_python3.9_toolchain.sh | 0 .../install/install_pi_python3x_toolchain.sh | 0 .../ci_build/install/install_pi_toolchain.sh | 0 .../install/install_pip_packages_remote.sh | 0 .../tools/ci_build/install/install_proto3.sh | 0 .../install/install_python3.6_pip_packages.sh | 0 .../ci_build/install/install_python37.sh | 0 .../ci_build/install/install_yum_packages.sh | 0 .../tools/ci_build/linux/cmake/run.sh | 0 .../tools/ci_build/linux/cpu/run_cc_core.sh | 0 .../tools/ci_build/linux/cpu/run_mkl.sh | 0 .../tools/ci_build/linux/cpu/run_py2_core.sh | 0 .../ci_build/linux/cpu/run_py3_contrib.sh | 0 .../tools/ci_build/linux/cpu/run_py3_core.sh | 0 .../tools/ci_build/linux/gpu/run_cc_core.sh | 0 .../tools/ci_build/linux/gpu/run_py3_core.sh | 0 .../tools/ci_build/linux/libtensorflow.sh | 0 .../tools/ci_build/linux/libtensorflow_cpu.sh | 0 .../ci_build/linux/libtensorflow_docker.sh | 0 .../tools/ci_build/linux/libtensorflow_gpu.sh | 0 .../ci_build/linux/libtensorflow_rocm.sh | 0 .../ci_build/linux/mkl/Dockerfile.devel-mkl | 0 .../ci_build/linux/mkl/basic-mkl-gpu-test.sh | 0 .../ci_build/linux/mkl/basic-mkl-test.sh | 0 .../ci_build/linux/mkl/build-dev-container.sh | 0 .../linux/mkl/install_openmpi_horovod.sh | 0 .../ci_build/linux/ppc64le/cpu/run_py2.sh | 0 .../ci_build/linux/ppc64le/cpu/run_py3.sh | 0 .../ci_build/linux/ppc64le/gpu/run_py2.sh | 0 .../ci_build/linux/ppc64le/gpu/run_py3.sh | 0 .../ci_build/linux/rocm/rocm_py310_pip.sh | 0 .../ci_build/linux/rocm/rocm_py36_pip.sh | 0 .../ci_build/linux/rocm/rocm_py37_pip.sh | 0 .../ci_build/linux/rocm/rocm_py38_pip.sh | 0 .../ci_build/linux/rocm/rocm_py39_pip.sh | 0 .../ci_build/linux/rocm/run_gpu_multi.sh | 0 .../tools/ci_build/osx/cpu/run_contrib.sh | 0 .../tools/ci_build/osx/cpu/run_py2_cc_core.sh | 0 .../tools/ci_build/osx/cpu/run_py3_cc_core.sh | 0 .../tools/ci_build/osx/libtensorflow_cpu.sh | 0 .../tools/ci_build/osx/libtensorflow_gpu.sh | 0 .../tools/ci_build/osx/libtensorflow_rocm.sh | 0 .../tools/ci_build/pi/build_raspberry_pi.sh | 0 .../protobuf/protobuf_optimized_pip.sh | 0 .../tensorflow/tools/gcs_test/gcs_smoke.sh | 0 .../tools/gcs_test/gcs_smoke_wrapper.sh | 0 .../tools/lib_package/concat_licenses.sh | 0 .../lib_package/libtensorflow_java_test.sh | 0 .../builder.devtoolset/fixlinks.sh | 0 .../builder.devtoolset/rpm-patch.sh | 0 .../devel.usertools/get_test_list.sh | 0 .../devel.usertools/repack_libtensorflow.sh | 0 .../devel.usertools/squash_testlogs.py | 0 .../tf_sig_build_dockerfiles/setup.cuda.sh | 0 .../setup.packages.sh | 0 tensorflow/tensorflow/workspace2.bzl | 6 +- .../grpc/upb_gcc10_compile_fix.patch | 11 ++ tensorflow/third_party/kblas/BUILD | 21 +++ tensorflow/third_party/kblas/build_defs.bzl | 8 + tensorflow/third_party/kblas/kblas.BUILD | 12 ++ tensorflow/third_party/kblas/workspace.bzl | 10 ++ tensorflow/third_party/ktfop/BUILD | 21 +++ tensorflow/third_party/ktfop/build_defs.bzl | 8 + tensorflow/third_party/ktfop/ktfop.BUILD | 14 ++ tensorflow/third_party/ktfop/workspace.bzl | 10 ++ .../xla/.kokoro/generate_index_html.sh | 0 .../gpu_build/parallel_gpu_execute.sh | 0 190 files changed, 438 insertions(+), 2 deletions(-) create mode 100644 .gitignore mode change 100644 => 100755 tensorflow/ci/official/code_check_changed_files.sh mode change 100644 => 100755 tensorflow/ci/official/containers/linux_arm64/builder.devtoolset/build_devtoolset.sh mode change 100644 => 100755 tensorflow/ci/official/containers/linux_arm64/builder.devtoolset/fixlinks_aarch64.sh mode change 100644 => 100755 tensorflow/ci/official/containers/linux_arm64/builder.devtoolset/rpm-patch.sh mode change 100644 => 100755 tensorflow/ci/official/containers/linux_arm64/builder.patchelf/build_patchelf.sh mode change 100644 => 100755 tensorflow/ci/official/containers/linux_arm64/devel.usertools/get_test_list.sh mode change 100644 => 100755 tensorflow/ci/official/containers/linux_arm64/devel.usertools/repack_libtensorflow.sh mode change 100644 => 100755 tensorflow/ci/official/containers/linux_arm64/devel.usertools/squash_testlogs.py mode change 100644 => 100755 tensorflow/ci/official/containers/linux_arm64/setup.packages.sh mode change 100644 => 100755 tensorflow/ci/official/utilities/cleanup_docker.sh mode change 100644 => 100755 tensorflow/ci/official/utilities/generate_index_html.sh mode change 100644 => 100755 tensorflow/configure mode change 100644 => 100755 tensorflow/tensorflow/c/experimental/ops/update_cpp_ops.sh mode change 100644 => 100755 tensorflow/tensorflow/c/generate-pc.sh mode change 100644 => 100755 tensorflow/tensorflow/cc/saved_model/testdata/half_plus_two/00000123/saved_model.pb mode change 100644 => 100755 tensorflow/tensorflow/cc/saved_model/testdata/half_plus_two/00000123/variables/variables.data-00000-of-00001 mode change 100644 => 100755 tensorflow/tensorflow/cc/saved_model/testdata/half_plus_two/00000123/variables/variables.index mode change 100644 => 100755 tensorflow/tensorflow/cc/saved_model/testdata/half_plus_two_pbtxt/00000123/saved_model.pbtxt mode change 100644 => 100755 tensorflow/tensorflow/cc/saved_model/testdata/half_plus_two_pbtxt/00000123/variables/variables.data-00000-of-00001 mode change 100644 => 100755 tensorflow/tensorflow/cc/saved_model/testdata/half_plus_two_pbtxt/00000123/variables/variables.index mode change 100644 => 100755 tensorflow/tensorflow/compiler/mlir/tensorflow/tests/reducer/unsupported-op-test.sh mode change 100644 => 100755 tensorflow/tensorflow/compiler/tf2tensorrt/utils/trt_engine_utils.cc mode change 100644 => 100755 tensorflow/tensorflow/core/api_def/update_api_def.sh create mode 100644 tensorflow/tensorflow/core/kernels/ktfop/BUILD create mode 100644 tensorflow/tensorflow/core/kernels/ktfop/embedding_lookup_op.cc create mode 100644 tensorflow/tensorflow/core/kernels/ktfop/softmax.cc mode change 100644 => 100755 tensorflow/tensorflow/core/kernels/mlir_generated/build_test.sh create mode 100644 tensorflow/tensorflow/core/ops/ktfop_ops.cc mode change 100644 => 100755 tensorflow/tensorflow/examples/multibox_detector/data/surfers.jpg mode change 100644 => 100755 tensorflow/tensorflow/go/test.sh mode change 100644 => 100755 tensorflow/tensorflow/lite/acceleration/configuration/prev_is_different_than_current_test.sh mode change 100644 => 100755 tensorflow/tensorflow/lite/delegates/gpu/cl/run_tests.sh mode change 100644 => 100755 tensorflow/tensorflow/lite/delegates/gpu/cl/testing/run_delegate_testing.sh mode change 100644 => 100755 tensorflow/tensorflow/lite/delegates/gpu/cl/testing/run_internal_api_samples.sh mode change 100644 => 100755 tensorflow/tensorflow/lite/delegates/gpu/cl/testing/run_memory_sharing_sample.sh mode change 100644 => 100755 tensorflow/tensorflow/lite/delegates/utils/dummy_delegate/external_delegate_test.sh mode change 100644 => 100755 tensorflow/tensorflow/lite/examples/ios/download_models.sh mode change 100644 => 100755 tensorflow/tensorflow/lite/experimental/acceleration/mini_benchmark/models/blazeface_mlkit_v1.tfl mode change 100644 => 100755 tensorflow/tensorflow/lite/ios/build_frameworks.sh mode change 100644 => 100755 tensorflow/tensorflow/lite/ios/hide_symbols_with_allowlist.sh mode change 100644 => 100755 tensorflow/tensorflow/lite/ios/hide_xcframework_symbols_with_allowlist.sh mode change 100644 => 100755 tensorflow/tensorflow/lite/java/build_aar_for_release.sh mode change 100644 => 100755 tensorflow/tensorflow/lite/java/demo/gradlew mode change 100644 => 100755 tensorflow/tensorflow/lite/java/ovic/demo/gradlew mode change 100644 => 100755 tensorflow/tensorflow/lite/lib_package/concat_licenses.sh mode change 100644 => 100755 tensorflow/tensorflow/lite/python/convert_file_to_c_source_test.sh mode change 100644 => 100755 tensorflow/tensorflow/lite/swift/docsgen/TensorFlowLiteSwift/scripts/download_frameworks.sh mode change 100644 => 100755 tensorflow/tensorflow/lite/testdata/sub.bin mode change 100644 => 100755 tensorflow/tensorflow/lite/tools/benchmark/ios/build_benchmark_framework.sh mode change 100644 => 100755 tensorflow/tensorflow/lite/tools/cmake/download_toolchains.sh mode change 100644 => 100755 tensorflow/tensorflow/lite/tools/delegates/experimental/stable_delegate/test_invalid_settings.json mode change 100644 => 100755 tensorflow/tensorflow/lite/tools/delegates/experimental/stable_delegate/test_missing_delegate_path_settings.json mode change 100644 => 100755 tensorflow/tensorflow/lite/tools/delegates/experimental/stable_delegate/test_missing_stable_delegate_settings.json mode change 100644 => 100755 tensorflow/tensorflow/lite/tools/delegates/experimental/stable_delegate/test_sample_stable_delegate_settings.json mode change 100644 => 100755 tensorflow/tensorflow/lite/tools/delegates/experimental/stable_delegate/test_stable_xnnpack_settings.json mode change 100644 => 100755 tensorflow/tensorflow/lite/tools/pip_package/debian/rules mode change 100644 => 100755 tensorflow/tensorflow/lite/tools/pip_package/update_sources.sh mode change 100644 => 100755 tensorflow/tensorflow/lite/tools/pip_package/with_the_same_user mode change 100644 => 100755 tensorflow/tensorflow/python/autograph/pyct/inspect_utils_test.sh mode change 100644 => 100755 tensorflow/tensorflow/python/compiler/tensorrt/test/testdata/gen_tf_readvariableop_model.py mode change 100644 => 100755 tensorflow/tensorflow/python/debug/examples/v1/examples_v1_debug_errors_test.sh mode change 100644 => 100755 tensorflow/tensorflow/python/debug/examples/v1/examples_v1_debug_fibonacci_test.sh mode change 100644 => 100755 tensorflow/tensorflow/python/debug/examples/v1/examples_v1_debug_keras_test.sh mode change 100644 => 100755 tensorflow/tensorflow/python/debug/examples/v1/examples_v1_debug_mnist_test.sh mode change 100644 => 100755 tensorflow/tensorflow/python/debug/examples/v1/examples_v1_offline_analyzer_test.sh mode change 100644 => 100755 tensorflow/tensorflow/python/debug/examples/v1/examples_v1_test.sh mode change 100644 => 100755 tensorflow/tensorflow/python/debug/examples/v2/examples_v2_test.sh mode change 100644 => 100755 tensorflow/tensorflow/python/tools/no_xla_multithread_symbols_test.sh mode change 100644 => 100755 tensorflow/tensorflow/python/tools/skip_test.sh mode change 100644 => 100755 tensorflow/tensorflow/python/tools/xla_multithread_symbols_test.sh mode change 100644 => 100755 tensorflow/tensorflow/python/tpu/client/pip_package/build_pip_package.sh mode change 100644 => 100755 tensorflow/tensorflow/python/tpu/profiler/pip_package/build_pip_package.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/android/test/gradlew mode change 100644 => 100755 tensorflow/tensorflow/tools/android/test/jni/object_tracking/gl_utils.h mode change 100644 => 100755 tensorflow/tensorflow/tools/android/test/jni/object_tracking/utils_neon.cc mode change 100644 => 100755 tensorflow/tensorflow/tools/android/test/jni/rgb2yuv.cc mode change 100644 => 100755 tensorflow/tensorflow/tools/android/test/jni/rgb2yuv.h mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/build_rbe.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/builds/android.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/builds/android_full.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/builds/benchmark.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/builds/check_system_libs.py mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/builds/cmake.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/builds/configured mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/builds/docker_cpu_pip.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/builds/docker_test.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/builds/integration_tests.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/builds/libtensorflow.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/builds/pip_new.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/builds/print_build_info.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/builds/py_test_delegate.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/builds/run_pip_tests.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/builds/test_tutorials.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/builds/test_user_ops.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/builds/with_the_same_user mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/ci_build.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/code_link_check.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/copy_binary.py mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/devtoolset/build_devtoolset.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/devtoolset/fixlinks.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/devtoolset/rpm-patch.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/gpu_build/parallel_gpu_execute.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/install/build_and_install_python.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/install/install_auditwheel.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/install/install_bootstrap_deb_packages.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/install/install_buildifier.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/install/install_buildifier_from_source.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/install/install_centos_python36.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/install/install_clang.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/install/install_cmake.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/install/install_gcc6.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/install/install_golang.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/install/install_golang_centos.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/install/install_golang_ppc64le.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/install/install_latest_clang.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/install/install_mpi.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/install/install_openblas_ppc64le.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/install/install_pi_python3.9_toolchain.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/install/install_pi_python3x_toolchain.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/install/install_pi_toolchain.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/install/install_pip_packages_remote.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/install/install_proto3.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/install/install_python3.6_pip_packages.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/install/install_python37.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/install/install_yum_packages.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/linux/cmake/run.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/linux/cpu/run_cc_core.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/linux/cpu/run_mkl.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/linux/cpu/run_py2_core.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/linux/cpu/run_py3_contrib.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/linux/cpu/run_py3_core.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/linux/gpu/run_cc_core.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/linux/gpu/run_py3_core.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/linux/libtensorflow.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/linux/libtensorflow_cpu.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/linux/libtensorflow_docker.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/linux/libtensorflow_gpu.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/linux/libtensorflow_rocm.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/linux/mkl/Dockerfile.devel-mkl mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/linux/mkl/basic-mkl-gpu-test.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/linux/mkl/basic-mkl-test.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/linux/mkl/build-dev-container.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/linux/mkl/install_openmpi_horovod.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/linux/ppc64le/cpu/run_py2.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/linux/ppc64le/cpu/run_py3.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/linux/ppc64le/gpu/run_py2.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/linux/ppc64le/gpu/run_py3.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/linux/rocm/rocm_py310_pip.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/linux/rocm/rocm_py36_pip.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/linux/rocm/rocm_py37_pip.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/linux/rocm/rocm_py38_pip.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/linux/rocm/rocm_py39_pip.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/linux/rocm/run_gpu_multi.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/osx/cpu/run_contrib.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/osx/cpu/run_py2_cc_core.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/osx/cpu/run_py3_cc_core.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/osx/libtensorflow_cpu.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/osx/libtensorflow_gpu.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/osx/libtensorflow_rocm.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/pi/build_raspberry_pi.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/ci_build/protobuf/protobuf_optimized_pip.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/gcs_test/gcs_smoke.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/gcs_test/gcs_smoke_wrapper.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/lib_package/concat_licenses.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/lib_package/libtensorflow_java_test.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/tf_sig_build_dockerfiles/builder.devtoolset/fixlinks.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/tf_sig_build_dockerfiles/builder.devtoolset/rpm-patch.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/tf_sig_build_dockerfiles/devel.usertools/get_test_list.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/tf_sig_build_dockerfiles/devel.usertools/repack_libtensorflow.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/tf_sig_build_dockerfiles/devel.usertools/squash_testlogs.py mode change 100644 => 100755 tensorflow/tensorflow/tools/tf_sig_build_dockerfiles/setup.cuda.sh mode change 100644 => 100755 tensorflow/tensorflow/tools/tf_sig_build_dockerfiles/setup.packages.sh create mode 100644 tensorflow/third_party/grpc/upb_gcc10_compile_fix.patch create mode 100644 tensorflow/third_party/kblas/BUILD create mode 100644 tensorflow/third_party/kblas/build_defs.bzl create mode 100644 tensorflow/third_party/kblas/kblas.BUILD create mode 100644 tensorflow/third_party/kblas/workspace.bzl create mode 100644 tensorflow/third_party/ktfop/BUILD create mode 100644 tensorflow/third_party/ktfop/build_defs.bzl create mode 100644 tensorflow/third_party/ktfop/ktfop.BUILD create mode 100644 tensorflow/third_party/ktfop/workspace.bzl mode change 100644 => 100755 tensorflow/third_party/xla/.kokoro/generate_index_html.sh mode change 100644 => 100755 tensorflow/third_party/xla/tools/ci_build/gpu_build/parallel_gpu_execute.sh diff --git a/.gitignore b/.gitignore new file mode 100644 index 000000000..f33aa4703 --- /dev/null +++ b/.gitignore @@ -0,0 +1,3 @@ +logs +/models +.vscode diff --git a/serving/.bazelrc b/serving/.bazelrc index b0c8fb5d2..103da3a30 100644 --- a/serving/.bazelrc +++ b/serving/.bazelrc @@ -52,6 +52,10 @@ build:mkl_aarch64 --copt=-O3 build --define=build_with_onednn_v2=true build --define=xnn_enable_avxvnni=false +build:ktfop --define=build_with_ktfop=true +build:ktfop --define=build_with_kblas=true +build:ktfop -c opt + # Processor native optimizations (depends on build host capabilities). build:nativeopt --copt=-march=native build:nativeopt --host_copt=-march=native diff --git a/tensorflow/.bazelrc b/tensorflow/.bazelrc index d18b2f8a8..58d527d7d 100644 --- a/tensorflow/.bazelrc +++ b/tensorflow/.bazelrc @@ -223,6 +223,11 @@ build:mkl_aarch64 -c opt build:mkl_aarch64_threadpool --define=build_with_mkl_aarch64=true build:mkl_aarch64_threadpool -c opt +# Config setting to build ktfop. +build:ktfop --define=build_with_ktfop=true +build:ktfop --define=build_with_kblas=true +build:ktfop -c opt + # CUDA: This config refers to building CUDA op kernels with nvcc. build:cuda --repo_env TF_NEED_CUDA=1 build:cuda --crosstool_top=@local_config_cuda//crosstool:toolchain diff --git a/tensorflow/ci/official/code_check_changed_files.sh b/tensorflow/ci/official/code_check_changed_files.sh old mode 100644 new mode 100755 diff --git a/tensorflow/ci/official/containers/linux_arm64/builder.devtoolset/build_devtoolset.sh b/tensorflow/ci/official/containers/linux_arm64/builder.devtoolset/build_devtoolset.sh old mode 100644 new mode 100755 diff --git a/tensorflow/ci/official/containers/linux_arm64/builder.devtoolset/fixlinks_aarch64.sh b/tensorflow/ci/official/containers/linux_arm64/builder.devtoolset/fixlinks_aarch64.sh old mode 100644 new mode 100755 diff --git a/tensorflow/ci/official/containers/linux_arm64/builder.devtoolset/rpm-patch.sh b/tensorflow/ci/official/containers/linux_arm64/builder.devtoolset/rpm-patch.sh old mode 100644 new mode 100755 diff --git a/tensorflow/ci/official/containers/linux_arm64/builder.patchelf/build_patchelf.sh b/tensorflow/ci/official/containers/linux_arm64/builder.patchelf/build_patchelf.sh old mode 100644 new mode 100755 diff --git a/tensorflow/ci/official/containers/linux_arm64/devel.usertools/get_test_list.sh b/tensorflow/ci/official/containers/linux_arm64/devel.usertools/get_test_list.sh old mode 100644 new mode 100755 diff --git a/tensorflow/ci/official/containers/linux_arm64/devel.usertools/repack_libtensorflow.sh b/tensorflow/ci/official/containers/linux_arm64/devel.usertools/repack_libtensorflow.sh old mode 100644 new mode 100755 diff --git a/tensorflow/ci/official/containers/linux_arm64/devel.usertools/squash_testlogs.py b/tensorflow/ci/official/containers/linux_arm64/devel.usertools/squash_testlogs.py old mode 100644 new mode 100755 diff --git a/tensorflow/ci/official/containers/linux_arm64/setup.packages.sh b/tensorflow/ci/official/containers/linux_arm64/setup.packages.sh old mode 100644 new mode 100755 diff --git a/tensorflow/ci/official/utilities/cleanup_docker.sh b/tensorflow/ci/official/utilities/cleanup_docker.sh old mode 100644 new mode 100755 diff --git a/tensorflow/ci/official/utilities/generate_index_html.sh b/tensorflow/ci/official/utilities/generate_index_html.sh old mode 100644 new mode 100755 diff --git a/tensorflow/configure b/tensorflow/configure old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/c/experimental/ops/update_cpp_ops.sh b/tensorflow/tensorflow/c/experimental/ops/update_cpp_ops.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/c/generate-pc.sh b/tensorflow/tensorflow/c/generate-pc.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/cc/saved_model/testdata/half_plus_two/00000123/saved_model.pb b/tensorflow/tensorflow/cc/saved_model/testdata/half_plus_two/00000123/saved_model.pb old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/cc/saved_model/testdata/half_plus_two/00000123/variables/variables.data-00000-of-00001 b/tensorflow/tensorflow/cc/saved_model/testdata/half_plus_two/00000123/variables/variables.data-00000-of-00001 old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/cc/saved_model/testdata/half_plus_two/00000123/variables/variables.index b/tensorflow/tensorflow/cc/saved_model/testdata/half_plus_two/00000123/variables/variables.index old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/cc/saved_model/testdata/half_plus_two_pbtxt/00000123/saved_model.pbtxt b/tensorflow/tensorflow/cc/saved_model/testdata/half_plus_two_pbtxt/00000123/saved_model.pbtxt old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/cc/saved_model/testdata/half_plus_two_pbtxt/00000123/variables/variables.data-00000-of-00001 b/tensorflow/tensorflow/cc/saved_model/testdata/half_plus_two_pbtxt/00000123/variables/variables.data-00000-of-00001 old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/cc/saved_model/testdata/half_plus_two_pbtxt/00000123/variables/variables.index b/tensorflow/tensorflow/cc/saved_model/testdata/half_plus_two_pbtxt/00000123/variables/variables.index old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/compiler/mlir/tensorflow/tests/reducer/unsupported-op-test.sh b/tensorflow/tensorflow/compiler/mlir/tensorflow/tests/reducer/unsupported-op-test.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/compiler/tf2tensorrt/utils/trt_engine_utils.cc b/tensorflow/tensorflow/compiler/tf2tensorrt/utils/trt_engine_utils.cc old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/core/BUILD b/tensorflow/tensorflow/core/BUILD index 538574360..e2da99cc6 100644 --- a/tensorflow/tensorflow/core/BUILD +++ b/tensorflow/tensorflow/core/BUILD @@ -111,6 +111,10 @@ load( "//third_party/mkl:build_defs.bzl", "if_mkl", ) +load( + "//third_party/ktfop:build_defs.bzl", + "if_ktfop", +) load("@local_config_tensorrt//:build_defs.bzl", "if_tensorrt") package( @@ -650,6 +654,9 @@ cc_library( "//tensorflow/core/kernels/mkl:mkl_matmul_op", "//tensorflow/core/kernels/mkl:mkl_tmp_bf16_ops", "//tensorflow/core/kernels/mkl:mkl_deprecated_ops", + ]) + if_ktfop([ + "//tensorflow/core/kernels/ktfop:fused_embedding_ops", + "//tensorflow/core/kernels/ktfop:softmax_ops", ]) + if_cuda_or_rocm([ "//tensorflow/core/kernels:cudnn_rnn_kernels", ]) + if_cuda([ diff --git a/tensorflow/tensorflow/core/api_def/update_api_def.sh b/tensorflow/tensorflow/core/api_def/update_api_def.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/core/kernels/BUILD b/tensorflow/tensorflow/core/kernels/BUILD index 22617ef8a..60ce87f44 100644 --- a/tensorflow/tensorflow/core/kernels/BUILD +++ b/tensorflow/tensorflow/core/kernels/BUILD @@ -38,6 +38,10 @@ load( "if_mkl", "mkl_deps", ) +load( + "//third_party/ktfop:build_defs.bzl", + "if_ktfop", +) load( "@local_config_rocm//rocm:build_defs.bzl", "if_rocm", diff --git a/tensorflow/tensorflow/core/kernels/ktfop/BUILD b/tensorflow/tensorflow/core/kernels/ktfop/BUILD new file mode 100644 index 000000000..a6fa0e71a --- /dev/null +++ b/tensorflow/tensorflow/core/kernels/ktfop/BUILD @@ -0,0 +1,36 @@ +load("//tensorflow:tensorflow.default.bzl", "tf_kernel_library") + +package( + # copybara:uncomment default_applicable_licenses = ["//tensorflow:license"], + default_visibility = ["//tensorflow:internal"], +) + +DYNAMIC_DEPS = [ + "//tensorflow/core/framework:bounds_check", + "//tensorflow/core:core_cpu", + "//tensorflow/core:framework", + "//tensorflow/core:lib", + "//tensorflow/core:lib_internal", +] + +tf_kernel_library( + name = "fused_embedding_ops", + srcs = [ + "embedding_lookup_op.cc", + ], + deps = [ + "@eigen_archive//:eigen3", + "@ktfop_archive//:ktfop", + ] + DYNAMIC_DEPS +) + +tf_kernel_library( + name = "softmax_ops", + srcs = [ + "softmax.cc", + ], + deps = [ + "@eigen_archive//:eigen3", + "@ktfop_archive//:ktfop", + ] + DYNAMIC_DEPS +) \ No newline at end of file diff --git a/tensorflow/tensorflow/core/kernels/ktfop/embedding_lookup_op.cc b/tensorflow/tensorflow/core/kernels/ktfop/embedding_lookup_op.cc new file mode 100644 index 000000000..a6f69458a --- /dev/null +++ b/tensorflow/tensorflow/core/kernels/ktfop/embedding_lookup_op.cc @@ -0,0 +1,149 @@ +#define EIGEN_USE_THREADS + +#include "tensorflow/core/framework/op.h" +#include "tensorflow/core/framework/resource_mgr.h" +#include "tensorflow/core/framework/shape_inference.h" +#include "tensorflow/core/framework/op_kernel.h" +#include "tensorflow/core/framework/resource_var.h" +#include "tensorflow/core/framework/bounds_check.h" + +#include "ktfop.h" + +namespace tensorflow { + +typedef Eigen::ThreadPoolDevice CPUDevice; + +class KPFusedEmbeddingOp : public OpKernel { +public: + explicit KPFusedEmbeddingOp(OpKernelConstruction* context) + : OpKernel(context) { + OP_REQUIRES_OK(context, context->GetAttr("combiner", &combiner_)); + node_name = context->def().name(); + } + + ~KPFusedEmbeddingOp() {} + + void Compute(OpKernelContext* context) override { + float *weight; + const Tensor* weight_tensor = &context->input(0); + + if (weight_tensor->dtype() == DT_RESOURCE) { + Var* variable; + OP_REQUIRES_OK(context, + LookupResource(context, HandleFromInput(context, 0), + &variable)); + core::ScopedUnref s(variable); + weight_tensor = variable->tensor(); + OP_REQUIRES(context, weight_tensor->dtype() == DT_FLOAT, + errors::InvalidArgument("Expect float weight in ", + node_name)); + } + + weight = (float *)weight_tensor->tensor_data().data(); + + const Tensor& input_tensor = context->input(1); + int64 *input = (int64 *)input_tensor.tensor_data().data(); + const Tensor& shape_tensor = context->input(2); + int64 *shape = (int64 *)shape_tensor.tensor_data().data(); + + OP_REQUIRES(context, (shape_tensor.dims() == 1), + errors::InvalidArgument("Shape tensor is not valid (dims != 1)")); + OP_REQUIRES(context, (shape_tensor.dim_size(0) >= 2), + errors::InvalidArgument("Shape tensor is not valid (dim_size(0) < 2)")); + + int64 input_size = 1; + for (int i = 0; i < input_tensor.dims(); ++i) { + input_size *= input_tensor.dim_size(i); + } + int input_dims = shape_tensor.dim_size(0); + int cols = shape[input_dims - 1]; + int batch_size = 1; + for (int i = 0; i < input_dims - 1; ++i) { + batch_size *= shape[i]; + } + OP_REQUIRES(context, (input_size == batch_size * cols), + errors::InvalidArgument("input id is dense")); + int embedding_dims = weight_tensor->dim_size(1); + bool is_mean = (combiner_ == 1); + + Tensor* output_tensor = NULL; + TensorShape output_shape({batch_size, embedding_dims}); + OP_REQUIRES_OK(context, context->allocate_output(0, output_shape, + &output_tensor)); + float *output = (float *)output_tensor->tensor_data().data(); + ktfop::EmbeddingParams params(input, + batch_size, + cols, + weight, + embedding_dims, + is_mean); + int result = ktfop::FusedEmbedding(params, output); + OP_REQUIRES(context, (result == 0), + errors::InvalidArgument("Invalid argument, error code: ", result)); + } + +private: + int combiner_; + std::string node_name; +}; + +REGISTER_KERNEL_BUILDER(Name("KPFusedEmbedding").Device(DEVICE_CPU), KPFusedEmbeddingOp); + +class KPFusedEmbeddingWithHashBucketOp : public OpKernel { + public: + explicit KPFusedEmbeddingWithHashBucketOp(OpKernelConstruction* context) + : OpKernel(context) { + OP_REQUIRES_OK(context, context->GetAttr("num_buckets", &num_buckets_)); + node_name = context->def().name(); + } + + void Compute(OpKernelContext* context) override { + float *weight; + const Tensor& input_tensor = context->input(0); + const Tensor* weight_tensor = &context->input(1); + + if (weight_tensor->dtype() == DT_RESOURCE) { + Var* variable; + OP_REQUIRES_OK(context, + LookupResource(context, HandleFromInput(context, 1), + &variable)); + core::ScopedUnref s(variable); + weight_tensor = variable->tensor(); + OP_REQUIRES(context, weight_tensor->dtype() == DT_FLOAT, + errors::InvalidArgument("Expect float weight in ", + node_name)); + } + + auto input = input_tensor.flat(); + weight = (float *)weight_tensor->tensor_data().data(); + int64_t batch = input_tensor.dim_size(0); + int64_t embedding_dims = weight_tensor->dim_size(1); + uintptr_t cstr_addresses[batch]; + size_t cstr_length[batch]; + for (int i = 0; i < batch; ++i) { + cstr_addresses[i] = reinterpret_cast(input(i).c_str()); + cstr_length[i] = input(i).length(); + } + Tensor* output_tensor = nullptr; + OP_REQUIRES_OK(context, + context->allocate_output( + 0, TensorShape({batch, embedding_dims}), + &output_tensor)); + float *output = (float *)output_tensor->tensor_data().data(); + ktfop::EmbeddingParamsWithHash params(cstr_addresses, + cstr_length, + batch, + weight, + num_buckets_, + embedding_dims); + int result = ktfop::FusedEmbeddingWithHashBucket(params, output); + OP_REQUIRES(context, (result == 0), + errors::InvalidArgument("Invalid argument, error code: ", result)); + } + + private: + int64_t num_buckets_; + std::string node_name; +}; +REGISTER_KERNEL_BUILDER(Name("KPFusedEmbeddingWithHashBucket").Device(DEVICE_CPU), KPFusedEmbeddingWithHashBucketOp); +} // namespace tensorflow diff --git a/tensorflow/tensorflow/core/kernels/ktfop/softmax.cc b/tensorflow/tensorflow/core/kernels/ktfop/softmax.cc new file mode 100644 index 000000000..401aa2abc --- /dev/null +++ b/tensorflow/tensorflow/core/kernels/ktfop/softmax.cc @@ -0,0 +1,37 @@ +#include "tensorflow/core/framework/op.h" +#include "tensorflow/core/framework/op_kernel.h" +#include "tensorflow/core/framework/tensor.h" +#include "tensorflow/core/framework/tensor_shape.h" + +#include "unsupported/Eigen/CXX11/Tensor" // from @eigen_archive +#include "ktfop.h" + +namespace tensorflow { +class KPSoftmaxOp : public OpKernel { + public: + explicit KPSoftmaxOp(OpKernelConstruction* context) : OpKernel(context) {} + + void Compute(OpKernelContext* context) override { + const Tensor& logits_in = context->input(0); + OP_REQUIRES(context, TensorShapeUtils::IsVectorOrHigher(logits_in.shape()), + errors::InvalidArgument("logits must have >= 1 dimension, got ", + logits_in.shape().DebugString())); + Tensor* softmax_out = nullptr; + OP_REQUIRES_OK(context, context->forward_input_or_allocate_output( + {0}, 0, logits_in.shape(), &softmax_out)); + if (logits_in.NumElements() > 0) { + typename TTypes::ConstMatrix input = logits_in.flat_inner_dims(); + float* input_data = (float *)logits_in.data(); + float* output_data = (float *)softmax_out->data(); + int result = ktfop::Softmax(input_data, output_data, input.dimension(0), input.dimension(1)); + OP_REQUIRES(context, (result == 0), + errors::InvalidArgument("Invalid argument, error code: ", result)); + } + } +}; + +REGISTER_KERNEL_BUILDER( + Name("KPSoftmax").Device(DEVICE_CPU).TypeConstraint("T"), + KPSoftmaxOp); + +} // namespace tensorflow \ No newline at end of file diff --git a/tensorflow/tensorflow/core/kernels/mlir_generated/build_test.sh b/tensorflow/tensorflow/core/kernels/mlir_generated/build_test.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/core/ops/BUILD b/tensorflow/tensorflow/core/ops/BUILD index 91d80b6c2..61e586433 100644 --- a/tensorflow/tensorflow/core/ops/BUILD +++ b/tensorflow/tensorflow/core/ops/BUILD @@ -18,7 +18,10 @@ load( "//third_party/mkl:build_defs.bzl", "if_mkl", ) - +load( + "//third_party/ktfop:build_defs.bzl", + "if_ktfop", +) # A lot of packages try to minimize binary size by depending on individual ops,\ # so they need access here. package( @@ -60,6 +63,7 @@ tf_gen_op_libs( "functional_ops", "image_ops", "io_ops", + "ktfop_ops", "linalg_ops", "list_ops", "map_ops", @@ -344,6 +348,8 @@ cc_library( }) + if_mkl([ ":mkl_array_ops_op_lib", ":mkl_nn_ops_op_lib", + ]) + if_ktfop([ + ":ktfop_ops_op_lib", ]), alwayslink = 1, ) diff --git a/tensorflow/tensorflow/core/ops/ktfop_ops.cc b/tensorflow/tensorflow/core/ops/ktfop_ops.cc new file mode 100644 index 000000000..c2f63fae4 --- /dev/null +++ b/tensorflow/tensorflow/core/ops/ktfop_ops.cc @@ -0,0 +1,66 @@ +#include + +#include "tensorflow/core/framework/op.h" +#include "tensorflow/core/framework/shape_inference.h" +#include "tensorflow/core/framework/common_shape_fns.h" + +namespace tensorflow { + +using shape_inference::DimensionHandle; +using shape_inference::InferenceContext; +using shape_inference::ShapeHandle; +using shape_inference::UnchangedShape; + +REGISTER_OP("KPFusedEmbedding") + .Input("weights: float") + .Input("lookup: int64") + .Input("dense_shape: int64") + .Input("indices: int64") + .Output("output: float") + .Attr("combiner: int") + + .SetShapeFn([](InferenceContext* ctx) { + ShapeHandle temp; + TF_RETURN_IF_ERROR(ctx->WithRank(ctx->input(1), 1, &temp)); + TF_RETURN_IF_ERROR(ctx->WithRank(ctx->input(3), 2, &temp)); + TF_RETURN_IF_ERROR(ctx->WithRank(ctx->input(2), 1, &temp)); + ShapeHandle emb_var_shape; + TF_RETURN_IF_ERROR(ctx->WithRank(ctx->input(0), 2, &emb_var_shape)); + + DimensionHandle emb_size_dim = ctx->Dim(emb_var_shape, 1); + DimensionHandle batch_dim = ctx->UnknownDim(); + + ShapeHandle output_shape = ctx->MakeShape({batch_dim, emb_size_dim}); + ctx->set_output(0, output_shape); + + return OkStatus(); + }); + +REGISTER_OP("KPFusedEmbeddingWithHashBucket") + .Input("lookup: string") + .Input("weights: T_weight") + .Attr("num_buckets: int >= 1") + .Attr("combiner: int") + .Attr("T_weight: {resource, float}") + .Output("output: float") + .SetShapeFn([](InferenceContext* ctx) { + ShapeHandle temp; + TF_RETURN_IF_ERROR(ctx->WithRank(ctx->input(0), 1, &temp)); + DimensionHandle emb_size_dim = ctx->UnknownDim(); + DimensionHandle batch_dim = ctx->UnknownDim(); + + ShapeHandle output_shape = ctx->MakeShape({batch_dim, emb_size_dim}); + ctx->set_output(0, output_shape); + + return OkStatus(); + }); + +REGISTER_OP("KPSoftmax") + .Input("logits: T") + .Output("softmax: T") + .Attr("T: {float}") + .SetShapeFn([](InferenceContext* c) { + return shape_inference::UnchangedShapeWithRankAtLeast(c, 1); + }); + +} // namespace tensorflow \ No newline at end of file diff --git a/tensorflow/tensorflow/examples/multibox_detector/data/surfers.jpg b/tensorflow/tensorflow/examples/multibox_detector/data/surfers.jpg old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/go/test.sh b/tensorflow/tensorflow/go/test.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/lite/acceleration/configuration/prev_is_different_than_current_test.sh b/tensorflow/tensorflow/lite/acceleration/configuration/prev_is_different_than_current_test.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/lite/delegates/gpu/cl/run_tests.sh b/tensorflow/tensorflow/lite/delegates/gpu/cl/run_tests.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/lite/delegates/gpu/cl/testing/run_delegate_testing.sh b/tensorflow/tensorflow/lite/delegates/gpu/cl/testing/run_delegate_testing.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/lite/delegates/gpu/cl/testing/run_internal_api_samples.sh b/tensorflow/tensorflow/lite/delegates/gpu/cl/testing/run_internal_api_samples.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/lite/delegates/gpu/cl/testing/run_memory_sharing_sample.sh b/tensorflow/tensorflow/lite/delegates/gpu/cl/testing/run_memory_sharing_sample.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/lite/delegates/utils/dummy_delegate/external_delegate_test.sh b/tensorflow/tensorflow/lite/delegates/utils/dummy_delegate/external_delegate_test.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/lite/examples/ios/download_models.sh b/tensorflow/tensorflow/lite/examples/ios/download_models.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/lite/experimental/acceleration/mini_benchmark/models/blazeface_mlkit_v1.tfl b/tensorflow/tensorflow/lite/experimental/acceleration/mini_benchmark/models/blazeface_mlkit_v1.tfl old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/lite/ios/build_frameworks.sh b/tensorflow/tensorflow/lite/ios/build_frameworks.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/lite/ios/hide_symbols_with_allowlist.sh b/tensorflow/tensorflow/lite/ios/hide_symbols_with_allowlist.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/lite/ios/hide_xcframework_symbols_with_allowlist.sh b/tensorflow/tensorflow/lite/ios/hide_xcframework_symbols_with_allowlist.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/lite/java/build_aar_for_release.sh b/tensorflow/tensorflow/lite/java/build_aar_for_release.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/lite/java/demo/gradlew b/tensorflow/tensorflow/lite/java/demo/gradlew old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/lite/java/ovic/demo/gradlew b/tensorflow/tensorflow/lite/java/ovic/demo/gradlew old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/lite/lib_package/concat_licenses.sh b/tensorflow/tensorflow/lite/lib_package/concat_licenses.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/lite/python/convert_file_to_c_source_test.sh b/tensorflow/tensorflow/lite/python/convert_file_to_c_source_test.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/lite/swift/docsgen/TensorFlowLiteSwift/scripts/download_frameworks.sh b/tensorflow/tensorflow/lite/swift/docsgen/TensorFlowLiteSwift/scripts/download_frameworks.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/lite/testdata/sub.bin b/tensorflow/tensorflow/lite/testdata/sub.bin old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/lite/tools/benchmark/ios/build_benchmark_framework.sh b/tensorflow/tensorflow/lite/tools/benchmark/ios/build_benchmark_framework.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/lite/tools/cmake/download_toolchains.sh b/tensorflow/tensorflow/lite/tools/cmake/download_toolchains.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/lite/tools/delegates/experimental/stable_delegate/test_invalid_settings.json b/tensorflow/tensorflow/lite/tools/delegates/experimental/stable_delegate/test_invalid_settings.json old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/lite/tools/delegates/experimental/stable_delegate/test_missing_delegate_path_settings.json b/tensorflow/tensorflow/lite/tools/delegates/experimental/stable_delegate/test_missing_delegate_path_settings.json old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/lite/tools/delegates/experimental/stable_delegate/test_missing_stable_delegate_settings.json b/tensorflow/tensorflow/lite/tools/delegates/experimental/stable_delegate/test_missing_stable_delegate_settings.json old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/lite/tools/delegates/experimental/stable_delegate/test_sample_stable_delegate_settings.json b/tensorflow/tensorflow/lite/tools/delegates/experimental/stable_delegate/test_sample_stable_delegate_settings.json old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/lite/tools/delegates/experimental/stable_delegate/test_stable_xnnpack_settings.json b/tensorflow/tensorflow/lite/tools/delegates/experimental/stable_delegate/test_stable_xnnpack_settings.json old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/lite/tools/pip_package/debian/rules b/tensorflow/tensorflow/lite/tools/pip_package/debian/rules old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/lite/tools/pip_package/update_sources.sh b/tensorflow/tensorflow/lite/tools/pip_package/update_sources.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/lite/tools/pip_package/with_the_same_user b/tensorflow/tensorflow/lite/tools/pip_package/with_the_same_user old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/python/autograph/pyct/inspect_utils_test.sh b/tensorflow/tensorflow/python/autograph/pyct/inspect_utils_test.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/python/compiler/tensorrt/test/testdata/gen_tf_readvariableop_model.py b/tensorflow/tensorflow/python/compiler/tensorrt/test/testdata/gen_tf_readvariableop_model.py old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/python/debug/examples/v1/examples_v1_debug_errors_test.sh b/tensorflow/tensorflow/python/debug/examples/v1/examples_v1_debug_errors_test.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/python/debug/examples/v1/examples_v1_debug_fibonacci_test.sh b/tensorflow/tensorflow/python/debug/examples/v1/examples_v1_debug_fibonacci_test.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/python/debug/examples/v1/examples_v1_debug_keras_test.sh b/tensorflow/tensorflow/python/debug/examples/v1/examples_v1_debug_keras_test.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/python/debug/examples/v1/examples_v1_debug_mnist_test.sh b/tensorflow/tensorflow/python/debug/examples/v1/examples_v1_debug_mnist_test.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/python/debug/examples/v1/examples_v1_offline_analyzer_test.sh b/tensorflow/tensorflow/python/debug/examples/v1/examples_v1_offline_analyzer_test.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/python/debug/examples/v1/examples_v1_test.sh b/tensorflow/tensorflow/python/debug/examples/v1/examples_v1_test.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/python/debug/examples/v2/examples_v2_test.sh b/tensorflow/tensorflow/python/debug/examples/v2/examples_v2_test.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/python/tools/no_xla_multithread_symbols_test.sh b/tensorflow/tensorflow/python/tools/no_xla_multithread_symbols_test.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/python/tools/skip_test.sh b/tensorflow/tensorflow/python/tools/skip_test.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/python/tools/xla_multithread_symbols_test.sh b/tensorflow/tensorflow/python/tools/xla_multithread_symbols_test.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/python/tpu/client/pip_package/build_pip_package.sh b/tensorflow/tensorflow/python/tpu/client/pip_package/build_pip_package.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/python/tpu/profiler/pip_package/build_pip_package.sh b/tensorflow/tensorflow/python/tpu/profiler/pip_package/build_pip_package.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/android/test/gradlew b/tensorflow/tensorflow/tools/android/test/gradlew old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/android/test/jni/object_tracking/gl_utils.h b/tensorflow/tensorflow/tools/android/test/jni/object_tracking/gl_utils.h old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/android/test/jni/object_tracking/utils_neon.cc b/tensorflow/tensorflow/tools/android/test/jni/object_tracking/utils_neon.cc old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/android/test/jni/rgb2yuv.cc b/tensorflow/tensorflow/tools/android/test/jni/rgb2yuv.cc old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/android/test/jni/rgb2yuv.h b/tensorflow/tensorflow/tools/android/test/jni/rgb2yuv.h old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/build_rbe.sh b/tensorflow/tensorflow/tools/ci_build/build_rbe.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/builds/android.sh b/tensorflow/tensorflow/tools/ci_build/builds/android.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/builds/android_full.sh b/tensorflow/tensorflow/tools/ci_build/builds/android_full.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/builds/benchmark.sh b/tensorflow/tensorflow/tools/ci_build/builds/benchmark.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/builds/check_system_libs.py b/tensorflow/tensorflow/tools/ci_build/builds/check_system_libs.py old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/builds/cmake.sh b/tensorflow/tensorflow/tools/ci_build/builds/cmake.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/builds/configured b/tensorflow/tensorflow/tools/ci_build/builds/configured old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/builds/docker_cpu_pip.sh b/tensorflow/tensorflow/tools/ci_build/builds/docker_cpu_pip.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/builds/docker_test.sh b/tensorflow/tensorflow/tools/ci_build/builds/docker_test.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/builds/integration_tests.sh b/tensorflow/tensorflow/tools/ci_build/builds/integration_tests.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/builds/libtensorflow.sh b/tensorflow/tensorflow/tools/ci_build/builds/libtensorflow.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/builds/pip_new.sh b/tensorflow/tensorflow/tools/ci_build/builds/pip_new.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/builds/print_build_info.sh b/tensorflow/tensorflow/tools/ci_build/builds/print_build_info.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/builds/py_test_delegate.sh b/tensorflow/tensorflow/tools/ci_build/builds/py_test_delegate.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/builds/run_pip_tests.sh b/tensorflow/tensorflow/tools/ci_build/builds/run_pip_tests.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/builds/test_tutorials.sh b/tensorflow/tensorflow/tools/ci_build/builds/test_tutorials.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/builds/test_user_ops.sh b/tensorflow/tensorflow/tools/ci_build/builds/test_user_ops.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/builds/with_the_same_user b/tensorflow/tensorflow/tools/ci_build/builds/with_the_same_user old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/ci_build.sh b/tensorflow/tensorflow/tools/ci_build/ci_build.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/code_link_check.sh b/tensorflow/tensorflow/tools/ci_build/code_link_check.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/copy_binary.py b/tensorflow/tensorflow/tools/ci_build/copy_binary.py old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/devtoolset/build_devtoolset.sh b/tensorflow/tensorflow/tools/ci_build/devtoolset/build_devtoolset.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/devtoolset/fixlinks.sh b/tensorflow/tensorflow/tools/ci_build/devtoolset/fixlinks.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/devtoolset/rpm-patch.sh b/tensorflow/tensorflow/tools/ci_build/devtoolset/rpm-patch.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/gpu_build/parallel_gpu_execute.sh b/tensorflow/tensorflow/tools/ci_build/gpu_build/parallel_gpu_execute.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/install/build_and_install_python.sh b/tensorflow/tensorflow/tools/ci_build/install/build_and_install_python.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/install/install_auditwheel.sh b/tensorflow/tensorflow/tools/ci_build/install/install_auditwheel.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/install/install_bootstrap_deb_packages.sh b/tensorflow/tensorflow/tools/ci_build/install/install_bootstrap_deb_packages.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/install/install_buildifier.sh b/tensorflow/tensorflow/tools/ci_build/install/install_buildifier.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/install/install_buildifier_from_source.sh b/tensorflow/tensorflow/tools/ci_build/install/install_buildifier_from_source.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/install/install_centos_python36.sh b/tensorflow/tensorflow/tools/ci_build/install/install_centos_python36.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/install/install_clang.sh b/tensorflow/tensorflow/tools/ci_build/install/install_clang.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/install/install_cmake.sh b/tensorflow/tensorflow/tools/ci_build/install/install_cmake.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/install/install_gcc6.sh b/tensorflow/tensorflow/tools/ci_build/install/install_gcc6.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/install/install_golang.sh b/tensorflow/tensorflow/tools/ci_build/install/install_golang.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/install/install_golang_centos.sh b/tensorflow/tensorflow/tools/ci_build/install/install_golang_centos.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/install/install_golang_ppc64le.sh b/tensorflow/tensorflow/tools/ci_build/install/install_golang_ppc64le.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/install/install_latest_clang.sh b/tensorflow/tensorflow/tools/ci_build/install/install_latest_clang.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/install/install_mpi.sh b/tensorflow/tensorflow/tools/ci_build/install/install_mpi.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/install/install_openblas_ppc64le.sh b/tensorflow/tensorflow/tools/ci_build/install/install_openblas_ppc64le.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/install/install_pi_python3.9_toolchain.sh b/tensorflow/tensorflow/tools/ci_build/install/install_pi_python3.9_toolchain.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/install/install_pi_python3x_toolchain.sh b/tensorflow/tensorflow/tools/ci_build/install/install_pi_python3x_toolchain.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/install/install_pi_toolchain.sh b/tensorflow/tensorflow/tools/ci_build/install/install_pi_toolchain.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/install/install_pip_packages_remote.sh b/tensorflow/tensorflow/tools/ci_build/install/install_pip_packages_remote.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/install/install_proto3.sh b/tensorflow/tensorflow/tools/ci_build/install/install_proto3.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/install/install_python3.6_pip_packages.sh b/tensorflow/tensorflow/tools/ci_build/install/install_python3.6_pip_packages.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/install/install_python37.sh b/tensorflow/tensorflow/tools/ci_build/install/install_python37.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/install/install_yum_packages.sh b/tensorflow/tensorflow/tools/ci_build/install/install_yum_packages.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/linux/cmake/run.sh b/tensorflow/tensorflow/tools/ci_build/linux/cmake/run.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/linux/cpu/run_cc_core.sh b/tensorflow/tensorflow/tools/ci_build/linux/cpu/run_cc_core.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/linux/cpu/run_mkl.sh b/tensorflow/tensorflow/tools/ci_build/linux/cpu/run_mkl.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/linux/cpu/run_py2_core.sh b/tensorflow/tensorflow/tools/ci_build/linux/cpu/run_py2_core.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/linux/cpu/run_py3_contrib.sh b/tensorflow/tensorflow/tools/ci_build/linux/cpu/run_py3_contrib.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/linux/cpu/run_py3_core.sh b/tensorflow/tensorflow/tools/ci_build/linux/cpu/run_py3_core.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/linux/gpu/run_cc_core.sh b/tensorflow/tensorflow/tools/ci_build/linux/gpu/run_cc_core.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/linux/gpu/run_py3_core.sh b/tensorflow/tensorflow/tools/ci_build/linux/gpu/run_py3_core.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/linux/libtensorflow.sh b/tensorflow/tensorflow/tools/ci_build/linux/libtensorflow.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/linux/libtensorflow_cpu.sh b/tensorflow/tensorflow/tools/ci_build/linux/libtensorflow_cpu.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/linux/libtensorflow_docker.sh b/tensorflow/tensorflow/tools/ci_build/linux/libtensorflow_docker.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/linux/libtensorflow_gpu.sh b/tensorflow/tensorflow/tools/ci_build/linux/libtensorflow_gpu.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/linux/libtensorflow_rocm.sh b/tensorflow/tensorflow/tools/ci_build/linux/libtensorflow_rocm.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/linux/mkl/Dockerfile.devel-mkl b/tensorflow/tensorflow/tools/ci_build/linux/mkl/Dockerfile.devel-mkl old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/linux/mkl/basic-mkl-gpu-test.sh b/tensorflow/tensorflow/tools/ci_build/linux/mkl/basic-mkl-gpu-test.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/linux/mkl/basic-mkl-test.sh b/tensorflow/tensorflow/tools/ci_build/linux/mkl/basic-mkl-test.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/linux/mkl/build-dev-container.sh b/tensorflow/tensorflow/tools/ci_build/linux/mkl/build-dev-container.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/linux/mkl/install_openmpi_horovod.sh b/tensorflow/tensorflow/tools/ci_build/linux/mkl/install_openmpi_horovod.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/linux/ppc64le/cpu/run_py2.sh b/tensorflow/tensorflow/tools/ci_build/linux/ppc64le/cpu/run_py2.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/linux/ppc64le/cpu/run_py3.sh b/tensorflow/tensorflow/tools/ci_build/linux/ppc64le/cpu/run_py3.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/linux/ppc64le/gpu/run_py2.sh b/tensorflow/tensorflow/tools/ci_build/linux/ppc64le/gpu/run_py2.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/linux/ppc64le/gpu/run_py3.sh b/tensorflow/tensorflow/tools/ci_build/linux/ppc64le/gpu/run_py3.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/linux/rocm/rocm_py310_pip.sh b/tensorflow/tensorflow/tools/ci_build/linux/rocm/rocm_py310_pip.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/linux/rocm/rocm_py36_pip.sh b/tensorflow/tensorflow/tools/ci_build/linux/rocm/rocm_py36_pip.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/linux/rocm/rocm_py37_pip.sh b/tensorflow/tensorflow/tools/ci_build/linux/rocm/rocm_py37_pip.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/linux/rocm/rocm_py38_pip.sh b/tensorflow/tensorflow/tools/ci_build/linux/rocm/rocm_py38_pip.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/linux/rocm/rocm_py39_pip.sh b/tensorflow/tensorflow/tools/ci_build/linux/rocm/rocm_py39_pip.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/linux/rocm/run_gpu_multi.sh b/tensorflow/tensorflow/tools/ci_build/linux/rocm/run_gpu_multi.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/osx/cpu/run_contrib.sh b/tensorflow/tensorflow/tools/ci_build/osx/cpu/run_contrib.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/osx/cpu/run_py2_cc_core.sh b/tensorflow/tensorflow/tools/ci_build/osx/cpu/run_py2_cc_core.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/osx/cpu/run_py3_cc_core.sh b/tensorflow/tensorflow/tools/ci_build/osx/cpu/run_py3_cc_core.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/osx/libtensorflow_cpu.sh b/tensorflow/tensorflow/tools/ci_build/osx/libtensorflow_cpu.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/osx/libtensorflow_gpu.sh b/tensorflow/tensorflow/tools/ci_build/osx/libtensorflow_gpu.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/osx/libtensorflow_rocm.sh b/tensorflow/tensorflow/tools/ci_build/osx/libtensorflow_rocm.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/pi/build_raspberry_pi.sh b/tensorflow/tensorflow/tools/ci_build/pi/build_raspberry_pi.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/ci_build/protobuf/protobuf_optimized_pip.sh b/tensorflow/tensorflow/tools/ci_build/protobuf/protobuf_optimized_pip.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/gcs_test/gcs_smoke.sh b/tensorflow/tensorflow/tools/gcs_test/gcs_smoke.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/gcs_test/gcs_smoke_wrapper.sh b/tensorflow/tensorflow/tools/gcs_test/gcs_smoke_wrapper.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/lib_package/concat_licenses.sh b/tensorflow/tensorflow/tools/lib_package/concat_licenses.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/lib_package/libtensorflow_java_test.sh b/tensorflow/tensorflow/tools/lib_package/libtensorflow_java_test.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/tf_sig_build_dockerfiles/builder.devtoolset/fixlinks.sh b/tensorflow/tensorflow/tools/tf_sig_build_dockerfiles/builder.devtoolset/fixlinks.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/tf_sig_build_dockerfiles/builder.devtoolset/rpm-patch.sh b/tensorflow/tensorflow/tools/tf_sig_build_dockerfiles/builder.devtoolset/rpm-patch.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/tf_sig_build_dockerfiles/devel.usertools/get_test_list.sh b/tensorflow/tensorflow/tools/tf_sig_build_dockerfiles/devel.usertools/get_test_list.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/tf_sig_build_dockerfiles/devel.usertools/repack_libtensorflow.sh b/tensorflow/tensorflow/tools/tf_sig_build_dockerfiles/devel.usertools/repack_libtensorflow.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/tf_sig_build_dockerfiles/devel.usertools/squash_testlogs.py b/tensorflow/tensorflow/tools/tf_sig_build_dockerfiles/devel.usertools/squash_testlogs.py old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/tf_sig_build_dockerfiles/setup.cuda.sh b/tensorflow/tensorflow/tools/tf_sig_build_dockerfiles/setup.cuda.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/tools/tf_sig_build_dockerfiles/setup.packages.sh b/tensorflow/tensorflow/tools/tf_sig_build_dockerfiles/setup.packages.sh old mode 100644 new mode 100755 diff --git a/tensorflow/tensorflow/workspace2.bzl b/tensorflow/tensorflow/workspace2.bzl index 8601192ed..e722d3f7a 100644 --- a/tensorflow/tensorflow/workspace2.bzl +++ b/tensorflow/tensorflow/workspace2.bzl @@ -26,6 +26,8 @@ load("//third_party/dlpack:workspace.bzl", dlpack = "repo") load("//third_party/ducc:workspace.bzl", ducc = "repo") load("//third_party/eigen3:workspace.bzl", eigen3 = "repo") load("//third_party/farmhash:workspace.bzl", farmhash = "repo") +load("//third_party/ktfop:workspace.bzl", ktfop = "repo") +load("//third_party/kblas:workspace.bzl", kblas = "repo") load("//third_party/flatbuffers:workspace.bzl", flatbuffers = "repo") load("//third_party/gemmlowp:workspace.bzl", gemmlowp = "repo") load("//third_party/hexagon:workspace.bzl", hexagon_nn = "repo") @@ -67,6 +69,8 @@ def _initialize_third_party(): dlpack() eigen3() farmhash() + kblas() + ktfop() flatbuffers() gemmlowp() hexagon_nn() @@ -803,7 +807,7 @@ def _tf_repositories(): name = "upb", sha256 = "61d0417abd60e65ed589c9deee7c124fe76a4106831f6ad39464e1525cef1454", strip_prefix = "upb-9effcbcb27f0a665f9f345030188c0b291e32482", - patch_file = ["//third_party/grpc:upb_platform_fix.patch"], + patch_file = ["//third_party/grpc:upb_platform_fix.patch", "//third_party/grpc:upb_gcc10_compile_fix.patch"], urls = tf_mirror_urls("https://github.com/protocolbuffers/upb/archive/9effcbcb27f0a665f9f345030188c0b291e32482.tar.gz"), ) diff --git a/tensorflow/third_party/grpc/upb_gcc10_compile_fix.patch b/tensorflow/third_party/grpc/upb_gcc10_compile_fix.patch new file mode 100644 index 000000000..dd38391c9 --- /dev/null +++ b/tensorflow/third_party/grpc/upb_gcc10_compile_fix.patch @@ -0,0 +1,11 @@ +--- a/upb/upb.c 2025-05-30 17:01:35.956845750 +0800 ++++ b/upb/upb.c 2025-05-30 16:54:07.768845750 +0800 +@@ -37,7 +37,7 @@ + void upb_status_seterrmsg(upb_status *status, const char *msg) { + if (!status) return; + status->ok = false; +- strncpy(status->msg, msg, sizeof(status->msg)); ++ strncpy(status->msg, msg, sizeof(status->msg) - 1); + nullz(status); + } + diff --git a/tensorflow/third_party/kblas/BUILD b/tensorflow/third_party/kblas/BUILD new file mode 100644 index 000000000..1a1c21654 --- /dev/null +++ b/tensorflow/third_party/kblas/BUILD @@ -0,0 +1,21 @@ +load("@bazel_skylib//:bzl_library.bzl", "bzl_library") + +package( + # copybara:uncomment default_applicable_licenses = ["//tensorflow:license"], + default_visibility = ["//visibility:public"], + licenses = ["notice"], +) + +exports_files(["LICENSE"]) + +config_setting( + name = "build_with_kblas", + define_values = { + "build_with_kblas": "true", + }, +) + +bzl_library( + name = "build_defs_bzl", + srcs = ["build_defs.bzl"], +) diff --git a/tensorflow/third_party/kblas/build_defs.bzl b/tensorflow/third_party/kblas/build_defs.bzl new file mode 100644 index 000000000..77a211b88 --- /dev/null +++ b/tensorflow/third_party/kblas/build_defs.bzl @@ -0,0 +1,8 @@ +"""Starlark macros for kblas. +""" + +def if_kblas(if_true, if_false = []): + return select({ + "@org_tensorflow//third_party/kblas:build_with_kblas": if_true, + "//conditions:default": if_false, + }) diff --git a/tensorflow/third_party/kblas/kblas.BUILD b/tensorflow/third_party/kblas/kblas.BUILD new file mode 100644 index 000000000..04a6d512b --- /dev/null +++ b/tensorflow/third_party/kblas/kblas.BUILD @@ -0,0 +1,12 @@ +cc_import( + name = "kblas_so", + shared_library = "lib/sve/kblas/locking/libkblas.so", + visibility = ["//visibility:public"], +) + +cc_library( + name = "kblas", + hdrs = ["include/kblas.h"], + includes = ["include"], + visibility = ["//visibility:public"], +) \ No newline at end of file diff --git a/tensorflow/third_party/kblas/workspace.bzl b/tensorflow/third_party/kblas/workspace.bzl new file mode 100644 index 000000000..c438aa090 --- /dev/null +++ b/tensorflow/third_party/kblas/workspace.bzl @@ -0,0 +1,10 @@ +"""Provides the repository macro to import kblas.""" + +def repo(): + """Imports kblas.""" + + native.new_local_repository( + name = "kblas_archive", + build_file = "@org_tensorflow//third_party/kblas:kblas.BUILD", + path = "/usr/local/kml", + ) diff --git a/tensorflow/third_party/ktfop/BUILD b/tensorflow/third_party/ktfop/BUILD new file mode 100644 index 000000000..7aec8b6cf --- /dev/null +++ b/tensorflow/third_party/ktfop/BUILD @@ -0,0 +1,21 @@ +load("@bazel_skylib//:bzl_library.bzl", "bzl_library") + +package( + # copybara:uncomment default_applicable_licenses = ["//tensorflow:license"], + default_visibility = ["//visibility:public"], + licenses = ["notice"], +) + +exports_files(["LICENSE"]) + +config_setting( + name = "build_with_ktfop", + define_values = { + "build_with_ktfop": "true", + }, +) + +bzl_library( + name = "build_defs_bzl", + srcs = ["build_defs.bzl"], +) diff --git a/tensorflow/third_party/ktfop/build_defs.bzl b/tensorflow/third_party/ktfop/build_defs.bzl new file mode 100644 index 000000000..41cef1eb5 --- /dev/null +++ b/tensorflow/third_party/ktfop/build_defs.bzl @@ -0,0 +1,8 @@ +"""Starlark macros for ktfop. +""" + +def if_ktfop(if_true, if_false = []): + return select({ + "@org_tensorflow//third_party/ktfop:build_with_ktfop": if_true, + "//conditions:default": if_false, + }) diff --git a/tensorflow/third_party/ktfop/ktfop.BUILD b/tensorflow/third_party/ktfop/ktfop.BUILD new file mode 100644 index 000000000..063365f69 --- /dev/null +++ b/tensorflow/third_party/ktfop/ktfop.BUILD @@ -0,0 +1,14 @@ +cc_import( + name = "ktfop_so", + shared_library = "lib/sve/libktfop.so", + deps = ["@kblas_archive//:kblas_so"], +) + +cc_library( + name = "ktfop", + hdrs = ["include/ktfop.h"], + includes = ["include"], + deps = [":ktfop_so", + "@kblas_archive//:kblas"], + visibility = ["//visibility:public"], +) diff --git a/tensorflow/third_party/ktfop/workspace.bzl b/tensorflow/third_party/ktfop/workspace.bzl new file mode 100644 index 000000000..76163ae89 --- /dev/null +++ b/tensorflow/third_party/ktfop/workspace.bzl @@ -0,0 +1,10 @@ +"""Provides the repository macro to import ktfop.""" + +def repo(): + """Imports ktfop.""" + + native.new_local_repository( + name = "ktfop_archive", + build_file = "@org_tensorflow//third_party/ktfop:ktfop.BUILD", + path = "/usr/local/sra_inference", + ) diff --git a/tensorflow/third_party/xla/.kokoro/generate_index_html.sh b/tensorflow/third_party/xla/.kokoro/generate_index_html.sh old mode 100644 new mode 100755 diff --git a/tensorflow/third_party/xla/tools/ci_build/gpu_build/parallel_gpu_execute.sh b/tensorflow/third_party/xla/tools/ci_build/gpu_build/parallel_gpu_execute.sh old mode 100644 new mode 100755 -- Gitee From 61feb9f3ae7e79cbef17b2d47685112989458f8b Mon Sep 17 00:00:00 2001 From: ruanding <410226737@qq.com> Date: Tue, 3 Jun 2025 17:56:21 +0800 Subject: [PATCH 2/2] fix the bug in the test script --- tensorflow/unit_test.sh | 4 ++-- 1 file changed, 2 insertions(+), 2 deletions(-) diff --git a/tensorflow/unit_test.sh b/tensorflow/unit_test.sh index a71a44239..b76aa98f9 100644 --- a/tensorflow/unit_test.sh +++ b/tensorflow/unit_test.sh @@ -1,9 +1,9 @@ export TF_NEED_CUDA=0 export TF_NEED_ROCM=0 export TF_NEED_CLANG=0 -export CC_OPT_FLAGS='march=armv8.3-a+crc' +export CC_OPT_FLAGS='-march=armv8.3-a+crc' -export PYTHON_BIN_PATH='which python' +export PYTHON_BIN_PATH=$(which python) yes "" | $PYTHON_BIN_PATH configure.py bazel --output_user_root=./test_output test --distdir=../serving/download --test_tag_filters=-no_oss,-oss_excluded,-gpu,-tpu,-benchmark-test --test_lang_filters=cc,java -k --test_timeout 300,450,1200,3600 --config=opt --test_output=errors --test_size_filters=small,medium,large --build_tests_only -- //tensorflow/core/... //tensorflow/compiler/jit/... -//tensorflow/core/tpu/... \ No newline at end of file -- Gitee