Files
pytorch/test/cpp/jit/CMakeLists.txt
Shivam Raikundalia 3ebbeb75fd [Profiler] Make Kineto traces export ns granularity for finer timestamps (#122425) (#123650)
Summary:

Kineto traces use microsecond level granularity because of chrome tracing defaults to that precision. Fix by adding preprocessor flag to TARGETS and BUCK files. Also remove any unnecessary ns to us conversions made in the profiler itself.

This diff contains profiler changes only. Libkineto changes found in D54964435.

Test Plan:
Check JSON and chrome tracing to make sure values are as expected. Tracing with flags enabled should have ns precision. Tracings without flags should be same as master.
Zoomer: https://www.internalfb.com/intern/zoomer/?profiling_run_fbid=796886748550189
Ran key_averages() to make sure FunctionEvent code working as expected:
--  ------------  ------------
                                                   Name    Self CPU %      Self CPU   CPU total %     CPU total  CPU time avg     Self CUDA   Self CUDA %    CUDA total  CUDA time avg    # of Calls

                                          ProfilerStep*         0.74%       3.976ms        64.40%     346.613ms      69.323ms       0.000us         0.00%      61.710ms      12.342ms             5
                      Optimizer.zero_grad#SGD.zero_grad         0.76%       4.109ms         0.76%       4.109ms     821.743us       0.000us         0.00%       0.000us       0.000us             5
                                          ## forward ##         6.89%      37.057ms        27.19%     146.320ms      29.264ms       0.000us         0.00%      58.708ms      11.742ms             5
                                           aten::conv2d         0.22%       1.176ms         7.74%      41.658ms     157.199us       0.000us         0.00%      27.550ms     103.962us           265
                                      aten::convolution         0.79%       4.273ms         7.52%      40.482ms     152.762us       0.000us         0.00%      27.550ms     103.962us           265
                                     aten::_convolution         0.69%       3.688ms         6.73%      36.209ms     136.637us       0.000us         0.00%      27.550ms     103.962us           265
                                aten::cudnn_convolution         6.04%      32.520ms         6.04%      32.520ms     122.719us      27.550ms         8.44%      27.550ms     103.962us           265
                                             aten::add_         2.42%      13.045ms         2.42%      13.045ms      30.694us      12.700ms         3.89%      12.700ms      29.882us           425
                                       aten::batch_norm         0.19%       1.027ms         8.12%      43.717ms     164.971us       0.000us         0.00%      16.744ms      63.185us           265
                           aten::_batch_norm_impl_index         0.31%       1.646ms         7.93%      42.691ms     161.096us       0.000us         0.00%      16.744ms      63.185us           265
-------------------------------------------------------  ------------  ------------  ------------  ------------  ------------  ------------  ------------  ------------  ------------  ------------

Differential Revision: D55925068

Pull Request resolved: https://github.com/pytorch/pytorch/pull/123650
Approved by: https://github.com/aaronenyeshi
2024-04-11 04:29:20 +00:00

163 lines
5.5 KiB
CMake

set(JIT_TEST_ROOT ${TORCH_ROOT}/test/cpp/jit)
# Build separate libraries the define custom classes/operators used from our Python tests.
# These are intended to be used with torch.ops.load_library() in our Python test suite.
add_library(torchbind_test SHARED
${JIT_TEST_ROOT}/test_custom_class_registrations.h
${JIT_TEST_ROOT}/test_custom_class_registrations.cpp
)
target_link_libraries(torchbind_test torch)
add_library(jitbackend_test SHARED ${JIT_TEST_ROOT}/test_backend_lib.cpp)
target_link_libraries(jitbackend_test torch)
set(BACKEND_WITH_COMPILER_SRCS
${JIT_TEST_ROOT}/test_backend_compiler_lib.cpp
${JIT_TEST_ROOT}/test_backend_compiler_preprocess.cpp
)
if(USE_KINETO)
# Testing edge profiler for backend use
# profiler_edge should only be aded when USE_KINETO flag is on
list(APPEND BACKEND_WITH_COMPILER_SRCS
${TORCH_SRC_DIR}/csrc/jit/mobile/profiler_edge.cpp)
endif()
add_library(backend_with_compiler SHARED
${BACKEND_WITH_COMPILER_SRCS}
)
if(USE_KINETO)
set_target_properties(backend_with_compiler PROPERTIES COMPILE_FLAGS
"-DUSE_KINETO -DTMP_LIBKINETO_NANOSECOND")
endif()
target_link_libraries(backend_with_compiler torch)
if(INSTALL_TEST)
install(TARGETS torchbind_test DESTINATION lib)
install(TARGETS jitbackend_test DESTINATION lib)
install(TARGETS backend_with_compiler DESTINATION lib)
endif()
# Build the cpp gtest binary containing the cpp-only tests.
set(JIT_TEST_SRCS
${JIT_TEST_ROOT}/test_add_if_then_else.cpp
${JIT_TEST_ROOT}/test_alias_analysis.cpp
${JIT_TEST_ROOT}/test_argument_spec.cpp
${JIT_TEST_ROOT}/test_autodiff.cpp
${JIT_TEST_ROOT}/test_load_upgraders.cpp
${JIT_TEST_ROOT}/test_op_replacement.cpp
${JIT_TEST_ROOT}/test_upgrader_utils.cpp
${JIT_TEST_ROOT}/test_backend.cpp
${JIT_TEST_ROOT}/test_class_import.cpp
${JIT_TEST_ROOT}/test_class_parser.cpp
${JIT_TEST_ROOT}/test_class_type.cpp
${JIT_TEST_ROOT}/test_code_template.cpp
${JIT_TEST_ROOT}/test_concat_opt.cpp
${JIT_TEST_ROOT}/test_constant_pooling.cpp
${JIT_TEST_ROOT}/test_cleanup_passes.cpp
${JIT_TEST_ROOT}/test_create_autodiff_subgraphs.cpp
${JIT_TEST_ROOT}/test_custom_class.cpp
${JIT_TEST_ROOT}/test_custom_class_registrations.h
${JIT_TEST_ROOT}/test_custom_class_registrations.cpp
${JIT_TEST_ROOT}/test_custom_operators.cpp
${JIT_TEST_ROOT}/test_dce.cpp
${JIT_TEST_ROOT}/test_fuser.cpp
${JIT_TEST_ROOT}/test_graph_executor.cpp
${JIT_TEST_ROOT}/test_graph_iterator.cpp
${JIT_TEST_ROOT}/test_cs_debug_info_serialization.cpp
${JIT_TEST_ROOT}/test_inliner.cpp
${JIT_TEST_ROOT}/test_interface.cpp
${JIT_TEST_ROOT}/test_interpreter.cpp
${JIT_TEST_ROOT}/test_ir.cpp
${JIT_TEST_ROOT}/test_irparser.cpp
${JIT_TEST_ROOT}/test_jit_type.cpp
${JIT_TEST_ROOT}/test_lite_interpreter.cpp
${JIT_TEST_ROOT}/test_lite_interpreter_direct.cpp
${JIT_TEST_ROOT}/test_lite_trainer.cpp
${JIT_TEST_ROOT}/test_memory_dag.cpp
${JIT_TEST_ROOT}/test_misc.cpp
${JIT_TEST_ROOT}/test_mobile_type_parser.cpp
${JIT_TEST_ROOT}/test_module_api.cpp
${JIT_TEST_ROOT}/test_peephole_optimize.cpp
${JIT_TEST_ROOT}/test_qualified_name.cpp
${JIT_TEST_ROOT}/test_save_load.cpp
${JIT_TEST_ROOT}/test_schema_info.cpp
${JIT_TEST_ROOT}/test_schema_matching.cpp
${JIT_TEST_ROOT}/test_stack_opt.cpp
${JIT_TEST_ROOT}/test_subgraph_matcher.cpp
${JIT_TEST_ROOT}/test_subgraph_rewriter.cpp
${JIT_TEST_ROOT}/test_subgraph_utils.cpp
${JIT_TEST_ROOT}/test_union.cpp
${JIT_TEST_ROOT}/test_utils.cpp
${JIT_TEST_ROOT}/test_script_profile.cpp
${JIT_TEST_ROOT}/test_shape_analysis.cpp
${JIT_TEST_ROOT}/test_jit_logging_levels.cpp
${JIT_TEST_ROOT}/test_file_format.cpp
${JIT_TEST_ROOT}/test_flatbuffer.cpp
)
add_executable(test_jit
${TORCH_ROOT}/test/cpp/common/main.cpp
${JIT_TEST_SRCS}
)
# We also build with UBSAN flag in build_asan.h
if(USE_ASAN)
target_compile_options(test_jit PRIVATE "-fsanitize=undefined")
target_link_libraries(test_jit PRIVATE "-fsanitize=undefined")
endif()
target_link_libraries(
test_jit PRIVATE flatbuffers)
# TODO temporary until we can delete the old gtest polyfills.
target_compile_definitions(test_jit PRIVATE USE_GTEST)
if(USE_SYSTEM_ONNX)
target_link_libraries(test_jit PRIVATE onnx_proto onnx)
endif()
if(USE_MKLDNN)
target_link_libraries(test_jit PRIVATE caffe2::mkldnn)
endif()
set(JIT_TEST_DEPENDENCIES torch gtest jitbackend_test backend_with_compiler)
if(MSVC)
list(APPEND JIT_TEST_DEPENDENCIES onnx_library)
endif(MSVC)
target_link_libraries(test_jit PRIVATE ${JIT_TEST_DEPENDENCIES})
target_include_directories(test_jit PRIVATE ${ATen_CPU_INCLUDE})
if(NOT MSVC)
target_compile_options(test_jit PRIVATE $<$<COMPILE_LANGUAGE:CXX>:-Wno-unused-variable>)
endif()
if(LINUX)
#Update to target_link_options when CMake version can be upgraded
target_link_libraries(test_jit PRIVATE "-Wl,--no-as-needed,$<TARGET_FILE:jitbackend_test>,$<TARGET_FILE:backend_with_compiler>,--as-needed")
endif()
if(USE_CUDA)
target_compile_definitions(test_jit PRIVATE USE_CUDA)
# Suppress sign compare checks for NVFUSER JIT tests
if(NOT MSVC)
target_compile_options(test_jit PRIVATE $<$<COMPILE_LANGUAGE:CXX>:-Wno-sign-compare>)
endif()
elseif(USE_ROCM)
target_link_libraries(test_jit PRIVATE
${ROCM_HIPRTC_LIB}
${PYTORCH_HIP_LIBRARIES}
${TORCH_CUDA_LIBRARIES})
target_compile_definitions(test_jit PRIVATE USE_ROCM)
endif()
if(INSTALL_TEST)
install(TARGETS test_jit DESTINATION bin)
# Install PDB files for MSVC builds
if(MSVC AND BUILD_SHARED_LIBS)
install(FILES $<TARGET_PDB_FILE:test_jit> DESTINATION bin OPTIONAL)
endif()
endif()