AI/rocm-systems
2
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
Files
efb6d766afd455f20af259cd47fb4182e51d0d32
rocm-systems/projects/rocprofiler-systems/cmake/Packages.cmake
T
Jonathan R. Madsen efb6d766af Reorganization and critical trace support (#17) 
* Roctracer wall clock integration (#16)

* Integrates roctracer values into wall-clock

* Fixed scoping + timemory roctracer

* Fixed data race in roctracer

* Synchronized HIP API on main thread

- Cache hip activity callbacks and execute on main thread
- Minor updates to transpose

* Debugging + MPI + transpose updates

* PTL + HSA and timemory + kernel timing

- PTL usage fixed HSA + timemory issues bc we could control the thread destruction
- Fixed laps counting in roctracer callbacks

* Ignore select HIP API types

- The ignored API types are ignored because there appears to be a bug
  which causes the "end" callback to be labeled as begin
- hipDeviceEnablePeerAccess
- hipImportExternalMemory
- hipDestroyExternalMemory

* Tweaks to PTL config

* Timemory update + pid-prefix w/ mpi headers

- %pid%- prefix with mpi headers
- timemory submodule update

* CMake + critical trace + reorganize library source

- clang-tidy tweaks
- cmake function updates to use hosttrace_ prefix
- update gitignore
- cmake HOSTTRACE_MAX_THREADS option
- Formatting.cmake
- cleaned up MacroUtilities.cmake
- PTL submodule + usage
- tweak to Findroctracer.cmake
- MT transpose
- Updated PTL submodule
- Updated timemory submodule
- fix to hosttrace return value type if type not found
- reorganized library source code
- support for critical trace

* Remove bits/stdint-uintn.h headers

* Rename + config + depth + critical path

- rename hosttrace_timemory_data to instrumentation_bundles
- rename hosttrace_bundle_t to main_bundle_t
- rename bundle_t to instrumentation_bundle_t
- rework of configuration setup
- critical_trace write directly to file option
- tweaked depth calculation
- updated timemory submodule
- improved parallel support in roctracer callbacks
- working critical_trace
- perfetto device-critical-trace and host-critical-trace categories
- made transpose example parallel
- made parallel-overhead example a bit uneven
- relocated LTO activation

* Fixed duplicates in perfetto critical-trace

* reworked critical trace support

- substantial perf improvement (30-45 min -> 30 sec)
- changes to configuration (new and removed options)

* Removed "%pid%-" output prefix in mpi_gotcha

* Update timemory submodule

[ROCm/rocprofiler-systems commit: 752424efc2]
2021-11-23 02:53:14 -06:00

377 lines
14 KiB
CMake
Raw Blame History

# include guard
include_guard(DIRECTORY)
# ########################################################################################
#
# External Packages are found here
#
# ########################################################################################
hosttrace_add_interface_library(
hosttrace-headers "Provides minimal set of include flags to compile with hosttrace")
hosttrace_add_interface_library(hosttrace-threading "Enables multithreading support")
hosttrace_add_interface_library(
hosttrace-dyninst
"Provides flags and libraries for Dyninst (dynamic instrumentation)")
hosttrace_add_interface_library(hosttrace-roctracer
"Provides flags and libraries for roctracer")
hosttrace_add_interface_library(hosttrace-mpi "Provides MPI or MPI headers")
hosttrace_add_interface_library(hosttrace-ptl "Enables PTL support (tasking)")
target_include_directories(hosttrace-headers INTERFACE ${PROJECT_SOURCE_DIR}/include
${PROJECT_BINARY_DIR}/include)
# include threading because of rooflines
target_link_libraries(hosttrace-headers INTERFACE hosttrace-threading)
# ----------------------------------------------------------------------------------------#
#
# Threading
#
# ----------------------------------------------------------------------------------------#
if(NOT WIN32)
set(CMAKE_THREAD_PREFER_PTHREAD ON)
set(THREADS_PREFER_PTHREAD_FLAG OFF)
endif()
find_library(pthread_LIBRARY NAMES pthread pthreads)
find_package_handle_standard_args(pthread-library REQUIRED_VARS pthread_LIBRARY)
find_package(Threads ${hosttrace_FIND_QUIETLY} ${hosttrace_FIND_REQUIREMENT})
if(Threads_FOUND)
target_link_libraries(hosttrace-threading INTERFACE ${CMAKE_THREAD_LIBS_INIT})
endif()
if(pthread_LIBRARY AND NOT WIN32)
target_link_libraries(hosttrace-threading INTERFACE ${pthread_LIBRARY})
endif()
# ----------------------------------------------------------------------------------------#
#
# roctracer
#
# ----------------------------------------------------------------------------------------#
if(HOSTTRACE_USE_ROCTRACER)
list(APPEND CMAKE_PREFIX_PATH /opt/rocm)
find_package(roctracer ${hosttrace_FIND_QUIETLY} REQUIRED)
find_package(hip ${hosttrace_FIND_QUIETLY} REQUIRED)
target_compile_definitions(hosttrace-roctracer INTERFACE HOSTTRACE_USE_ROCTRACER)
target_link_libraries(hosttrace-roctracer INTERFACE hip::host roctracer::roctracer)
set(CMAKE_INSTALL_RPATH "${CMAKE_INSTALL_RPATH}:${roctracer_LIBRARY_DIRS}")
endif()
# ----------------------------------------------------------------------------------------#
#
# MPI
#
# ----------------------------------------------------------------------------------------#
if(HOSTTRACE_USE_MPI)
find_package(MPI ${hosttrace_FIND_QUIETLY} REQUIRED)
target_link_libraries(hosttrace-mpi INTERFACE MPI::MPI_C MPI::MPI_CXX)
elseif(HOSTTRACE_USE_MPI_HEADERS)
find_package(MPI-Headers ${hosttrace_FIND_QUIETLY} REQUIRED)
target_compile_definitions(hosttrace-mpi INTERFACE TIMEMORY_USE_MPI_HEADERS
HOSTTRACE_USE_MPI_HEADERS)
target_link_libraries(hosttrace-mpi INTERFACE MPI::MPI_HEADERS)
endif()
# ----------------------------------------------------------------------------------------#
#
# Dyninst
#
# ----------------------------------------------------------------------------------------#
if(HOSTTRACE_BUILD_DYNINST)
hosttrace_checkout_git_submodule(
RELATIVE_PATH external/dyninst
WORKING_DIRECTORY ${PROJECT_SOURCE_DIR}
REPO_URL https://github.com/jrmadsen/dyninst.git
REPO_BRANCH formatting)
set(DYNINST_OPTION_PREFIX ON)
set(DYNINST_BUILD_DOCS OFF)
set(DYNINST_QUIET_CONFIG
ON
CACHE BOOL "Suppress dyninst cmake messages")
set(DYNINST_BUILD_PARSE_THAT
OFF
CACHE BOOL "Build dyninst parseThat executable")
set(DYNINST_BUILD_SHARED_LIBS
ON
CACHE BOOL "Build shared dyninst libraries")
set(DYNINST_BUILD_STATIC_LIBS
OFF
CACHE BOOL "Build static dyninst libraries")
set(DYNINST_ENABLE_LTO
OFF
CACHE BOOL "Enable LTO for dyninst libraries")
hosttrace_save_variables(PIC VARIABLES CMAKE_POSITION_INDEPENDENT_CODE)
set(CMAKE_POSITION_INDEPENDENT_CODE ON)
add_subdirectory(external/dyninst)
hosttrace_restore_variables(PIC VARIABLES CMAKE_POSITION_INDEPENDENT_CODE)
add_library(Dyninst::Dyninst INTERFACE IMPORTED)
foreach(_LIB common dyninstAPI parseAPI instructionAPI symtabAPI stackwalk Boost TBB)
target_link_libraries(Dyninst::Dyninst INTERFACE Dyninst::${_LIB})
endforeach()
target_link_libraries(hosttrace-dyninst INTERFACE Dyninst::Dyninst)
set(HOSTTRACE_DYNINST_API_RT
${PROJECT_BINARY_DIR}/external/dyninst/dyninstAPI_RT/libdyninstAPI_RT${CMAKE_SHARED_LIBRARY_SUFFIX}
)
if(HOSTTRACE_DYNINST_API_RT)
target_compile_definitions(
hosttrace-dyninst
INTERFACE
DYNINST_API_RT="${CMAKE_INSTALL_PREFIX}/${CMAKE_INSTALL_LIBDIR}:$<TARGET_FILE_DIR:Dyninst::dyninstAPI_RT>:${CMAKE_INSTALL_PREFIX}/lib/$<TARGET_FILE_NAME:Dyninst::dyninstAPI_RT>:$<TARGET_FILE:Dyninst::dyninstAPI_RT>"
)
endif()
else()
find_package(Dyninst ${hosttrace_FIND_QUIETLY} REQUIRED
COMPONENTS dyninstAPI parseAPI instructionAPI symtabAPI)
if(TARGET Dyninst::Dyninst) # updated Dyninst CMake system was found
# useful for defining the location of the runtime API
find_library(
HOSTTRACE_DYNINST_API_RT dyninstAPI_RT
HINTS ${Dyninst_ROOT_DIR} ${Dyninst_DIR}
PATHS ${Dyninst_ROOT_DIR} ${Dyninst_DIR}
PATH_SUFFIXES lib)
if(HOSTTRACE_DYNINST_API_RT)
target_compile_definitions(
hosttrace-dyninst INTERFACE DYNINST_API_RT="${HOSTTRACE_DYNINST_API_RT}")
endif()
hosttrace_add_rpath(${Dyninst_LIBRARIES})
target_link_libraries(hosttrace-dyninst INTERFACE Dyninst::Dyninst)
else() # updated Dyninst CMake system was not found
set(_BOOST_COMPONENTS atomic system thread date_time)
set(hosttrace_BOOST_COMPONENTS
"${_BOOST_COMPONENTS}"
CACHE STRING "Boost components used by Dyninst in hosttrace")
set(Boost_NO_BOOST_CMAKE ON)
find_package(Boost QUIET REQUIRED COMPONENTS ${hosttrace_BOOST_COMPONENTS})
# some installs of dyninst don't set this properly
if(EXISTS "${DYNINST_INCLUDE_DIR}" AND NOT DYNINST_HEADER_DIR)
get_filename_component(DYNINST_HEADER_DIR "${DYNINST_INCLUDE_DIR}" REALPATH
CACHE)
else()
find_path(
DYNINST_HEADER_DIR
NAMES BPatch.h dyninstAPI_RT.h
HINTS ${Dyninst_ROOT_DIR} ${Dyninst_DIR} ${Dyninst_DIR}/../../..
PATHS ${Dyninst_ROOT_DIR} ${Dyninst_DIR} ${Dyninst_DIR}/../../..
PATH_SUFFIXES include)
endif()
# useful for defining the location of the runtime API
find_library(
HOSTTRACE_DYNINST_API_RT dyninstAPI_RT
HINTS ${Dyninst_ROOT_DIR} ${Dyninst_DIR}
PATHS ${Dyninst_ROOT_DIR} ${Dyninst_DIR}
PATH_SUFFIXES lib)
# try to find TBB
find_package(TBB QUIET)
# if fail try to use the Dyninst installed FindTBB.cmake
if(NOT TBB_FOUND)
list(APPEND CMAKE_MODULE_PATH ${Dyninst_DIR}/Modules)
find_package(TBB QUIET)
endif()
if(NOT TBB_FOUND)
find_path(
TBB_INCLUDE_DIR
NAMES tbb/tbb.h
PATH_SUFFIXES include)
endif()
if(HOSTTRACE_DYNINST_API_RT)
target_compile_definitions(
hosttrace-dyninst INTERFACE DYNINST_API_RT="${HOSTTRACE_DYNINST_API_RT}")
endif()
if(Boost_DIR)
get_filename_component(Boost_RPATH_DIR "${Boost_DIR}" DIRECTORY)
get_filename_component(Boost_RPATH_DIR "${Boost_RPATH_DIR}" DIRECTORY)
if(EXISTS "${Boost_RPATH_DIR}" AND IS_DIRECTORY "${Boost_RPATH_DIR}")
set(CMAKE_INSTALL_RPATH "${CMAKE_INSTALL_RPATH}:${Boost_RPATH_DIR}")
endif()
endif()
hosttrace_add_rpath(${DYNINST_LIBRARIES} ${Boost_LIBRARIES})
target_link_libraries(hosttrace-dyninst INTERFACE ${DYNINST_LIBRARIES}
${Boost_LIBRARIES})
foreach(
_TARG
dyninst
dyninstAPI
instructionAPI
symtabAPI
parseAPI
headers
atomic
system
thread
date_time
TBB)
if(TARGET Dyninst::${_TARG})
target_link_libraries(hosttrace-dyninst INTERFACE Dyninst::${_TARG})
elseif(TARGET Boost::${_TARG})
target_link_libraries(hosttrace-dyninst INTERFACE Boost::${_TARG})
elseif(TARGET ${_TARG})
target_link_libraries(hosttrace-dyninst INTERFACE ${_TARG})
endif()
endforeach()
target_include_directories(
hosttrace-dyninst SYSTEM INTERFACE ${TBB_INCLUDE_DIR} ${Boost_INCLUDE_DIRS}
${DYNINST_HEADER_DIR})
target_compile_definitions(hosttrace-dyninst INTERFACE hosttrace_USE_DYNINST)
endif()
endif()
# ----------------------------------------------------------------------------------------#
#
# Perfetto
#
# ----------------------------------------------------------------------------------------#
set(perfetto_DIR ${PROJECT_SOURCE_DIR}/external/perfetto)
hosttrace_checkout_git_submodule(
RELATIVE_PATH external/perfetto
WORKING_DIRECTORY ${PROJECT_SOURCE_DIR}
REPO_URL https://android.googlesource.com/platform/external/perfetto
REPO_BRANCH v17.0
TEST_FILE sdk/perfetto.cc)
# ----------------------------------------------------------------------------------------#
#
# ELFIO
#
# ----------------------------------------------------------------------------------------#
if(HOSTTRACE_BUILD_DEVICETRACE)
hosttrace_checkout_git_submodule(
RELATIVE_PATH external/elfio
WORKING_DIRECTORY ${PROJECT_SOURCE_DIR}
REPO_URL https://github.com/jrmadsen/ELFIO.git
REPO_BRANCH set-offset-support)
add_subdirectory(external/elfio)
endif()
# ----------------------------------------------------------------------------------------#
#
# timemory submodule
#
# ----------------------------------------------------------------------------------------#
set(TIMEMORY_INSTALL_HEADERS
OFF
CACHE BOOL "Disable timemory header install")
set(TIMEMORY_INSTALL_CONFIG
OFF
CACHE BOOL "Disable timemory cmake configuration install")
set(TIMEMORY_INSTALL_ALL
OFF
CACHE BOOL "Disable install target depending on all target")
set(TIMEMORY_BUILD_C
OFF
CACHE BOOL "Disable timemory C library")
set(TIMEMORY_BUILD_FORTRAN
OFF
CACHE BOOL "Disable timemory Fortran library")
set(TIMEMORY_BUILD_TOOLS
OFF
CACHE BOOL "Ensure timem executable is built")
set(TIMEMORY_BUILD_EXCLUDE_FROM_ALL
ON
CACHE BOOL "Set timemory to only build dependencies")
set(TIMEMORY_QUIET_CONFIG
ON
CACHE BOOL "Make timemory configuration quieter")
# timemory feature settings
set(TIMEMORY_USE_GOTCHA
ON
CACHE BOOL "Enable GOTCHA support in timemory")
set(TIMEMORY_USE_PERFETTO
OFF
CACHE BOOL "Disable perfetto support in timemory")
# timemory feature build settings
set(TIMEMORY_BUILD_GOTCHA
ON
CACHE BOOL "Enable building GOTCHA library from submodule")
# timemory build settings
set(TIMEMORY_TLS_MODEL
"global-dynamic"
CACHE STRING "Thread-local static model" FORCE)
hosttrace_checkout_git_submodule(
RELATIVE_PATH external/timemory
WORKING_DIRECTORY ${PROJECT_SOURCE_DIR}
REPO_URL https://github.com/NERSC/timemory.git
REPO_BRANCH gpu-kernel-instrumentation)
hosttrace_save_variables(BUILD_CONFIG VARIABLES BUILD_SHARED_LIBS BUILD_STATIC_LIBS
CMAKE_POSITION_INDEPENDENT_CODE)
# ensure timemory builds PIC static libs so that we don't have to install timemory shared
# lib
set(BUILD_SHARED_LIBS ON)
set(BUILD_STATIC_LIBS OFF)
set(CMAKE_POSITION_INDEPENDENT_CODE ON)
add_subdirectory(external/timemory)
hosttrace_restore_variables(BUILD_CONFIG VARIABLES BUILD_SHARED_LIBS BUILD_STATIC_LIBS
CMAKE_POSITION_INDEPENDENT_CODE)
# ----------------------------------------------------------------------------------------#
#
# PTL (Parallel Tasking Library) submodule
#
# ----------------------------------------------------------------------------------------#
# timemory might provide PTL::ptl-shared
if(NOT TARGET PTL::ptl-shared)
hosttrace_checkout_git_submodule(
RELATIVE_PATH external/PTL
WORKING_DIRECTORY ${PROJECT_SOURCE_DIR}
REPO_URL https://github.com/jrmadsen/PTL.git
REPO_BRANCH master)
set(PTL_BUILD_EXAMPLES OFF)
set(PTL_USE_TBB OFF)
set(PTL_USE_GPU OFF)
set(PTL_DEVELOPER_INSTALL OFF)
hosttrace_save_variables(
BUILD_CONFIG
VARIABLES BUILD_SHARED_LIBS BUILD_STATIC_LIBS CMAKE_POSITION_INDEPENDENT_CODE
CMAKE_CXX_VISIBILITY_PRESET CMAKE_VISIBILITY_INLINES_HIDDEN)
set(BUILD_SHARED_LIBS ON)
set(BUILD_STATIC_LIBS OFF)
set(CMAKE_POSITION_INDEPENDENT_CODE ON)
set(CMAKE_VISIBILITY_INLINES_HIDDEN ON)
add_subdirectory(external/PTL)
hosttrace_restore_variables(
BUILD_CONFIG
VARIABLES BUILD_SHARED_LIBS BUILD_STATIC_LIBS CMAKE_POSITION_INDEPENDENT_CODE
CMAKE_CXX_VISIBILITY_PRESET CMAKE_VISIBILITY_INLINES_HIDDEN)
endif()
target_link_libraries(hosttrace-ptl INTERFACE PTL::ptl-shared)
Reference in New Issue View Git Blame Copy Permalink