a23bf28aaa2d5e2a5c6b7b4b5ea77f879d4bd776
3 Коммитов
| Автор | SHA1 | Сообщение | Дата | |
|---|---|---|---|---|
|
|
4ae26e2d08 |
rocm-smi and KokkosTools support (#23)
* renamed omnitrace_thread_data to thread_data
* initial implementation
* Numerous fixes and updates
- Updated timemory submodule
- Updated perfetto submodule (pulls in fixes for TRACE_EVENT)
- pthread_gotcha only after omnitrace_init_tooling
- omnitrace banner
- config settings for rocm-smi freq and devices
- critical_trace::get_entries
- OMNITRACE_BASIC_PRINT
- rocm_smi perfetto category
- redirect roctracer warnings for ROCm 4.5.0
- property specializations for rocm-smi components
- units fixes data_tracker types
- roctracer entries for pthread_create and start_thread
- omnitrace-avail defaults to settings, not components
- settings have conforming names
- settings warn about duplicates
- ptl named threads
- decreased max freq for sampler SIGALRM
- rocm-smi names thread
- rocm-smi avoids call to hipGetDeviceCount
- name roctracer activity callback threads
- fixed binary rewrite test output names
* Update lulesh example
- supports non-UVM GPU
* Lulesh tweaks + formatting
* KokkosP + Mode + Roctracer sampling deadlock fix
- kokkosp support
- omnitrace_init_library
- config::print_settings()
- config::get_mode()
- omnitrace::Mode
- omnitrace-avail improvements (removes settings)
- handle get_verbose() < 0
- disable dyninst InstrStackFrames by default
- handle perf_event_paranoid > 1 by disabling PAPI
- SIGALRM max freq to 5.0
- Name threads
- rocm-smi handles get_use_perfetto() and get_use_timemory()
- HSA_ENABLE_INTERRUPT=0 when roctracer + sampling (fixes deadlock)
* Tests, API renaming, roctracer
- disable renaming of thread 0
- verbprintf_bare
- enable dyninst merge tramp
- tweaked some omnitrace exe verbose levels
- reworked roctracer::setup and roctracer::shutdown
- rocm_smi::data::poll checks get_state()
- omnitrace_trace_finalize -> omnitrace_finalize
- omnitrace_trace_init -> omnitrace_init
- omnitrace_trace_set_env -> omnitrace_set_env
- omnitrace_trace_set_mpi -> omnitrace_set_mpi
- sampling mode does not disable timemory
- disable roctracer before shutting down rocm-smi
- lulesh tests w/ and w/o kokkosp
- lulesh tests for perfetto only
- with --dynamic-callsites --traps --allow-overlapping
- lulesh tests for timemory only
- with --stdlib --dynamic-callsites --traps --allow-overlapping
* Update timemory submodule
- fix for TIMEMORY_PROPERTY_SPECIALIZATION
* get_verbose() handling + timemory submodule update
- Findroctracer.cmake uses find_package(hsakmt)
* Stability fixes + rework roctracer + perfetto
- reworked roctracer start up
- critical_trace perfetto basic values
- perfetto sampling category
- sampler checks signals
- peak_rss in sampling
- pthread_gotcha::shutdown()
- rocm_smi::device_count()
- HSA_TOOLS_LIB is set
- HSA_ENABLE_INTERRUPT in omnitrace exe
- omnitrace exe verbosity level changes
- Avoid instrumenting Impl ns in Kokkos
- gpu::device_count prefers rocm_smi instead of hip
- ptl blocks signals
- fixed pthread_gotcha roctracer_data values
- removed runtime-instrument-sampling tests
- timemory submodule update
* cmake formatting
* timemory + roctracer updates
- fix timemory issue with papi_common
- fix timemory issue with units
- define roctracer::is_setup()
* Miscellaneous tweaks
- Disable sampling during runtime instrument
- Fixed warnings about dynamic callsites
- Fixed backtrace output when timemory disabled
- Test tweaks
* cmake-format
* omnitrace_target_compile_definitions
* timemory submodule update
* config, omnitrace, State, mpi_gotcha updates
- use OMNITRACE_THROW instead of direct throw
- is_attached()
- is_binary_rewrite()
- get_is_continuous_integration()
- get_debug_init()
- get_debug_finalize()
- max_thread_bookmarks default to 1
- State::Init
- app_thread oneTimeCode
- runtime instrumentation uses waitpid
- fixed init_names
- include main in MPI runs
- fixed sampling setup when disabled
- reworked mpi_gotcha
- disabled critical trace in transpose test
* cmake-format
* handle rocm_smi::device_count() exception
* CI timeouts
* Re-enable runtime-instrument + sampling
[ROCm/rocprofiler-systems commit:
|
||
|
|
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:
|
||
|
|
cdd2707058 |
v0.0.3: MPI wrappers w/o full MPI support + setup-env.sh (#15)
* v0.0.3: MPI wrappers w/o full MPI support + setup-env.sh
- bumped to v0.0.3
- enabled gotcha wrapping of MPI functions w/o enabling MPI
- added setup-env.sh script
- minor updates to testing
- Update timemory submodule
- fixes tim::component::configure_mpip undefined symbol
- Script updates
[ROCm/rocprofiler-systems commit:
|