Use rocprofiler-sdk for RCCL-API tracing (#126)

- Add support for RCCL API tracing through rocprofiler-sdk.
- Refactored the comm_data code to use the SDK RCCL_API callbacks.
- Add a runtime version check for SDK to gate callback enablement, rather than just the compile-time check.
- Fixed: SAMPLING_TIMEOUT was not being handled correctly in add_test.
This commit is contained in:
David Galiffi
2025-06-06 11:36:17 -04:00
کامیت شده توسط GitHub
والد c5507e3740
کامیت af77d93f75
31فایلهای تغییر یافته به همراه350 افزوده شده و 1252 حذف شده
@@ -1359,6 +1359,14 @@ configure_disabled_settings(const std::shared_ptr<settings>& _config)
_config->find("ROCPROFSYS_USE_AMD_SMI")->second->set_hidden(true);
for(const auto& itr : _config->disable_category("amd_smi"))
_config->find(itr)->second->set_hidden(true);
_config->find("ROCPROFSYS_USE_RCCLP")->second->set_hidden(true);
for(const auto& itr : _config->disable_category("rcclp"))
_config->find(itr)->second->set_hidden(true);
_config->find("ROCPROFSYS_USE_ROCM")->second->set_hidden(true);
for(const auto& itr : _config->disable_category("rocm"))
_config->find(itr)->second->set_hidden(true);
#endif
#if defined(ROCPROFSYS_USE_OMPT) || ROCPROFSYS_USE_OMPT == 0