Use logical_node_id for mapping rocprofiler agents to HSA agents (#708)

* Temp: Fixing node id

* source formatting (clang-format v11) (#709)

Co-authored-by: ammarwa <3832908+ammarwa@users.noreply.github.com>

* Using logical node id

* Update agent.cpp

* Update agent.cpp

* Python formatting

---------

Co-authored-by: Ammar ELWazir <aelwazir@rocprofiler1.amd.com>
Co-authored-by: github-actions[bot] <41898282+github-actions[bot]@users.noreply.github.com>
Co-authored-by: ammarwa <3832908+ammarwa@users.noreply.github.com>
Co-authored-by: Ammar ELWazir <aelwazir@rocprofiler2.amd.com>
Co-authored-by: Jonathan R. Madsen <jonathanrmadsen@gmail.com>

[ROCm/rocprofiler-sdk commit: 62625d0aa1]

projects/rocprofiler-sdk/source/lib/rocprofiler-sdk/agent.cpp

@@ -755,7 +755,9 @@ construct_agent_cache(::HsaApiTable* table)
 
                 for(const auto* ritr : rocp_agents)
                 {
-                    if(ritr->node_id == internal_node_id)
+                    // TODO(aelwazir): To be changed back to use node id once ROCR fixes
+                    // the hsa_agents to use the real node id
+                    if(ritr->logical_node_id == static_cast<int64_t>(internal_node_id))
                     {
                         rocp_hsa_agent_node_ids.erase(internal_node_id);
                         break;
@@ -799,9 +801,11 @@ construct_agent_cache(::HsaApiTable* table)
                    static_cast<hsa_agent_info_t>(HSA_AMD_AGENT_INFO_DRIVER_NODE_ID),
                    &node_id) == HSA_STATUS_SUCCESS)
             {
-                if(ritr->node_id == node_id)
+                // TODO(aelwazir): To be changed back to use node id once ROCR fixes
+                // the hsa_agents to use the real node id
+                if(ritr->logical_node_id == static_cast<int64_t>(node_id))
                 {
-                    agent_map.emplace(ritr->node_id, std::make_tuple(ritr, hitr));
+                    agent_map.emplace(ritr->logical_node_id, std::make_tuple(ritr, hitr));
                     get_agent_mapping().emplace_back(agent_pair{ritr, hitr});
                     break;
                 }
@@ -888,8 +892,10 @@ construct_agent_cache(::HsaApiTable* table)
         {
             if(rocp_agent->type == ROCPROFILER_AGENT_TYPE_GPU)
             {
+                // TODO(aelwazir): To be changed back to use node id once ROCR fixes
+                // the hsa_agents to use the real node id
                 LOG(ERROR) << fmt::format("rocprofiler agent <-> HSA agent mapping failed: {} ({})",
-                                          rocp_agent->node_id,
+                                          rocp_agent->logical_node_id,
                                           err.what());
             }
         }

projects/rocprofiler-sdk/source/lib/rocprofiler-sdk/tests/agent.cpp

@@ -244,7 +244,9 @@ TEST(rocprofiler_lib, agent)
         EXPECT_EQ(std::string_view{agent->product_name},
                   std::string_view{hsa_agent->device_mkt_name})
             << msg;
-        EXPECT_EQ(agent->node_id, hsa_agent->internal_node_id) << msg;
+        // TODO(aelwazir): To be changed back to use node id once ROCR fixes the hsa_agents to use
+        // the real node id
+        EXPECT_EQ(agent->logical_node_id, hsa_agent->internal_node_id) << msg;
         EXPECT_EQ(agent->location_id, hsa_agent->bdf_id) << msg;
         EXPECT_EQ(agent->device_id, hsa_agent->chip_id) << msg;
         EXPECT_EQ(agent->simd_count, hsa_agent->compute_unit * hsa_agent->simds_per_cu) << msg;

projects/rocprofiler-sdk/tests/counter-collection/validate.py

@@ -30,9 +30,17 @@ def test_counter_values(input_data):
             scaling_factor = 64 / itr["wave_front_size"]
             break
 
-    for itr in data["rocprofiler-sdk-json-tool"]["buffer_records"]["counter_collection"]:
-        value = itr["counter_value"]
-        assert int(round(value, 0)) == int(round(1 * scaling_factor, 0)), str(data)
+        scaling_factor = 64 / itr["wave_front_size"]
+
+        for itr in data["rocprofiler-sdk-json-tool"]["buffer_records"][
+            "counter_collection"
+        ]:
+            value = itr["counter_value"]
+            assert int(round(value, 0)) == int(round(1 * scaling_factor, 0)), (
+                f"Failure on agent "
+                f"{str(itr)} expected {1 * scaling_factor} but got {value} "
+                f"Debug Info:\n {str(data)}"
+            )
 
 
 if __name__ == "__main__":

projects/rocprofiler-sdk/tests/kernel-tracing/CMakeLists.txt

@@ -32,7 +32,7 @@ set(kernel-tracing-env
 
 set_tests_properties(
     test-kernel-tracing-execute
-    PROPERTIES TIMEOUT 45 LABELS "integration-tests" ENVIRONMENT "${kernel-tracing-env}"
+    PROPERTIES TIMEOUT 100 LABELS "integration-tests" ENVIRONMENT "${kernel-tracing-env}"
                FAIL_REGULAR_EXPRESSION "${ROCPROFILER_DEFAULT_FAIL_REGEX}")
 
 foreach(FILENAME validate.py pytest.ini conftest.py)