{ "schema": "csl.perf03-ts01-single-process-correlation/1", "purpose": "Non-reported, process-scoped CPU sampling for PERF-03; not part of published latency statistics.", "identity": { "issue": 220, "source_commit": "2a5ab835295dfd22f02b19d440bb8158267d985a", "baseline_harness_commit": "4fc8e28587a7fe0e4130f1cb435cf6502ba0e4e2", "original_harness_sha256": "EDE47392E0C17C1A537251C2F1DA9989BC13D998F3459021DC5E05AB7AC399DA", "profile_harness_sha256": "A8F2679D4DE76B194A13CAFDDA63A56F2287A1AE884BBE497B73AB43BA77906B", "harness_delta": "Only the main repetition count changed from 7 to 30000; one_run() and its T-S01 validation body are unchanged." }, "build": { "compiler": "MSVC 19.44.35215.0", "visual_studio": "17.14.19 build tools", "cmake": "3.31.6-msvc6", "generator": "Visual Studio 17 2022 x64", "configuration": "Release", "flags": [ "/O2 (MaxSpeed)", "/Zi", "/DEBUG", "NDEBUG", "/MD" ], "highs": "1.11.0", "executable_sha256": "BC4BE7647964F51A0458BBAC0A7E81356C4F5E2B610507035FD2A58E5C78EBFA", "executable_bytes": 6714880, "pdb_sha256": "808E5B8453401245A9C02ABDEE98BEB9EBF5E7CFE762DEFA9BCE0CCD59326FB8", "pdb_bytes": 66424832, "symbol_resolution": "xperf resolved target executable functions from the colocated matching PDB; no source line-level DWARF/PDB line report was used.", "pinned_highs_archive_sha256": "2b44b074cf41439325ce4d0bbdac2d51379f56faf17ba15320a410d3c1f07275" }, "workload": { "profile": "ts01_exact_integrator", "single_process_pid": 23164, "warmups": 1, "measured_repetitions": 30000, "per_repetition_work": "The unchanged one_run() body constructs/resolves the T-S01 problem, prepares registered Strategy B and HiGHS LP/convex providers, executes StrategyPort, verifies the registered complete optimal family and trajectory anchor, and checks the literal canonical H-relation/membership points.", "successes": 30000, "failures": 0, "stdout_records_including_warmup": 30001 }, "paired_execution": { "unprofiled": { "command": "csl_syn04_strategy_b_report.exe", "exit_code": 0, "wall_ms": 36136, "stdout_bytes": 21313038, "stderr_bytes": 0, "stdout_sha256": "E7DA5CD508DE145E733145DAD72FA544E5E5C249865582EB74B35E43F0D58423", "per_call": { "n": 30000, "median_ns": 1130300.0, "mean_ns": 1155787.4, "min_ns": 800700, "max_ns": 2333700, "range_width_ns": 1533000 } }, "profiled": { "command": "VSDiagnostics session 89 -> attach PID 23164 -> run same executable -> stop session", "exit_code": 0, "wall_ms": 33309, "stdout_bytes": 21302977, "stderr_bytes": 0, "stdout_sha256": "AACB489AFC03E89A063A2811C9754975491E1449461ABD9C8FB2BA767941CAD4", "per_call": { "n": 30000, "median_ns": 975100.0, "mean_ns": 1059753.1, "min_ns": 840200, "max_ns": 2184200, "range_width_ns": 1344000 } }, "assessment": "Inconclusive. This is one sequential unprofiled/profiled pair under a changing, busy host. The profiled median is lower but the distributions overlap substantially; no profiler-overhead ratio, speedup, or latency difference is claimed." }, "profiler": { "collector": "Visual Studio DiagnosticsHub VSDiagnostics 17.14.36412.1", "agent": "DiagnosticsHub.CpuAgent.dll", "config": "AgentConfigs/CpuUsageHigh.json", "sample_rate_hz": 4000, "mode": "Process-scoped sampled CPU stacks; collector started before target launch and attached only to PID 23164.", "privilege": "No SeSystemProfilePrivilege; process-scoped sampling succeeded. WPR was not used in this capture.", "commands": [ "VSDiagnostics.exe start 89 /loadConfig:AgentConfigs\\CpuUsageHigh.json /scratchLocation: /package:dir", "VSDiagnostics.exe update 89 /attach:23164", "VSDiagnostics.exe stop 89 /output:profiled.diagsession", "VSDiagnostics.exe expandDiagSession profiled.diagsession", "xperf.exe -i sc.user_aux.etl -symbols verbose -o target-stack.html -a stack -pid 23164 -butterfly 1", "xperf.exe -i sc.user_aux.etl -a profile -util" ], "trace": { "duration_seconds": 33.786631, "etl_bytes": 48037888, "etl_sha256": "87E6B8C444E62EE22DC346B9DF76DFE633F0E28C64C57FFA1D3243F182F31A46", "buffers_processed": 733, "total_events": 468186, "sampleprof_events": 211895, "stackwalk_events": 137835, "lost_events": 0, "target_process_stack_hits": 130813, "target_process_root_hits": 130813, "target_stack_report_sha256": "CECDE625DAB92A3BE008E771D2EF982E31A22962C887D1BA18930AFE3A41859C", "raw_etl_and_stack_html": "Retained locally; not committed. Hashes above identify the exact local artifacts." }, "session_id": 89, "session_guid": "5997E42F-003D-4F91-A845-6404CF289E84", "config_sha256": "2CA721E4D6CDA0D46521F40E5D330F8FA179C01530E409D9E22855748340920A", "symbol_search_path": "/Release (contains the exact matching executable PDB)" }, "host": { "cpu": "12th Gen Intel Core i5-12400F", "physical_cores": 6, "logical_processors": 12, "os_build": "Windows NT 10.0.22631 (build 22631)", "trace_load_observation": "xperf per-logical-CPU utilization over 34 intervals (33.79 s): mean 12.45%, range 0.03–69.32%; includes the target plus unrelated background work.", "load_context": "Host was concurrently running other compiler/build and desktop processes; system utilization is retained as trace context, not normalized away." }, "sampled_cpu_attribution": [ { "rank": 1, "symbol": "`anonymous namespace`::one_run", "inclusive_hits": 126542, "share_of_target_stack_hits_percent": 96.74, "source": "reports/algorithm-results/tooling/syn04/strategy_b_measure.cpp:60" }, { "rank": 2, "symbol": "csl::registry::StrategyPort::run", "inclusive_hits": 118950, "share_of_target_stack_hits_percent": 90.93, "source": "src/registry/registry.cpp:65" }, { "rank": 3, "symbol": "csl::synthesis::strategy_b::DirectJointWorkspace::solve", "inclusive_hits": 116354, "share_of_target_stack_hits_percent": 88.95, "source": "src/synthesis/strategy_b/direct_joint.cpp:218" }, { "rank": 4, "symbol": "csl::adapters::convex::Workspace::solve_native", "inclusive_hits": 104178, "share_of_target_stack_hits_percent": 79.64, "source": "src/adapters/convex/highs.cpp:320" }, { "rank": 5, "symbol": "Highs::optimizeModel", "inclusive_hits": 60301, "share_of_target_stack_hits_percent": 46.1, "source": "HiGHS 1.11.0 highs/lp_data/Highs.cpp:958" }, { "rank": 6, "symbol": "Highs::callSolveQp", "inclusive_hits": 47804, "share_of_target_stack_hits_percent": 36.54, "source": "HiGHS 1.11.0 highs/lp_data/Highs.cpp:3837" } ], "conclusion": "This single-process profile provides source-resolved evidence that sampled CPU work during the repeated, individually timed T-S01 one_run() operation is concentrated in the public StrategyPort -> Strategy B DirectJointWorkspace -> convex HiGHS QP path. It supports identifying that path as the sampled CPU hotspot for this exact diagnostic workload. It does not quantify exact unprofiled single-call latency attribution, generalize beyond T-S01, or support speedup, comparison, application-scale, or scalability claims. Independent review is still required for issue closure.", "limitations": [ "One profiled and one unprofiled execution only; order was unprofiled then profiled, not alternated.", "The one_run() body is unchanged, but the diagnostic harness repetition count is increased to 30,000 to obtain usable in-process samples.", "PDB-enabled Release semantics were retained, but this is not the exact no-PDB published binary.", "Per-call timings include synchronous microsecond-scale operations; median and range vary substantially, and the paired wall-time difference is inconclusive.", "Windows host utilization includes background work." ] }