Diagnostics & Profiling with SMonitor

MolSysMT integrates a built-in diagnostics and profiling engine managed through SMonitor to track execution timelines, memory pressure, and performance telemetry during complex calculations.

---

Performance Diagnostics and Profiling

Analyzing large molecular systems involves non-trivial execution costs split across file I/O, selection parsing, form conversion, and numeric kernel evaluation.

MolSysMT uses SMonitor to provide structured execution profiling:

• Kernel vs. I/O Profiling: Measures the precise time spent reading coordinate blocks from disk versus executing numerical kernels (e.g. distance evaluations or SASA integrations).

• Execution Timelines: Tracks function entry and exit events across nested API calls to pinpoint bottlenecks in custom workflows.

• External Documentation & Governance: SMonitor’s standalone documentation is hosted at https://www.uibcdf.org/smonitor, and its architectural role in MolSysMT is detailed in the Governance module.

---

Controlling Diagnostics and Telemetry

Users can configure SMonitor telemetry and monitor diagnostic events programmatically:

import molsysmt as msm

# 1. Enable progress and decision telemetry events during chunked execution
msm.configure.emit_heavy_telemetry = True

# 2. Set RAM memory pressure warning threshold (80% of max_ram_usage)
msm.configure.memory_pressure_threshold = 0.80

# 3. Running heavy analysis emits diagnostic events and MemoryPressureWarning if pressure is high
center = msm.structure.get_center('large_system.h5msm', selection='all')