Architecture & Systems
Performance Profiling and Bottlenecks
Profiling identifies where time is actually spent so optimisation efforts focus on real bottlenecks, not guesses.
Mental model
Measure where computation is spent before changing code to make it faster.
Profiling tools and timing metrics reveal hotspots in pricing and risk runs, such as repeated interpolations or slow loops.
The biggest gains often come from vectorisation, caching reusable artefacts, or batching operations across trades.
Explore the mechanism
Where does the program actually spend time and resources under a representative workload?
- Profiling tells you where time is actually spent.
- Big wins usually come from caching, batching, and removing repeated work.
- Numerical behaviour must remain controlled under regression tests.
Does speeding up one component improve the whole run?
What to notice
Any optimisation must be validated against regression tests to ensure numerical behaviour remains within acceptable tolerances.
Where it appears
- finding slow pricing paths
- measuring database and network waits
- prioritizing optimization work
Where the intuition breaks
What this view leaves out
- Profilers add overhead and can shift behavior; a non-representative workload produces precise measurements of the wrong bottleneck.
Where this concept sits
Follow the strongest prerequisites, applications, extensions, and contrasts.
Connections for Performance Profiling and Bottlenecks
Profiling identifies code paths and infrastructure bottlenecks that threaten latency budgets or timely risk controls.
SupportsScheduled Jobs and Worker QueuesProfiling helps identify throughput and queue bottlenecks in batch and worker-based pipelines.
Profiling identifies code paths and infrastructure bottlenecks that threaten latency budgets or timely risk controls.
Explore concept →