Allocation Churn and Garbage-Collection Pressure Reduction
Reduce allocation churn causing frequent collection pauses during an interactive flow
A profiling protocol locating the hot allocation sites, an attribution to code paths, and ranked reduction techniques with the measurement that confirms each.
Role
You are optimising interaction smoothness in {{app_name}}.
Task
Define a profiling protocol that locates the allocation sites driving collection pressure during the flow, attribute them to code paths, and rank reduction techniques with confirming measurements.
Context
{{app_name}} on {{framework_name}} shows {{pause_symptoms}} during {{interaction_flow}} on {{target_device}}. Current allocation profile: {{allocation_profile}}.
Inputs
- {{interaction_flow}}
- {{allocation_profile}}
- {{target_device}}
Constraints
- Record allocations during the flow itself rather than at page idle
- Separate short-lived churn from objects that survive and accumulate
- Attribute allocations to a constructor and call path, not just a total byte count
- Measure on the target device class, not on developer hardware
Output Format
Markdown: profiling protocol, hot allocation attribution table, ranked reduction techniques with confirming measurement, regression guard.
Quality Criteria
- Profiling isolates the flow under test
- Churn and retention are distinguished
- Attribution reaches specific call paths
- A regression guard is defined