PROMPT: A Fast and Extensible Memory Profiling Framework
summary
A framework that standardizes instrumentation and decouples event generation from analysis, PROMPT implements shared shadow memory, context management, buffering and double‑buffered queues and ports two state‑of‑the‑art profilers while preserving features. The paper reports >65% code reduction, runtime improvements of 5.3× and 7.1× faster versus originals, and a bespoke workflow implemented in 570 lines of code that cuts profiling overhead by over 90% and increases robustness.
AI-written summary · about
chapters (9)
- 0:00PROMPT framework intro
- 2:09Existing profiler problems
- 3:35Memory profiling challenges
- 7:05Profiler generalization
- 8:30PROMPT design and benefits
- 11:19Buffering for throughput
- 12:44Performance improvements
- 14:10Key takeaways
- 14:55Q&A
related talks
19mProfiling Your Code: 5 Tips to Significantly Boost Performance
27mP3: A Profiler Suite for Parallel Applications on the Java Virtual Machine
30mProfiling CPU and Memory on Linux, with Opensource Graphical Tools
26mProfiling Mastery: Decode Your Data and Elevate Observability
31mProfiler: Profiling Code
01h 03mPerformance loop—A practical guide to profiling and benchmarking
16mMemory Allocation Profiling upcoming features
14mWhen Sampling Lies: Trustworthy Performance Profiling for Flat Workloads with Blink