A5044232284- Software System Performance and Reliability
- Parallel Computing and Optimization Techniques
- Software Engineering Research
- Cloud Computing and Resource Management
- Software Testing and Debugging Techniques
We investigate the duet measurement procedure, which helps improve accuracy of performance comparison experiments conducted on shared machines by executing measured artifacts in parallel and evaluating their relative together, rather than individually. Specifically, we analyze behavior procedure multiple cloud environments use experimental evidence to answer research questions concerning assumption underlying procedure. demonstrate improvements ranging from 2.3x 12.5x (5.03x average) for...
Systematic testing of software performance during development is a persistent challenge, made increasingly important by the magnifying effect mass deployment on any savings. In practice, such systematic evaluation requires combination an efficient and reliable measurement procedure integrated into environment, coupled with automated results compact reporting detected anomalies.
For the past three years, we have used several public Java benchmarks (DaCapo, ScalaBench, Renaissance) to track performance changes introduced by daily development of Graal compiler. The talk will outline how tackle common measurement issues such as scheduling and change detection, summarize observed parameters themselves, then discuss factors that impact usefulness testing for compiler process, example (1) what are useful report (or not), (2) missed benchmarks, or (3) aspects behavior make...