| Popis: |
When running an analysis application distributedly, a crucial feature is the ability to monitor, debug and profile the execution of the application, a task that is rendered more complex in the presence of JIT-ed code. In the following report an overview of the power of Cling new JIT-ed profiling feature is given, underlining its strength and weaknesses. Afterwards, this important feature is exploited to devise and implement a profiling strategy for RDataFrame applications with JIT-ed code in the distributed case. Finally some tests are performed and the result compared to the local profiling. |
