HEP applications perform large amounts of memory allocations/deallocations within short time intervals, which can result in memory churn, poor locality and performance degradation. These can now be identified and examined with the recently extended FOM-tools, which through stored stack traces, pinpoints source code responsible for short-lived objects and distinctive lifetime patterns. The presentation will show and discuss the most important results that have been obtained from ATLAS jobs.
(CERN), Sami Kama
(Southern Methodist University (US)), Sami Kama
(Southern Methodist University (US))
Today, memory become again a precious resource for performance and for the memory consumption. MALT is a malloc profiling tool mapping your memory allocation onto your source code. Hence you can easily search for bad memory usage, memory leaks, large memory consumption and short live memory allocations.