Speaker
Mr
Sverre Jarp
(CERN)
Description
HEP programs commonly have very flat execution profiles, implying that the execution
ime is spread over many routines/methods. Consequently, compiler optimization should
be applied to the whole program and not just a few inner loops. In this talk I,
nevertheless, discuss the value of extracting some of the most solicited routines
(relatively speaking) and using them to gauge overall performance and potentially
“critique” the code generated by the compiler(s). An initial set of ten C++ routines
have been extracted from three HEP packages (ROOT, GEANT4 and CLHEP). One key
advantage of this approach is that the selected routines compile and execute in
seconds, allowing lots of testing of different platforms, compilers and compiler
options. The speaker will review the initial selection and show results with GNU gcc
and Intel icc on multiple hardware platforms.
Primary author
Mr
Sverre Jarp
(CERN)
