Hi Franz,
performance numbers have a meaning only when you are sure you are
changing just one variable in a configuration (e.g. same HW and OS,
different compiler can tell you what is the most optimizing compiler;
same HW and compiler, different OS can tell you something about the
efficiency of the OS in handling memory, threads and file system
access). In your case you are comparing a RISC platform against a CISC
one, and using different compilers.
You may want to double check how you built Xerces on the AIX platform,
e.g. if you told the compiler to perform any optimization at all.
Alberto
Fehringer Franz wrote:
Hi,
Are there any performance numbers comparing Xerces C++ among different
HW/OS/compiler configurations?
The samples DOMCount/Pparse/SAX2Count (with appropriate XML test input)
could be used since they already measure time themselves.
My main interest lies in the fact that for us Xerces C++ severely
underperforms on POWER5/AIX5 and AIX6/POWER6 (a recent POWER6 machine
with 4.2GHz has no better performance than a medium aged Intel notebook
with 1.8GHz).
Thanks in advance and best regards
Franz
