Here are the partial results for Windows: https://docs.google.com/spreadsheet/pub?key vVqQ9a3_pIudExQek1hOXRyTVVZZ0tMN1kzQjdhNGc&gid=1
Just like on Linux, the new QMutex is much worse than native and than Qt 4.8
for the contended case. Both contention cases are showing 700% worse than
Windows native. On Linux, the 1-mutex case was 100% worse but the 2-mutex case
was 1000%.
I haven't been able to test the new code yet because tst_QMutex crashes
immediately with a failed assertion. I'll need to debug this tomorrow.
Reference:
Intel i5-560M (Westmere)
2 cores x 2 threads, 2.66 GHz, turbo 3.0 GHz
CPU in "performance" mode
Windows 7
MSVC 2010, 64-bit mode
QtCore in release mode, using the standard compilation flags
Results are either the best of 6 runs, or the second best of 9 runs.
PS: I'm quite impressed by the performance of the native solution. The only
thing is that the benchmark is not testing the Win32 functions we're using in
QMutex...
--
Thiago Macieira - thiago.macieira (AT) intel.com
Software Architect - Intel Open Source Technology Center
Intel Sweden AB - Registration Number: 556189-6027
Knarrarnäsgatan 15, 164 40 Kista, Stockholm, Sweden
signature.asc
Description: This is a digitally signed message part.
_______________________________________________ Development mailing list Development@qt-project.org http://lists.qt-project.org/mailman/listinfo/development
