dragonegg vs xplor-nih

Project Infrastructure LLVM Dev List Archives
1

I was quite surprised to find that dragonegg svn can now compile all of
xplor-nih (which is a complex mix of c, c++ and fortran that is a regression
magnet for FSF gcc). The xplor-nih package was compiled at -O3 -ffast-math -funroll-loops
for all three compilers. The xplor testsuite passed without regressions and benchmarked
as follows...

dragonegg svn with llvm 2.9 and FSF gcc 4.5.3svn
Total CPU time: 46.0750

FSF gcc 4.5.3svn
Total CPU time: 43.2884

Pretty impressive. It will be very interesting to see how dragonegg performs once
-fplugin-arg-dragonegg-enable-gcc-optzns become usable at -ffast-math -O3.

2

Hi Jack,

Pretty impressive. It will be very interesting to see how dragonegg performs once
-fplugin-arg-dragonegg-enable-gcc-optzns become usable at -ffast-math -O3.

I originally intended -fplugin-arg-dragonegg-enable-gcc-optzns as an aid for
analysing performance differences between gcc and dragonegg. But it sounds like
you systematically want to run LLVM optimizations after all GCC optimizations,
presumably to get the fastest possible executables. If this gives good results
I will add -fplugin-arg-dragonegg-enable-gcc-optzns to my list of features that
need to be 100% robust.

Ciao, Duncan.