The goal of these runs is to compare the perfomance of the latest
versions of LLVM and GCC on x86_64. Most widely used common
options -O2 and -O3 is taken for all
runs.
The used machine is 3.0Ghz Intel Core2 with 4GB memory
under Fedora Core8. The compilers are configured with
--enable-checking=release.
All changes are given relative to LLVM-2.3.
-
Compilation Speed
-
Code Size
-
Base SPECInt2000
-
Peak SPECInt2000
-
Generated Code Performace
-
Base SPECInt2000
-
Peak SPECInt2000
We used -O0 for 176.gcc compilation by llvm-2.3 because
176.gcc crashes LLVM compiler in -O2 and -O3 mode.
Last modified: 06/13/2008 - vmakarov at redhat dot com
Return to index page.