The goal of these runs is to compare the performance of the latest versions of LLVM and GCC on ARM. I used similar options -Ofast for both compilers.

As usually LLVM can not produce a correct code for 254.gap on any optimization level. Therefore -O0 was used instead of -Ofast for both LLVM and GCC compilers for 254.gap.

I did not manage LLVM-3.4 to work with LTO as GOLD have a crash in arm.cc file (GOLD is needed for LLVM if you want to use LTO).

The used processor is Exynos 5410 (1.6GHz Cortex-A15) with 2GB memory under Fedora Core19.

GCC-4.9 was configured with --enable-checking=release --with-arch=armv7-a --with-fpu=vfpv3-d16 --with-float=hard. LLVM-3.4 with CLANG for ARM were built in release mode with--enable-optimized --with-arch=armv7-a --with-fpu=vfpv3-d16 --with-float=hard. The sources are taken from LLVM download site. All changes are given relative to LLVM-3.4 run.

Last modified: 06/23/2014 - vmakarov at redhat dot com

Return to index page.