GCC vs. LLVM Clang Compiler Performance On The AMD Threadripper 2990WX Benchmarks

Written by Michael Larabel in Software on 3 September 2018 at 03:38 PM EDT. Page 1 of 4. 24 Comments.

With many of our other initially planned Threadripper 2 Linux benchmarks out of the way, recently I carried out a compiler benchmarking comparison on the Zen+ Ryzen Threadripper 2990WX between GCC and Clang.

For those curious about the performance of code generated by GCC versus the LLVM Clang C/C++ compilers on AMD Zen+ hardware, I ran some tests using GCC 7.3.0, GCC 8.2.0, a GCC 9.0.0 development snapshot, LLVM Clang 6.0.1, and LLVM Clang 8.0 SVN. Tests of Clang 7.0 weren't bothered since it was fairly recently that LLVM Clang 7.0 was forked from trunk and not many changes between 7.0 and 8.0 in SVN for right now.

When testing of the five compilers, the CFLAGS/CXXFLAGS were maintained the same as "-O3 -march=znver1" throughout the entire testing process. This open-source compiler benchmarking was done on the AMD Ryzen Threadripper 2990WX 32-core / 64-thread system running Ubuntu 18.04 LTS with the Linux 4.19 kernel.

GCC + LLVM Clang Benchmarks Threadripper 2990WX

This GCC vs. Clang compiler benchmarking was carried out using the open-source and fully-automated Phoronix Test Suite benchmarking software.

Related Articles