GCC 8 & LLVM Clang 6.0 Compiler Performance On AMD EPYC - November 2017

Given the continuously evolving state of open-source code compilers, especially for the newer AMD Zen "znver1" architecture, here is the latest installment of our compiler benchmarks. Tested for this article from and AMD EPYC 7601 processor were GCC 7.2, GCC 8.0.0, LLVM Clang 5.0, and LLVM Clang 6.0 SVN.

Tests were done from the TYAN Transport SX TN70A-B8026 with AMD EPYC 7601 (32-core / 64-thread) processor and 8 x 16GB DDR4-2666 memory. Running on this system was Ubuntu 17.10 with the Linux 4.13 kernel to provide an updated Linux stack for best support.

Via the Phoronix Test Suite a range of benchmarks were run when testing GCC 7.2.0, GCC 8.0.0 20171030, LLVM Clang 5.0, and LLVM Clang 6.0 SVN. All tests were built under each compiler while the system CFLAGS/CXXFLAGS were set to "-O3 -march=znver1" and each compiler was built in its release/optimized mode. Clang 6.0 and GCC 8 will be released as stable in early 2018.