One of the areas I’ve been meaning to run more benchmarks on this season has been for the recently released Clang 21 compiler. Back in September when LLVM Clang 21 was debuting I ran some initial benchmarks and found it to deliver some nice performance gains on AMD EPYC Zen 5 but then have been busy with other benchmarks/articles for expanding that testing. Recently with having some spare cycles and gratis access still to the Microsoft Azure HBv5 instance for AMD EPYC Zen 4 with HBM3, I ran some Clang 20 vs. Clang 21 performance benchmarks there for those wondering about any performance benefits of this new compiler release on Zen 4.
Azure’s HBv5 instances offer incredible performance with being powered by the AMD EPYC 9V64H processor that is a Zen 4 design but paired with HBM3 high bandwidth memory. The HBM3 memory provides a powerful combination with very significant generational performance gains for HPC and other workloads as shown in prior benchmarks. With having access still to the VM, I ran some Clang 20.1 vs. Clang 21.1 comparison benchmarks to see how this latest LLVM/Clang C/C++ compiler is performing there to complement the prior Zen 5 compiler benchmarks.
The Clang 20 and Clang 21 builds were obtained using the LLVM.org APT repository. All the same CFLAGS/CXXFLAGS were kept the same during testing along with the other software settings with just swapping out the used compiler between rebuilding the benchmarks and running.
This Friday article is mainly intended for reference purposes for those wondering about any AMD Zen 4 benefits on Clang 21.
