AMD/Intel Benchmarks: Building The Mainline Linux x86_64 Kernel With LLVM Clang

Written by Michael Larabel in Software on 12 September 2019. Page 2 of 4. 23 Comments

First up were a few I/O tests between the same kernel built with GCC and then Clang. Again, all of the programs/workloads under test were built with the same (GCC) compiler.

Under the SQLite insertion test, the Clang built kernel seemed to yield faintly better performance. There is differing storage between the two systems among other factors to keep in mind; the purpose of the comparison isn't for the 3900X vs. 9900K performance comparison but rather a look at the Clang kernel performance on two distinct systems.

The FIO performance meanwhile improved for the GCC kernel build on the AMD Zen 2 system.

The FS-Mark and CompileBench tests meanwhile were mixed on the AMD side while the Intel system was seeing about the same performance.

The AMD Ryzen system did se its performance affected in more benchmarks with the Clang-built kernel than the Core i9 9900K system. With the RAMspeed memory benchmark using the GCC kernel build did provide noticeably better performance.

While in the t-test1 memory allocation tests, the Clang built kernel on the Ryzen 9 3900X system did see a slight performance advantage.

Related Articles