Some performance differences are not due to languages, but to processors
glandium opened this issue · 2 comments
glandium commented
For instance, the C and C++ code are strictly identical, and compile to the exact same machine code. Yet, the C++ version is faster... only because it's not at the same address. See https://news.ycombinator.com/item?id=18092684
This likely explains the observations in #28.
drujensen commented
That is an amazing observation. I've made a note in the readme.
mratsim commented
Another article that goes in-depth into code alignment issues in benchmarking: https://dendibakh.github.io/blog/2018/01/18/Code_alignment_issues