SPEC benchmarks depend on undocumented C behaviour: http://t.co/9zkybqwM9N -- compiler benchmarking takes real language lawyering.
miniblog.
Related Posts
Benchmarking the cost of array bounds checks in Rust code: https://blog.readyset.io/bounds-checks/
The author even patches rustc to make sure *all* bounds checks are gone, in order to compare the performance!
Cute benchmarking technique for choosing sample programs in a benchmarking suite: perform principal component analysis to avoid redundancy!
I really like instruction counts for measuring performance. It's not perfect, but it's a very stable metric, even if I'm listening to music whilst developing and benchmarking.
