Fabian Klötzl

C++ has `enum class` to avoid these kinds of problems. But I am looking forward to the prefixed symbols in the next version.

Adding more characters definitely helps to prevent a collision. > Will there ever be a good time to change the existing constants? I fear the answer to that will always...

> The compiler is pretty good at resolving them into normal for loops. That is the issue here. Even with my patch applied the code is compiled into a normal,...

Ooops. Turns out my benchmarking code was wrong. Because I DoNotOptimize(foo) I basically force byte-wise generation. Storing the result in a vector is properly vectorized. ``` ------------------------------------------------------------------- Benchmark Time CPU...

Those are the Linux perf tools. Just `perf record ./command --flags` and it takes a profile. With `perf report` you get an overview of all functions and how much time...

Btw, next time you tag a release please add the `-a` flag. Otherwise git describe uses 3.0.0 as reference. (EDIT marehr: This will be tracked at https://github.com/seqan/product_backlog/issues/159)

Found a bug in [my code](https://github.com/kloetzl/libdna/commit/efb96e4fbe1545e8141ea93c6f126a059e746173). Here are some new numbers using [this patch](https://github.com/seqan/seqan3/issues/1970#issuecomment-654479994). ``` -------------------------------------------------------------------- Benchmark Time CPU Iterations -------------------------------------------------------------------- dnax_revcomp 1060246 ns 1008337 ns 685 bwa_bench 135160 ns...

I recomputed the table above with the latest master on my old laptop. ``` -------------------------------------------------------------------- Benchmark Time CPU Iterations -------------------------------------------------------------------- dnax_revcomp 1022730 ns 1020968 ns 572 bwa_bench 99187 ns 98739...

2022 update, numbers still look the same: ``` -------------------------------------------------------------------- Benchmark Time CPU Iterations -------------------------------------------------------------------- dnax_revcomp 778646 ns 774001 ns 887 bwa_bench 97834 ns 97409 ns 6986 bench/dna4_revcomp 92363 ns 91299...

> But before adding, it would probably be good to have a benchmark included, as well. What was the latest benchmark code you were using, @kloetzl ? Here you go:...