Add inline tags to commonly used functions

[joshuataylor]

I've gone through and added inline to some commonly used functions.

If everyone is happy, we could also add these to others.

We should judge what should be inlined. See https://nnethercote.github.io/perf-book/inlining.html

None. The compiler will decide itself if the function should be inlined. This will depend on the optimization level, the size of the function, etc. If you are not using link-time optimization, functions will never be inlined across crates. #[inline]. This suggests that the function should be inlined, including across crate boundaries.

Amazingly I saw great wins with just this, as I have to call some of these functions a lot when encoding a lot of data.

This is using lto: before

Operating System: macOS
CPU Information: Apple M1
Number of Available Cores: 8
Available memory: 16 GB
Elixir 1.13.4
Erlang 25.0.2

Benchmarking encode column ...

Name                    ips        average  deviation         median         99th %
encode columns         13.86       72.17 ms    ±21.40%       74.47 ms      114.95 ms

after:

Name                    ips        average  deviation         median         99th %
encode columns         15.12       66.13 ms    ±23.48%       68.36 ms      108.11 ms

[evnu]

@joshuataylor thanks for the PR. Can you share the code for your benchmark?

[hansihe]

This looks like a very good idea to me.

[evnu]

Interesting. I would not expect inline to make a real difference. Can we explain why we see a difference in the performance?