Evaluate using Profile-Guided Optimization (PGO)

[zamazan4ik]

Hi!

Writing this for the history. Maybe these results will be interesting to someone who trying to achieve better performance with lingua-rs since the project cares about performance.

I test Profile-Guided Optimization (PGO) on different kinds of software - the current results are available here(with a lot of other PGO-related information). That's why I tried to optimize lingua-rs with PGO too.

Test environment

I performed tests on two configurations: Linux and macOS.

Linux:

• Fedora 39
• Linux kernel 6.5.11
• AMD Ryzen 9 5900x
• 48 Gib RAM
• SSD Samsung 980 Pro 2 Tib
• Compiler - Rustc 1.73
• Lingua-rs version: the latest for now from the main branch on commit 5a1f4980162996319474552e319b211035fdd670
• Disabled Turbo boost

Benchmarks

As a benchmark, I use built-in benchmarks with cargo bench --all-features. For PGO optimization phase, I use cargo-pgo with cargo pgo optimize bench -- --all-features. For PGO training phase, I use the same benchmark with cargo pgo bench -- --all-features.

Results

I got the following results:

• PGO-optimized compared to Release: https://gist.github.com/zamazan4ik/fdbb9f8b2a1e16d6bbd162a01ea0f6b5
• (just for reference) PGO-instrumented compared to Release: https://gist.github.com/zamazan4ik/14973a900c780902462a29cccbe18ae8

As you see, in several cases lingua-rs performance can be improved with PGO. I think this information can be written somewhere into the documentation, so users will be aware about PGO effects on Lingua-rs performance and can decide to apply PGO for their Lingua-rs builds.

[pemistahl]

Hi, sorry for the late response and thanks for this valuable information. I'm sure there is still room for performance improvements. As soon as I want to work on performance, I will look at your results more closely.