Speeding up the Rust compiler without changing its code

kobzol.github.io

56 points·cpeterso··4 comments

Interesting idea. Not sure how much it worked in practice?

You'll never get order-of-magnitude speedups from things like LTO, but 4% here and 4% there can really add up over time. In a world where most software keeps getting slower, I'm always glad to see people focusing on performance.

The blog describes the technical principes but also how the optimization techniques are difficult to set up in the CI build and largely unfamiliar even to experts.

This shows how "performance focus" has become a niche. Large dev tools like Clang/LLVM set the defauts for performance opt and very few people wander beyond them.

> Large dev tools like Clang/LLVM set the defauts for performance opt

I think of performance the way I think of security. The basics are pretty easy to understand, easy to apply, and are generally applicable across a wide range of software applications and problem domains.

But then you get into stuff like setting kernel parameters to harden your system, and it's much like tinkering with build options to speed up your application. The choices and combinations of options start to become highly dependent on the specifics of your particular system, to require specialized knowledge to understand the implications of any particular change, and to have effects that are not necessarily easy to observe immediately unless you know what to look for and measure.

Reply on news.ycombinator.com