Recently I checked Profile-Guided Optimization (PGO) improvements on multiple projects. The results are here. E.g. PGO results for LLVM-related tooling are here. According to the tests, PGO usually helps with the compiler and compiler-like workloads (like static analysis) - e.g. Clang gets +20% compilation speed with PGO. That's why I think trying to optimize the Rust-Code-Analysis tool with PGO can be a good idea.
I can suggest the following action points:
Perform PGO benchmarks on Rust-Code-Analysis. And if it shows improvements - add a note about possible improvements in Rust-Code-Analysis's tool performance with PGO.
Providing an easier way (e.g. a build option) to build scripts with PGO can be helpful for the end-users and maintainers since they will be able to optimize the Rust-Code-Analysis tool according to their own workloads.
Optimize pre-built binaries
Testing Post-Link Optimization techniques (like LLVM BOLT) would be interesting too (Clang and Rustc already use BOLT as an addition to PGO) but I recommend starting from the usual PGO.
I recommend trying to use PGO with cargo-pgo for the Rust code.
Here are some examples of how PGO optimization is integrated in other projects:
marco-c
commented
8 months ago
Hi!
Recently I checked Profile-Guided Optimization (PGO) improvements on multiple projects. The results are here. E.g. PGO results for LLVM-related tooling are here. According to the tests, PGO usually helps with the compiler and compiler-like workloads (like static analysis) - e.g. Clang gets +20% compilation speed with PGO. That's why I think trying to optimize the Rust-Code-Analysis tool with PGO can be a good idea.
I can suggest the following action points:
Testing Post-Link Optimization techniques (like LLVM BOLT) would be interesting too (Clang and Rustc already use BOLT as an addition to PGO) but I recommend starting from the usual PGO.
I recommend trying to use PGO with cargo-pgo for the Rust code.
Here are some examples of how PGO optimization is integrated in other projects:
configurescript