Enhance interface with LLVM optimizations #2396
Labels
A-LLVM
Area: Code generation parts specific to LLVM. Both correctness bugs and optimization-related issues.
C-enhancement
Category: An issue proposing an enhancement or a PR with one.
Comments
ghost
assigned 
|
I think this should be done, but it seems like a "far future" thing. |
bors
added a commit
that referenced
this issue
May 29, 2013
Refactor the optimization passes to explicitly use the passes. This commit just re-implements the same passes as were already being run. It also adds an option (behind `-Z`) to run the LLVM lint pass on the unoptimized IR. This should close #2812. It is also the first step towards #2396 ---------------- This is pretty much just an initial "get it out there" PR. With finer control over the optimization passes coming later. I also just blindly copied the passes we were already doing, so there could almost certainly be some more work in paring it down. The other thing is the addition of the `mergefunctions` pass, which is currently enabled at `--opt-level=3` and does have a small impact on the code size. However the fact that it is at the end of the optimization pipeline is probably not ideal, so some more experimentation is in order.
bors
added a commit
that referenced
this issue
May 29, 2013
This refactors pass handling to use the argument names, so it can be used in a similar manner to `opt`. This may be slightly less efficient than the previous version, but it is much easier to maintain. It also adds in the ability to specify a custom pipeline on the command line, this overrides the normal passes, however. This should completely close #2396.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Three related FIXMEs in
back::linksuggest adding back-end optimization levels other than-O2, exposing and using lists of specific LLVM optimizations, or at least not duplicating the LLVM optimizer's list of-O2optimizations.The text was updated successfully, but these errors were encountered: