[experiment] Make incremental compilation respect the -Ccodegen-units flag.#67834
Conversation
|
r? @eddyb (rust_highfive has picked a reviewer for you, use r? to override) |
|
The job Click to expand the log.I'm a bot! I can only do what humans tell me to, so if this was not helpful or you have suggestions for improvements, please ping or otherwise contact |
|
|
|
⌛ Trying commit 197594b94d501724c1a361b22bf44fa44af17462 with merge e8ecb07a7235325dce4677b72367d94fae4c732b... |
|
☀️ Try build successful - checks-azure |
|
@rust-timer build e8ecb07a7235325dce4677b72367d94fae4c732b |
|
Queued e8ecb07a7235325dce4677b72367d94fae4c732b with parent 30ddb5a, future comparison URL. |
|
Hm, the only two benchmarks that were affected by the new limit in my local tests (style-servo and script-servo) are broken right now. Also, clap-rs seems to set an explicit codegen-unit count of 4 which now also affects incr. comp. |
|
@bors try @rust-timer queue |
|
Awaiting bors try build completion |
|
🔒 Merge conflict This pull request and the master branch diverged in a way that cannot be automatically merged. Please rebase on top of the latest master branch, and let the reviewer approve again. How do I rebase?Assuming
You may also read Git Rebasing to Resolve Conflicts by Drew Blessing for a short tutorial. Please avoid the "Resolve conflicts" button on GitHub. It uses Sometimes step 4 will complete without asking for resolution. This is usually due to difference between how Error message |
197594b to
c4831e1
Compare
|
@bors try @rust-timer queue |
|
Awaiting bors try build completion |
|
⌛ Trying commit c4831e11621bb57eaa588b2c435a91ca0fed3d5e with merge 6e4bb7983ea0f75fd97fc0f97a654625087439a2... |
|
The job Click to expand the log.I'm a bot! I can only do what humans tell me to, so if this was not helpful or you have suggestions for improvements, please ping or otherwise contact |
|
☀️ Try build successful - checks-azure |
|
Queued 6e4bb7983ea0f75fd97fc0f97a654625087439a2 with parent 9fe05e9, future comparison URL. |
|
The results are "interesting":
I think it makes sense to go forward with this change, but maybe with a slightly higher default for the number of CGUs. I'll do some more testing in order to find a better default. |
c4831e1 to
e4f568e
Compare
|
@bors try @rust-timer queue This uses a higher threshold and we have the webrender-wrench benchmark now. |
|
Awaiting bors try build completion |
|
⌛ Trying commit e4f568e with merge 25dfaf65f2b53636a8235a066d3c8afe5b8f9cc7... |
|
The job Click to expand the log.I'm a bot! I can only do what humans tell me to, so if this was not helpful or you have suggestions for improvements, please ping or otherwise contact |
|
☀️ Try build successful - checks-azure |
|
Queued 25dfaf65f2b53636a8235a066d3c8afe5b8f9cc7 with parent 2f688ac, future comparison URL. |
|
This is properly implemented in #70156. |
Some local testing I did last year showed that limiting the number of CGUs for incremental compilation can improve performance quite a bit for big crates. Let's look into that some more.