Open mstange opened 3 years ago
I don't think it's a regression. CPU usage being too high is something that have been observed without distributed compilation, even, and some of it is covered in multiple issues filed by froydnj and njn a few months ago.
Ok. I was mostly going by fan noise. I thought in the past my server machine was louder than my Macbook Pro during compilation, and that's no longer the case, but I could definitely be wrong.
Distributed compilation of mozilla-central on macOS causes very high CPU usage in the client's sccache process. I think this is a regression. I think it also slows down the build and underutilizes server resources but I haven't gathered evidence about that yet.
Some cursory profiling with Activity Monitor blames zero-initialization in a
read_to_end
function in tokio.