Skip to content

rustc is slow on Windows XP #4303

New issue

Have a question about this project? # for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “#”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? # to your account

Closed
ghost opened this issue Dec 28, 2012 · 5 comments
Closed

rustc is slow on Windows XP #4303

ghost opened this issue Dec 28, 2012 · 5 comments
Labels
I-slow Issue: Problems and improvements with respect to performance of generated code. O-windows Operating system: Windows
Milestone

Comments

@ghost
Copy link

ghost commented Dec 28, 2012

On my 1.7GHz x86 WinXP laptop, rustc 0.5 ( the official binary ) takes 7.2 seconds to compile the "Hello World" from http://www.rust-lang.org/ . Even simply running "rustc -v" takes 5.8 seconds. During these operations, the CPU is 100% utilized and my system performs hardly any disk I/O. I have spare physical memory and no other tasks are hogging the CPU. I'm using a fresh install of the recommended MSYS version, and otherwise don't have any performance issues with the system; in comparison, "g++ -v" takes around a tenth of a second. The results are from several consecutive runs of the programs, and are very predictable.

@ghost
Copy link
Author

ghost commented Dec 28, 2012

On the same machine under Linux, "rustc -v" takes 0.03 sec, and compiling the "Hello World" -- around 1.6 sec.

@ghost
Copy link
Author

ghost commented Dec 31, 2012

This seems to be an issue with MinGW itself. I'm observing a very similar timing difference when running clang compiled with MinGW, and comparing to a version built by Visual Studio. The latter is snappy, whereas the former takes seconds to display the version number.

This is relevant: http://forums.wxwidgets.org/viewtopic.php?f=19&t=35542

@bstrie
Copy link
Contributor

bstrie commented Jan 2, 2013

Related to #1768

@catamorphism
Copy link
Contributor

@h3r2tic 's observation suggests LLVM may be the issue. Probably the first step would be to run with -Z time-passes and figure out which stage is taking up most of the time.

@catamorphism
Copy link
Contributor

This is vague, and the eventual goal is to switch away from MinGW anyway. Closing.

RalfJung pushed a commit to RalfJung/rust that referenced this issue May 5, 2025
add -Zmiri-deterministic-concurrency flag and use it for concurrency tests
# for free to join this conversation on GitHub. Already have an account? # to comment
Labels
I-slow Issue: Problems and improvements with respect to performance of generated code. O-windows Operating system: Windows
Projects
None yet
Development

No branches or pull requests

2 participants