# Profiling build performance?
**URL:** https://discourse.cmake.org/t/profiling-build-performance/2443
**Category:** Usage
**Created:** [December 29, 2020, 8:33pm UTC](https://discourse.cmake.org/t/profiling-build-performance/2443 "2020-12-29T20:33:23Z")
**Posts on this page:** 3
**Page:** 1
### Author:  [@YMba9g8j9CJp0wLoQf5y](https://discourse.cmake.org/u/YMba9g8j9CJp0wLoQf5y)
#### Post date: [December 29, 2020, 8:33pm UTC](https://discourse.cmake.org/t/profiling-build-performance/2443/1 "2020-12-29T20:33:23Z")
I’d like to profile the performance of my cmake ninja build. To see how I can improve my build/linker speed.
Ideally I’d like to be able to distinguish compile vs linker time. Because my project is being bottlenecked at the linker phase.
Thoughts?
EDIT:
(I’m using MSVC as my compiler)
---
### Author:  [@McMartin](https://discourse.cmake.org/u/McMartin)
#### Post date: [December 31, 2020, 10:24am UTC](https://discourse.cmake.org/t/profiling-build-performance/2443/2 "2020-12-31T10:24:45Z")
Ninja writes a `.ninja_log` file where each build step is recorded. You can then convert that log file using [https://github.com/nico/ninjatracing](https://github.com/nico/ninjatracing) to a format that can be read by Chrome’s `about:tracing` utility.
I hope this helps.
---
### Author:  [@YMba9g8j9CJp0wLoQf5y](https://discourse.cmake.org/u/YMba9g8j9CJp0wLoQf5y)
#### Post date: [December 31, 2020, 6:17pm UTC](https://discourse.cmake.org/t/profiling-build-performance/2443/3 "2020-12-31T18:17:52Z")
Cool I didn’t know that Ninja did that!
I guess it makes sense this type of thing is dependent on the generator you use.
Seems like a lot of effort to make that sort of thing cross platform and user friendly in cmake.