These are the results from running Xcode on a non-trivial open source project using various Macs. The goal is to give developers a relative idea of how each computer model compares to one another. Read the specifics and contributing sections for more info.
๐ฅ | Computer Model | CPU | RAM | Fresh Build Time | Incremental Build Time | Date & Commit Hash | New Build System? |
---|---|---|---|---|---|---|---|
โจ๏ธ | Custom PC Early 2018 |
i7-8700K 4.70+ GHz | 32 GB | 0:36 | 0:07 | 2018-02-19 (commit) | โ |
โจ๏ธ | Custom PC Late 2015 |
i7-4790k 4.0 GHz | 32 GB | 0:40 | 0:07 | 2018-02-22 (commit) | โ |
๐ฅ | iMac Pro 10 Core, Late 2017 |
3 GHz Intel Xeon W | 64 GB | 0:41 | 0:07 | 2017-12-28 | โ |
๐ฅ | iMac 27" 5K 1TB SSD, Mid 2017 |
4,2 GHz i7 | 40 GB | 0:49 | 0:08 | 2018-01-18 | โ |
๐ป | MacBook Pro Retina, 15", 2017 |
2.9 GHz i7 | 16 GB | 0:49 | 0:16 | 2017-10-09 | โ๏ธ |
๐ป | MacBook Pro Retina, 15", 2017 |
2.9 GHz i7 | 16 GB | 0:50 | 0:15 | 2017-10-09 | โ |
๐ป | MacBook Pro Retina, 15", Late 2013 |
2.6 GHz i7 | 16 GB | 1:09 | 0:12 | 2018-02-28 | โ๏ธ |
๐ป | MacBook Pro Retina, 15", 2015 |
2.8 GHz i7 | 16 GB | 1:17 | 0:12 | 2018-01-02 | โ |
๐ป | MacBook Pro Retina, 13", 2014 |
2.6 GHz i5 | 8 GB | 2:30 | 0:23 | 2018-01-04 | โ |
Mac Mini Mid 2012, 512 SSD |
2.3 GHz Quad-Core i7 | 16 GB | 1:32 | 0:18 | 2017-10-20 | โ | |
๐ป | MacBook Pro Retina, 15", 2017 |
2.8 GHz i7 | 16 GB | 1:50 | 0:14 | โ | |
๐ป | MacBook Retina, 15", Mid 2012 |
2.6 GHz i7 | 8 GB | 2:26 | 0:23 | โ | |
๐ป | MacBook Pro 13", Mid 2012 |
2.9 GHz i7 | 8 GB | 2:30 | 0:23 | โ |
For the test, I decided to use an app that I actually work on: eidolon. Unlike most iOS apps, the project is stable and not under active development, so it's a prime candidate for accurate-ish benchmarking.
For "fresh" builds, I cleaned the build folder (โโงK) repeatedly until it worked with no permissions problems. Then I sat and waited for Xcode to index the project. I also made sure the simulator was closed, so these times include booting the simulator and launching the app. Then I hit โR and start a timer, only ending it when the app had fully launched.
"Incremental" builds represent a more common use case: changing one file and recompiling with the simulator already running. I added print("hello!")
to application(: didFinishLaunchingWithOptions:)
and hit โR, timing the time it took for the app to launch.
I repeated each test a few times and took their average times.
It would be super-cool if we could perform the above tests on a variety of machines and consolidate the results here. You can follow the instructions to download the code and the project dependencies, and send a pull request adding your own results. I'd super-appreciate it! ๐
Please note that this project is released with a Contributor Code of Conduct. By participating in this project, you agree to abide by its terms.
This work is licensed under a Creative Commons Attribution 4.0 International License.