![]() ![]() That’s quite significant but is in line with what you might expect from dropping two performance cores. There was a noticeable difference between the M1 Pro 8-core and the M1 Pro 10-core machine, with the 10-core being around 15% faster when building. It’s also worth noting that we saw no difference between an M1 Max with 32Gb and one with 64Gb, difference of 0.2 seconds on builds and 0.6 seconds on tests is well within the error margins. The GPU is not used for compilation or in these tests, so it makes perfect sense, but it’s worth mentioning. You may have noticed no difference between the benchmarks when switching between the M1 Pro and the M1 Max. M1 Pro vs M1 Max, 32Gb vs 64Gb, and 8-core vs 10-core ![]() Just less than 27 seconds on the M1 MacBook Air and just less than 26 seconds on the Mac Book Pro. The M1 Max’s extra cores don’t help much, and the best time between the regular M1 and the Pro/Max is inconsequential. The results are much less dramatic when running the tests. It’s been a long time since we’ve seen 2x improvements in a machine of the same class, but the best Intel build time is 60 seconds versus the above mentioned 31 seconds of the M1 Max. If you’re considering upgrading from an Intel machine to an M1 Pro or Max, you’ll see a dramatic reduction in build times. Those extra cores matter, and the ~35% improvement is in line with what you’d expect, going from a 4+4 performance/efficiency core setup to an 8+2 configuration. They improve on the M1 Air’s best result of 47 seconds with a build time of less than 31 seconds. Overall, it’s remarkable that the M1 MacBook Air already had the best performance before Apple introduced the new MacBook Pros, but the M1 Pro and Max chips take this further. A clean build will typically utilise all available cores, where this project’s tests use a dockerised database running on the same machine and are not heavily multi-threaded.įirst, we know what you all want to see. These tasks might seem similar, but they stress the machine in quite different ways. We are testing two performance aspects: building the project (clean build) and running the tests. The Swift Package Index is a sizeable open source project and makes for a good test when assessing the new chips, so we’ve created a benchmark tool to give some real-world results. You’ll have seen benchmarks appearing on various sites over the last week, but how do these machines stack up with a real-world Xcode project? ![]() First observed some 200 years ago by Jane Austen, this still holds today as lucky developers take delivery of new M1 machines. It is a truth universally acknowledged that a developer in possession of a good project must be in want of a fast compile. M1 Pro and M1 Max Xcode Build and Test Benchmarks Posted Monday, November 1, 2021. ![]()
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |