What kind of speed improvements did anyone who’s moved from an x86 Mac to an M1 Mac see in the time it takes them to run a debug build of their Xojo projects? I’m picking running a debug build as my benchmark because that’s where I spend the most time watching progress bars in my Xojo work.
I’ve just been running some tests, comparing the current version of Xojo on my current Mac (2017 iMac, 4.2GHz quad core i7, 32GB RAM) vs an M1 MacBook Air. It’s taking ~55 seconds to run a debug build with default optimisation on the M1 Mac, compared to ~75 seconds on my Intel iMac.
Is that ~25% speed improvement roughly what anyone else has been seeing?
FWIW, I looked back at an old spreadsheet of speed tests that I did when I bought my iMac back in 2017. Back then, it took ~22 seconds to launch a debug build on the then-current version of Xojo, although I think that was just before I switched to making 64-bit builds on macOS.
is it 1m15s or 115s ?
for me 55s to 115s is x2 increase ?
also hope you have an ssd drive in your iMac ?
Missing colon in 1:15
I’ve changed my original post to read 75 seconds, thanks for pointing out the silly mistake.
And, yes, SSD in the iMac.
My compare is not fair as in one case (Xojo 2015r1 / MBP 2014 i5) is faster than the later (Xojo 2020r2.1/ M1 MBP).
So in that case, and that project, the former runs in around 2seconds, the later 20seconds (or more), in the IDE.
The data folder that is scanned is in an external SSD (former) boot (faster) SSD (later).
The data folder holds 480 folders with 40 to 70 files (depends on folder, sometimes a bit less) and read the files names (and 1 file size in each folder when that specific file exists; an archive file).
BTW: the code returns a report of what it found in a TextArea as html.
I just got my new 16" MacBook Pro with the M1 Max and 64GB of memory. Just to satisfy my curiosity, tonight I did a rough speed test with Xojo.
The old MBP (2019) has the 2.3 GHz 8-core Intel Core i9, 32GB of RAM, and 4TB of SSD storage.
The new MBP (2021) has the M1 Max processor, 64GB of RAM, and 8TB of SSD storage.
Both are running the same current version of Xojo and both have the same set of plugins (a mix of MBS, Einhugur, and Xojo). The test project was Lightwright (which is very large), set for a Universal (Intel & ARM) build.
For a test, I separately timed how long it took to launch Xojo, then I timed how long Xojo took to load Lightwright, and finally how long it took for Xojo to go from hitting the Run button to “finish”.
Note - When I hit run on the old MBP, Catalina asks if I want Lightwright to have access to the Documents folder. This doesn’t happen on the new MBP, which is running Monterey. So on the old MBP my “finish” time is to the Documents folder dialog appearing, and on the new MBP “finish” is when Lightwright launches. Note: By the time my finger on the stopwatch hits “stop”, Lightwright will have already accessed the Documents folder.
The old MBP took 223.82 seconds, the new MBP took 149.46 seconds.
No doubt there are things I didn’t consider in this quick test, but it was a fun to do and definitely interesting! Next will be doing some video work, I’m hoping that has even better speed increases.
I put on Twitter that I did a test build that felt noticeably faster in the M1 Max compared to my 2017 MBP. No measurement, but it sure felt faster.
I got my MacBook Pro, too.
The most impressive thing is silence. The Intel i9 MacBook Pro mostly runs fans, I can hear. Even when sitting here on idle with 5K display attached and doing time machine or so.
The MacBook Pro with Apple Silicon can do compiling with Xcode and doesn’t even turn on the fan!
I love that.
Thanks, John: glad to hear that you’re getting a ~33% debug build speed increase, compared to the ~25% speed increase I saw with the M1 MBA that I tested.
I wish I had a careful record of things that reduced the speed of my Xojo debug build speeds between 2017 and now, though: the same iMac has gone from ~22 seconds to ~75 seconds in that time, which is much bigger than any improvement I can hope for with new hardware.
the precompiling on M1 Max is about half the time compared to the i9.
Agreed. I’ve been playing with mine all weekend and I haven’t heard its fans once. There is so much to like about this machine.
I’ll chime in here too. I have the new 10-core 16" M1 Pro and it is (a) a bit faster than the M1 MacBook Pro it’s replacing (more cores for compiling) but (b) way faster than the £5K iMac Pro that replaced.
Everything about the new MacBook Pro is nicer than its predecessor. Things like the display and the utter silence are joyful.
I hate you all …
[writing this on a 2010 Mac Pro ]
Prices for Intel Macs may soon drop significant and provide a buying opportunity for you to get a newer one.
I may be wrong, but there is no more New Intel Mac on sale right now. Only second hands Intel Mac computers.
Yes, and those used once got cheaper
I also now have my 16" MBP with M1 Max, and I have yet to hear the fans kick on. I don’t notice a significant performance improvement compared to my M1 mac mini, but either of them blows away all my Intel based macs. Other than VMs.
My Geekbench 5 scores for single core are nearly identical between my M1 mini and my MBP with M1 Max (which was expected). The multi-core score is 65.8% higher but for my workload that does not yet translate to a significant difference. Enabling high power mode also did not have much effect on my Geekbench 5 score.
From strictly a price / performance view for my workload, the M1 mini probably can’t be beat. But the new 16" MBP is hands down the nicest laptop I have ever used. The notch seems like a non-issue. The screen looks utterly fantastic, and the keyboard feels very nice and we are back to full size function keys, which I personally like.
Intel Mini, 27" iMac, Mac Pro.
Thinking of ditching my late 2019 MacBook Pro 15” Touchbar and just getting the M1 Air.
It’s just the virtualisation of Windows I need to sort out - I do have some awful Windows laptops I could leverage when the need arises.
Yes, especially mine … … now I hate you all even more …
Unless you really need the trade value, keep the 2019. Use that for your virtualization platform. Windows testing is more complicated than just “a version of Windows” and you’ll probably want to maintain Windows 7, Windows 10, and Windows 11 VMs.