I honestly don’t think throttling will be a big issue. From people’s testing so far (large video/audio renders, and gaming) the penalty appears to be quite small. These things are so fast that a hit of 15-20% will still leave you with a huge amount of processing power.
how is the cpu usage of the native arm64 version compared to rosetta 2 + intel in the case of vcvrack for some equal patch?
True heroes. The depth of courage, ingenuity and tenacity in the face of such overwhelming odds shows a wellspring of pure, indomitable will that is frankly staggering.
Along with the tireless hospital staffs in the trenches battling wave after pandemic wave, the VCV Rack dev community exemplifies that core of human excellence that gives us the inspiration, the hope to soldier on.
Will examine this weekend
Congrats, that is very exciting! Could you share the patches or zipped source tree (with all modified deps), and a precompiled binary, for early testers? (or upload to github)
I am about to prepare a public version which will be available later today… stay tuned!
For all early birds - here is a prerelease of Rack.arm64
No documentation etc. bundled with a couple of modules. Tomorrow I will release a better tested version with some additional notes.
+++ USE ON YOUR OWN RISC +++
Hm… what system you’re on? Sorry, was late yesterday… there will be a working version later
That is really really strange! Exact the same file will stop working after it’s downloaded from the internet… Currently I can’t explain myself what’s happen here. I tried .zip .tgz and .img - always the same. After downloaded from GitHub none of them could be run…
Well it seems that Apple protect unsigned Applications from running after downloaded from the internet. Thats really bad and I am kinda angry right now… I will search for an easy workaround, stay tuned!
Open Terminal App and paste the following lines:
cd ~/Download xattr -dr com.apple.quarantine ./Rack.app
Now you should run Rack without any issue.
Hey Lars, any news on the M1 Macbook Air with VCV? I’m very curious to see how it performs!
It wont, just buy a PC! All the fans and coolings you cant find on apple products
Hi there, please don’t dilute this thread, which is specifically about Apple Silicon M1. And preferably a good place to find information on how the M1 and possible consecutive models will perform with VCV Rack.
There are plenty of other places to debate on PC vs Mac. Which frankly is a very tiring debate on its own.
Have fun making music!
Ok, you kicked me, testing right now…
Ok, I’ve done some testing on the M1 Macbook Air, and these are the results.
Settings and conditions for all tests, unless noted otherwise:
- Good headphones directly into the headphone out of the computer.
- Audio module connected to standard, internal audio interface.
- 44.1 Khz, blocksize 1024.
- 1 engine thread used
- Native (Retina) screen resolution with 60FPS
- Rack window zoomed to whole screen
- Screen is on the whole time, powercable is connected.
- Used CPU is given for the whole machine as such.
- Temperature readings are given as two readings: 1) By touching the hottest strip of the cabinet, just above the function key row. 2) By touching the cabinet next to the trackpad.
- When nothing is remarked it means everything works, plays and sounds as it should, with no audio stuttering at all.
- Each patch is run for 15 minutes straight or more. Any thermal throttling will happen before this.
- It’s the normal VCV Rack for Mac downloaded and installed, which means an x86/Intel binary running under Rosetta2 on the M1 Mac.
- Unless otherwise noted, scrolling the Rack is buttery smooth, with no lagging whatsoever. Very graphically satisfying.
- Also, just as a reminder, the M1 Macbook Air has no fan and so makes no sound at all!
- This M1 Macbook Air is the base model: 8GB RAM and 256GB SSD. In other words: The smallest M1 machine that Apple makes
- All test patches are fully wired up and playing actual “music”. This to maximize the chance of getting actual, real-use results.
Before Rack is opened:
- Used CPU: 2%
- Rack default patch
- Used CPU: 11%
- Temperature: Skin temperature / Cold
Test 2: test-02.vcv (8.1 KB)
- The Test-1 patch with Midi module replaced with Bogaudio llfo, addr-seq and VCV Qnt:
- Used CPU: 11%
- Temperature: Slightly above skin temperature / Slightly below skin temperature
Test 3: test-03.vcv (20.4 KB)
- The Test-2 patch built out with another row of Plaits, Clouds, 2xRings plus Plateau reverb etc. So a modest and quite musically relevant 2-row patch with 4 voices.
- Used CPU: 12%
- Temperature: Slightly warm / Skin temperature
Test 4: test-04.vcv (76.7 KB)
- The Test-3 patch built out with big mixer+expander, GUI effects, and 3 instances of Dexter. A healthy 6-row patch with 7 voices.
- Used CPU: 15%. Rack is using about 75% of one core.
- Temperature: Slightly warm / Skin temperature
Test 5: test-05.vcv (152.9 KB)
- 3 voices of Dexter with 4 x Dexter in each voice, so 12 Dexters all in all, plus various mixers, reverb, etc. I’ve chosen Dexter for testing because it uses significant CPU.
- Used CPU: 20%
- Temperature: Warm / Slightly above skin temperature
- At this point I had been running Rack with this patch for a couple of hours. I felt out the underside of the cabinet and the side towards the screen is quite warm. Seems like it’s the main heat disipation point of the cabinet. So when you do something CPU intensive on this laptop for a long time you should not be having it on your lap, and it should definately be sitting on a level, hard surface with some airflow underneath.
Test 6: test-06.vcv (284.9 KB)
- The test-5 patch doubled up, so now 24 instances of Dexter in all. So that’s a 7-row patch of which 6 are overflowing with Dexters.
- When I added the 5th voice the audio finally started breaking up using 1 engine thread. I bumped it up to 2 threads and everything was fine. After 2 minutes with the full patch it started to break up a little again, and I bumped it up to 3 threads after which it was steady.
- Scrolling the patch is still completely fluid at this load. Good GPU on that chip.
- Used CPU: 40%. Rack is using about 280% of one core.
- Temperature: Hot / A bit above skin temperature. Both the strip above and the underside of the cabinet are hot, but not hotter than you can still touch it continuously without burning yourself or being uncomfortable. The machine is thermally stable also after playing this patch for half an hour.
- I consider this patch to be very close to the practical limit of this machine. You might get away with adding another row of Dexters (so 4 more) to it, but it will likely get quite hot and another background process starting up and your audio will instantly break up I think.
My conclusion is that this little machine is very capable at running Rack and has a shocking amount of power, runs very cool unless you really push it, with very capable graphics. I feel like with Test-6 I reached the practical limit and don’t plan on doing any further testing. Using more CPU efficient modules than Dexter, I reckon that I could run a 8-10 row patch on this machine. Whether that’s sufficient for you is for you to decide.
I wouldn’t run a PRO studio on this machine alone, but for home use running moderately beefy but non-extreme patches I think it does the job well. It’s been a very pleasant surprise and it’s a fantastic “little” laptop.
Very interesting to hear that Lars - thank you!
I bought a 2020 iMac a few months back as I wanted a new computer and from experience it has generally not been a great idea to buy into the first gen of a new chip architecture - wondering if I made the right choice now haha! (although tbh having a Mac that can run Windows in Bootcamp is useful to me too).
The iMac I have now is very capable with VCV but I imagine the M1 iMac I buy in a few years time will be an absolute beast if your MacBook Air test is anything to go on.
You’re welcome, glad to be of help. And I do find these numbers interesting and promising as well.
I know I’ve been watching a lot of reviews (playlist in my post above) and the M1’s seem to regularly embarrass also much bigger Intel Macs in various ways. My late-2013 iMac 27" with Core i7-4700 3.5Ghz and fast SSD still feels quite snappy, but my gut feeling, without having measured, says that this M1 Air is creaming it in all respects. But yeah, you won’t be running x86 Windows on these machines, not virtualized either, it’ll be emulated at best and very slow.
Yeah, if the recent leaks have any truth to them at all, it seems that Apple is full steam ahead on these chips, and that there’s some monster machines coming in the near future, we’ll see. For me it’s just a very pleasant minor miracle, that we have what seems like a quantum leap in processor and graphics efficiency and performance, which should be great for everyone. That Rack seems, so far, to run comfortably on a slim laptop like the Air without a fan is just… amazing. Quite the departure anyway from its Intel siblings of the last few years, which were really bad at running something like Rack. I’ll continue my testing with bigger patches to see just how far this little miracle will go, but so far I’m very impressed with it.
By the way, could someone help me with info on what our real CPU guzzlers are these days? Is it still “Portland Weather” and then “Dexter”, or…? I need to put some load on this machine and I don’t think CPU efficient modules are going to cut it.
Thanks so much! This is music to my ears - I’ve been teaching myself Rack for the last two months on my aging (but much loved) mid-2012 15-inch Macbook Pro and I keep running into the brick wall of performance limitations. I’ve been waiting quite some time for the new M1’s.
Just to confirm - is this the base model Air with 8GB RAM?
I think I’ll go ahead and grab an Air for now and wait for the possible ‘M1x’ generation and see if they do a ‘performance’ version of the Mini with better I/O etc. Finally a reason to be buying Apple again.