M1 Air performance test results, Logic, Bitwig, Live, Reaper, and DP11.

[machinesworking]

OK all of this is subject to change, especially the results I'm getting for Logic and Reaper.
The DAWs I own that have native M1 versions with the native M1 version of Diva in AU format. [except Bitwig of course, but native Bitwig VSTi anyway]

Quick breakdown, stress test for software instrument use in the M1 MacBook Air using Diva initial patch, simple 120bpm 8th note 4 bar run followed by a two bar pause and three note chord, 6 bars in total. 48khz, 256 buffer. M1 MacBook Air 16GB, 1 TB SSD, Big Sur 11.6.

Logic 10.7-21 instances before audio engine System Overload warning. 16 with only the Performance Cores turned on in Logics Preferences.

Bitwig 4.07- 21 instances before audio starts crackling etc.

Ableton Live 11.1b6 2021 10-29- 20 instances before audio crackles

DP11.02 - 43 instances before overload window pops up. 20 with DP11's "Live Performance Mode" turned on.

Reaper 6.40 Arm64 rev 021leaf- 24 instances before audio crackling.


Pretty crazy, I don't know what's going on with Logic and Reaper here? Normally Logic, Reaper and DP on Intel anyway are in the ballpark of each other with beating Live and Bitwig, which fair enough do not do any sort of buffering of unarmed tracks, but in this case Logic looks no better than the "uninterrupted audio engine" DAWs. Reaper usually has a one or two plug in advantage over DP11, not right now. whereas DP11 only drops to that level when put into it's version of Bitwig and Lives smoother real time operation with "Live Performance Mode". It's crazy how similar they all are when that's the case, and how much more CPU efficient DP11 is than all of them, with roughly twice as many tracks.

I have no idea how DP11 is beating Logic and Reaper with track count, Reaper having it's super slim code, and Logic having "home field advantage" with the new chips, but somehow MOTU's NextGen PreGen™ as they somewhat cheekily call their audio buffering on unarmed tracks etc. plays very nicely with Apples new M1 chips.

Side note, Logic being the only DAW that turns off "Efficiency Core" support, you can see the percentage gain with them on being around 25% per core, so basically these earlier 8 core M1's are essentially 5 core machines, with the new 10 core Max and Pro being 8.5 core machines, which all fits into the x1.7 gain the new chips give.

Anyway, bragging rights for MOTU right now.

[Passing Bye]

Interesting, thanks for doing this and posting results!

[antic604]

Quick breakdown, stress test for software instrument use in the M1 MacBook Air using Diva initial patch, simple 120bpm 8th note 4 bar run followed by a two bar pause and three note chord, 6 bars in total. 48khz, 256 buffer. M1 MacBook Air 16GB, 1 TB SSD, Big Sur 11.6.

Can you please share the MIDI file? Also settings for Diva (Multicore on or off? Which INIT patch - there's 10 or so of them?).

I'd like to run such test here, to see how my i7-8750h compares to M1 Air.

[machinesworking]

Quick breakdown, stress test for software instrument use in the M1 MacBook Air using Diva initial patch, simple 120bpm 8th note 4 bar run followed by a two bar pause and three note chord, 6 bars in total. 48khz, 256 buffer. M1 MacBook Air 16GB, 1 TB SSD, Big Sur 11.6.

Can you please share the MIDI file? Also settings for Diva (Multicore on or off? Which INIT patch - there's 10 or so of them?).

I'd like to run such test here, to see how my i7-8750h compares to M1 Air.

https://drive.google.com/file/d/12rWK1_ ... sp=sharing

[Tronam]

I'd be curious to know what the 'Process Buffer Range' is set to in Logic when testing. That's the buffer which affects playback performance. Medium is essentially 512 samples, whereas Large is 1024. The 'I/O Buffer Size' is primarily for the live/record enabled tracks. Great showing for DP11 though, regardless, and it's great to see them still going strong. MOTU Performer (and later Digital) was one of the original Mac-only sequencers going all the way back to when I was in college, especially in US schools. It was the backbone of all the electronic music classes there.

[machinesworking]

I'd be curious to know what the 'Process Buffer Range' is set to in Logic when testing. That's the buffer which affects playback performance. Medium is essentially 512 samples, whereas Large is 1024. The 'I/O Buffer Size' is primarily for the live/record enabled tracks. Great showing for DP11 though, regardless, and it's great to see them still going strong. MOTU Performer (and later Digital) was one of the original Mac-only sequencers going all the way back to when I was in college, especially in US schools. It was the backbone of all the electronic music classes there.

Yeah I used it right out of high school on my friends Mac Plus. They have a small but loyal following still.

Logics Process Buffer Range was set to Large. I instantiated all 8 cores because otherwise it was much worse. IMO MOTU did something right here the other two, Reaper and Logic are going to have to fix, because the typical scenario in similar tests on Intel Macs was for Reaper to pull say 35 plug ins, DP 33-4, and Logic 31-33. Bitwig and Live would typically get trounced pulling in maybe 24-28; the zero interest in non real time operation is the main reason for that.

^^^^ This has been consistent for years, with DP pulling slightly ahead of Logic from being behind before they did their whole NextGen PreGen buffering thing, but the rest, Live being 60-80% of what Logic could do, that's been that way forever.

[Passing Bye]

Now remember, when Logic 10.5 was released I asked people in that thread if anyone could do stress test like this and nobody was interested, so later on I had a chance to do that on Intel iMac and was shocked to see Logic performed same as Bitwig, but it was quick and dirty test, was interested does clip launching aspect affects the performance, think I even tested both in clip launchers. But here most of them have similar performance, so I'm puzzled even more.

[pdxindy]

Quick breakdown, stress test for software instrument use in the M1 MacBook Air using Diva initial patch, simple 120bpm 8th note 4 bar run followed by a two bar pause and three note chord, 6 bars in total. 48khz, 256 buffer. M1 MacBook Air 16GB, 1 TB SSD, Big Sur 11.6.

I'm curious what your results would be with a buffer of 64

[machinesworking]

Quick breakdown, stress test for software instrument use in the M1 MacBook Air using Diva initial patch, simple 120bpm 8th note 4 bar run followed by a two bar pause and three note chord, 6 bars in total. 48khz, 256 buffer. M1 MacBook Air 16GB, 1 TB SSD, Big Sur 11.6.

I'm curious what your results would be with a buffer of 64

I'm away from the studio but before I do it, I would guess it would even them out a bit. Hard to say, but I do know that DP gets substantial plug in Gaines from higher buffer settings.

The way these DAWs like Reaper, DP, etc. get more plug ins is more or less by having two buffers, one that's the one you set in preferences, and the other that's some multiplier of your buffer, so you get hardly noticeable latency starting a song that buffers (more or less preemptively records), the non record enabled tracks at a much higher buffer setting. It's also why Bitwig and Live always take more CPU for the same plug ins etc. they don't have much of a secondary buffer if at all. For a hot moment Live had a separate setting for "live instruments" they dropped it pretty quick don't think it lasted a full version.

This is also why Live and Bitwig are far far less likely to hiccup your audio when you ad an instrument, or do other tasks while the sequence is running, it's more or less like every track already has the record on at all times, is in real time anyway. Easy to test, there's hardly a difference in CPU hit when you record enable all tracks in a Bitwig or Live project, but in the rest it will quickly crackle the audio doing this in a song near 80% on their CPU meters.

[machinesworking]

Oh, and Logic cheats, it disables record enabled tracks when it detects no input, which of course means there's massive latency when you attempt to play that track. Haven't messed with it yet, but Logic finally allows MIDI input by port, which is going to throw people for a loop if this behavior is still the default. I recall you can turn it off, but it's not truly gone...Apple..

[anttimaatteri]

[machinesworking]

So he's testing 16GB lowest GPU M1 Pro against Nvidea GPUs. I'm halfway through but the bias and unbalanced benchmarking is already just hard to take. At least he mentions it I guess? Of all DAWs to benchmark performance with, he chooses FL studio.

[apoclypse]

So he's testing 16GB lowest GPU M1 Pro against Nvidea GPUs. I'm halfway through but the bias and unbalanced benchmarking is already just hard to take. At least he mentions it I guess? Of all DAWs to benchmark performance with, he chooses FL studio.

My biggest takeaway is that with all that patting on the back and complaining about Apple the Image-Line guy on this forums was doing they still managed to put out a crappy performing product. Well done Image-Line.

I would love to see more DAW apps tested in a M1 vs PC matchup (S1, Bitwig, Ableton whenever that comes out of beta).

[anttimaatteri]

So he's testing 16GB lowest GPU M1 Pro against Nvidea GPUs. I'm halfway through but the bias and unbalanced benchmarking is already just hard to take. At least he mentions it I guess? Of all DAWs to benchmark performance with, he chooses FL studio.

and you are exactly WHO?

abby normal doin some idiotic random bullshit testing with plugins smashed into some daws and call that a test or a day?

god knows why they tested with fl studio. hardware unboxed is one of the bigger channels testing, next to gamers nexus. they usually dont give a ratfuck about audio-benchmarking. they just put it in on demand for testing cpus with what they deem to be "average performance environment"

maybe they all should ask you mr. go-to-guy?
but according to the stupid text you wrote about that testing, you simply havent even watched the video, but zapped thru for buzzwords. that little gpu test was ONE of a dozen or more...

[syntonica]

The thing that annoys me most about these comparisons is that they never talk about build quality and usability. I have a MacBook Pro and a fujitso Windows 10 machine that are more or less equally specced. The MacBook is still solid after 5 years, but the Fujitsu is a plastic piece of crap and the trackpad is basically unusable. The Mac originally retailed for $1100 and the PC was $1400, iirc. (I got the Mac as a refurb for $900 and the PC three-years used for $300.)

All of the Macs I've owned have served me for 4-5 years each and I've never had any issues except for my original iBook G4 (my favorite Mac I've owned) which spontaneously self-destructed after like 4 years of really heavy use.