Audiobus: Use your music apps together.

What is Audiobus?Audiobus is an award-winning music app for iPhone and iPad which lets you use your other music apps together. Chain effects on your favourite synth, run the output of apps or Audio Units into an app like GarageBand or Loopy, or select a different audio interface output for each app. Route MIDI between apps — drive a synth from a MIDI sequencer, or add an arpeggiator to your MIDI keyboard — or sync with your external MIDI gear. And control your entire setup from a MIDI controller.

Download on the App Store

Audiobus is the app that makes the rest of your setup better.

New iPads performance?

Hi!

So I'm currently using the 6th gen iPad 9.7" for musicproduction along with some hardware. I really love this setup and I have no plans of moving to a computer.

But with all the new iPads that are coming out right now i'm definitely considering an update.

These are my current options:
iPad Pro 12.9" 2020
iPad Air 2020
Used iPad Pro 12.9" 2018

What I wish to get out of a new iPad:
Better performance in AUM (more plugins with less DSP-load)
Bigger screen (less scrolling in apps)
Better performance in general, in all musicapps.
Higher pixeldensity

So I really want to hear your experience with upgrading to one of these iPads? Can you actually notice a difference in for example AUMs performance compared to an older iPad?

My 9.7" iPad has the A10 Fusion chip and both the A12z(ipad pro) and A14(ipad air) is alot faster on paper but how does that really show in real life use?

Thanks in advance!

«1

Comments

  • I have all the iPad Pro 12.9 since the First way back in 2015. Since the 2018 model I don’t see any performance upgrade that are worth the money. If you care about the money buy 2018, if the money is not a problem go for the 2020 but sincerely you will not notice the boost that you would expect from it.

  • With the iPad Air 2020 at 599$, I don't see many reason to buy something else. Right now, the iPad Air is the fastest (on single core) iPad on earth for less money than the Pro. I really like my iPad Air 2020 :) Faster than my iPad Mini 5

  • @Montreal_Music said:
    With the iPad Air 2020 at 599$, I don't see many reason to buy something else. Right now, the iPad Air is the fastest (on single core) iPad on earth for less money than the Pro. I really like my iPad Air 2020 :) Faster than my iPad Mini 5

    Yeah, the only downside is the storage,
    $599 for 6 gbs, personally for me it’s definitely not enough storage for serious audio production, and even the next storage option at 256 gbs can get filled up if you’re an appoholic like some of us, lol. Don’t get me wrong, if you only install what you really need and use 256 gbs can go a long way, I used an Air 2 with 128 gbs for years but I usually ended up with 5 gigs of free space.

  • edited November 2020

    Is the Ipad 8 of this year a noticeable upgrade vs the 2018 by the way? Has anyone compared for music production/synth/aum etc?

    Not just to save money but to keep the headphone jack for mobile production, I like the standard ipads vs the air and pro for that reason at least. If only there was a high end pro model with a high end hp jack option lol

  • @donvito

    2020 iPad has A12 chip so it's performance for audio apps is basically same like 2019 Air3/mini5 and basically almost same like 2018 pro model .. so yeah, it's significant improvememt when compared to 2018 iPad with A10

  • exactly - there is clearly visible that for audio apps, there is almost no difference in performance of latest PRO (A12Z), larest Air, MINI and iPad 8th gen - they all have almost same single core performance...

  • Would love to see a comparison of current and older models in a real world case, i.e. multiple instances of an auv3 synth in AUM, until dsp is at max. Also how much does 6 vs 4 Gb RAM matter? I rarely encounter problems with memory usage but some apps like NOISE or bismarck (with large soundfonts) definitely complain less on my 10.5 iPad Pro than on my iPad 2018, and not because of single core performance...

  • @GLacey said:
    I like this comparison:
    https://browser.geekbench.com/ios-benchmarks

    @dendy said:

    exactly - there is clearly visible that for audio apps, there is almost no difference in performance of latest PRO (A12Z), larest Air, MINI and iPad 8th gen - they all have almost same single core performance...

    1585 vs 1110 is a significant difference to me , it is a 42% of more performance.

  • @dendy said:

    exactly - there is clearly visible that for audio apps, there is almost no difference in performance of latest PRO (A12Z), larest Air, MINI and iPad 8th gen - they all have almost same single core performance...

    Multi-core performance does have an impact for audio apps. Even if the main audio is handles in a single thread. Many audio apps are multi-threaded in ways that benefit from multiple cores. And starting with iOS 14, developers can start writing audio codes that is multithreaded.

  • edited November 2020

    Multi-core performance does have an impact for audio apps. Even if the main audio is handles in a single thread. Many audio apps are multi-threaded in ways that benefit from multiple cores.

    Which one ? None of currently available AU plugins. None of currently available DAWs. Maybe miRack (according to what dev said, but did't saw performance comparision lets say A12 vs A12X - in theory there should be twice as much performance )

    Anyway for workflow used by most users here (some host - DAW or AUM + AU plugins) there is near to zero usage of multicore performance...

    I would be happy to be wrong. There is nothing easier that to try it - to compare same setup (same plugins in same host, with same buffer size) let's say on A12 cpu (i can check it with my mini5 and A12x cpu or A12Z cpu (pro models)

  • @dendy said:

    Multi-core performance does have an impact for audio apps. Even if the main audio is handles in a single thread. Many audio apps are multi-threaded in ways that benefit from multiple cores.

    Which one ? None of currently available AU plugins. None of currently available DAWs. Maybe miRack (according to what dev said, but did't saw performance comparision lets say A12 vs A12X - in theory there should be twice as much performance )

    Anyway for workflow used by most users here (some host - DAW or AUM + AU plugins) there is near to zero usage of multicore performance...

    I would be happy to be wrong. There is nothing easier that to try it - to compare same setup (same plugins in same host, with same buffer size) let's say on A12 cpu (i can check it with my mini5 and A12x cpu or A12Z cpu (pro models)

    This has all been gone over before, so I'm not getting into it again. But there are two important things to consider when looking at multi-core and iOS. The first is that yes, having multiple cores (and threads) help all applications, even those that are not directly using them, perform better. Even the efficiency cores help with this. In playing with the new M1 Mac, one of the things that is really obvious is that most of the background bookkeeping type tasks are running on the little cores. This frees up the power cores to do the real work, including the real-time audio threads.

    As far as the second consideration goes, just look at this link, https://developer.apple.com/documentation/audiotoolbox/workgroup_management/understanding_audio_workgroups

  • edited November 2020

    i know about audioworkgroups api (i was first one who posted about this great improvement after WWDC here ;-)) - but it would be suprising if any current DAWs or plugins are using it... maybe in future, 2-3 years from now

    sorry for skepticism, but i'm observing iOS apps evolution since 2010, and i'm more and more skeptic about future timeframe of large scale adoption of audio workgroups api sooner than 2022-23...

    as i said, it's nothing easier than make some real-world sceniario and test it... just propose host/plugins... then make project which will need, for example, 80% of CPU on iPad with A12 cpu, and then try same project on A12x, A12z or even A13 and A14 CPU

    i think we can find people with devices with all of those CPU

    my proposal is:

    • Cubasis 3 (because it's available on both iPhone and iPad, som we are not limited just to iPads. Alternative is AUM or Nanostudio2)
    • test plugin: Model D (you don't need too much of them to reach 80% of CPU)
    • let's use same buffer size, for example 512

    i will try it on my Mini5, then somebody with latest PRO model will test it there, etc. etc.

    Let's do a bit of experimental science instead of endless debates ;-)

  • @dendy, it’s a very good idea to carry out real measurements! We should play the same project in CB3 on different devices. Same song, same AU instruments, same latency, same everything and check the dsp meter. The song should be ‘composed’ in a way that keeps the dsp nearly at a constant level.
    I can contribute with my Air 3. I’m fine with model D, but almost whatever AU instrument can be chosen for plug-in 😜

  • I did a test. Just simply added modelD instances one by one in Cubasis 3 and checked the DSP gauge.
    Settings: sample rate: 48kHz, Latency: 11.6 ms (I think it’s 512 bits)
    iPad model: Air3
    DSP load vs number of modelD instances:
    1: 14-15%
    2: 28-30%
    3: 45-48%
    4: 58-60%
    5: 74-80%
    6: 94-98%

  • @GLacey said:
    I did a test. Just simply added modelD instances one by one in Cubasis 3 and checked the DSP gauge.
    Settings: sample rate: 48kHz, Latency: 11.6 ms (I think it’s 512 bits)
    iPad model: Air3
    DSP load vs number of modelD instances:
    1: 14-15%
    2: 28-30%
    3: 45-48%
    4: 58-60%
    5: 74-80%
    6: 94-98%

    Did you try to push it more? (I mean until it starts to crackle like crazy?)

    I ask this since the newer CPUs cores need some time to 'activate/speed up' and I pretty sure the Air 3 should be able to run more that 6 instances of Model D.

    For example in AUM's the CPU meter can be at 90% and then when you add a few more instances it drops down to 50% when the high efficiency cores get activated on the more modern CPUs like the one in iPad Air 3.
    (I mean my ancient Air 2 pushes 4-5 instances before it 'cracks up' using a 512 buffer).

    I'm expecting that when I get my next iPad I should at least be able to run 10 instances om Model D without the iPad cracking up.

    Cheers!

  • @Samu said:

    @GLacey said:
    I did a test. Just simply added modelD instances one by one in Cubasis 3 and checked the DSP gauge.
    Settings: sample rate: 48kHz, Latency: 11.6 ms (I think it’s 512 bits)
    iPad model: Air3
    DSP load vs number of modelD instances:
    1: 14-15%
    2: 28-30%
    3: 45-48%
    4: 58-60%
    5: 74-80%
    6: 94-98%

    Did you try to push it more? (I mean until it starts to crackle like crazy?)

    I ask this since the newer CPUs cores need some time to 'activate/speed up' and I pretty sure the Air 3 should be able to run more that 6 instances of Model D.

    For example in AUM's the CPU meter can be at 90% and then when you add a few more instances it drops down to 50% when the high efficiency cores get activated on the more modern CPUs like the one in iPad Air 3.
    (I mean my ancient Air 2 pushes 4-5 instances before it 'cracks up' using a 512 buffer).

    I'm expecting that when I get my next iPad I should at least be able to run 10 instances om Model D without the iPad cracking up.

    Cheers!

    This is definitely true. DSP load meters in DAW's have always been pretty suspect and now they are even more so. Apple's audio system is very good at running until the system is just exhausted. To figure out how hard you can push one of these machines, you are going to need to load it up until either the audio starts to sputter and die or the UI becomes useless.

  • I’m holding out on the new pros before I decide on an upgrade
    Hopefully spring?

  • edited November 2020

    @GLacey said:
    I did a test. Just simply added modelD instances one by one in Cubasis 3 and checked the DSP gauge.
    Settings: sample rate: 48kHz, Latency: 11.6 ms (I think it’s 512 bits)
    iPad model: Air3
    DSP load vs number of modelD instances:
    1: 14-15%
    2: 28-30%
    3: 45-48%
    4: 58-60%
    5: 74-80%
    6: 94-98%

    in all cases bellow i used default patch, no notes played (which is ok, Model D eats basically same CPU even if it just sits in orojects and plays nothing when compared to playing some notes)

    Nanostudio 2

    • 48khz, buffer "high" (it's 512), iPad MINI5 (2019 model)
    • 6 instances of Model D at steady 81% CPU, clean sound no crackles
    • 7 instances at steady 94% CPU, crackles

    AUM (rest is same)

    • basically same numbers like in NS2

    Cubasis 3 (rest is same, buffer 11 ms which is equivalent to 512)

    • 6 instances CPU load 84-87%, no rackles
    • 7 instances 98-100%, crackles

    Beatmaker 3

    • 6 instances 79%
    • 7 instances 92% and crackles
    • note that in my experience BM3 CPU meter is unreliable, it shows always a bit less that other hosts but starts crackle with same amount of plugins, no matter which plugins you use for test, i tried it also with less cpu demanding plugs (to get higher granularity and precision) and looks like BM3 CPU meter shiws always 2-4% less that other hosts but REAL CPU load is approximately same

    now somebody with A12x (Pro 2018), or A12Z (pro 2020)

  • @dendy, great, thorough test.
    In my first test I didn’t hit play, just added the instances and opened one of them.
    If I hit play I get the same 85-ish dsp load with 6 instances and crakling with 7.

    However my experience is that I rarely have crackling issues with other au synths. So out of curiousity I tried the same excercise with synthmaster one, I made a one bar track with 4 quarter notes, hit play (loop) and started duplicating the track. I reached 30 instances, but DSP is not going above 50%..

  • edited November 2020

    @GLacey yeah model D is one of most CPU heavy plugins on iOS, and sadly it is not optimalised (i really don't see reason why plugin should take any CPU in case it is not playing anything).

    But also thanks to this, it is great tool for such CPU load tests, because you really don't need many instances of it to totally kill CPU :-)))

  • 11” pro 2020 here, same results as @dendy

  • edited November 2020

    @drcongo said:
    11” pro 2020 here, same results as @dendy

    Cool, thanks ! So this is consistent with what i said .. 11" pro as A12x which has twice as much high performance cores as A12 (mini5/air3) .. but audio apps performance is same ..

    IT would be interesting to try some miRack patch which takes CPU on the edge on A12 ... any tips which miRack modules are really CPU heavy ?

  • Anyone with Air 4?

  • @dendy said:
    @GLacey yeah model D is one of most CPU heavy plugins on iOS, and sadly it is not optimalised (i really don't see reason why plugin should take any CPU in case it is not playing anything).

    Korg Gadget is another example of that.
    This is without playing or touching anything:

  • edited November 2020

    @R_2

    what a shame... this is really big thing, because if all plugins would be properly optimalised to use CPU (or most of their CPU needs) really only when they're doing some job, in general you will be able to load lotmore plugins in project, because not all are playing simultaneously all the time...

  • edited November 2020

    This is a very interesting and important thread as many users do consider to buy a new device and need some guidance. I also think that CPU is not the only limiting factor. As comparisons between Intel and M1 Macs showed there was a significant difference in how many plugins you could run in Logic between the 16 GB M1 and the 32 GB Intel machine. So we should definitely also take a look at the RAM. Here a comparison I compiled from wikipedia article and the before mentioned geekbench link

    iPad 6 iPad 7 iPad 8 iPad Mini 5 iPad Air 3 iPad Air 4 iPad Pro 3 (11/2.9) iPad Pro 4 (11/2.9)
    Release 2018 2019 2020 2019 2019 2020 2018 2020
    CPU A10 A10 A12 A12 A12 A14 A12X A12Z
    Cores 4 4 6 6 6 6 8 8
    GB RAM 2 3 3 3 3 4 4 6
    Resolution 2048x1536 2160×1620 2160×1620 2048×1536 2224x1668 2360x1640 2388x1668 2732x2028 2388x1668 2732x2028
    Geekbench single 756 747 1111 1111 1113 1585 1118/1129 1121
    Geekbench multi-core 1403 1378 2413 2703 2828 4193 4610/4672 4649

    Please note that:

    • I ignored that the iPad pro 3 with 1TB storage has got 6GB RAM
    • I consider the iPad 6 as a good reference device as it is probably the youngest device where the owners (like myself and the OP) would already consider an upgrade
  • @krassmann thanks for table, very informative !!

    So for now looks like iPad Air 4 is far away best model for audio apps (no, size of memory really doesn't matter for our use case, albeit some people thinks it does. nope.)

  • @dendy said:
    So for now looks like iPad Air 4 is far away best model for audio apps (no, size of memory really doesn't matter for our use case, albeit some people thinks it does. nope.)

    Yes, I agree.

    • iPad Air 4 is the best general purpose choice if 256 GB is enough and the budget should not exceed $1000
    • iPad 8 is the best for the budget musician. CPU is almost on par with iPad Air 3 and double the power than the previous generation
    • Used iPad Pro 3 if you want more screen estate or storage and the budget should still be reasonable.
  • @Samu said:

    Did you try to push it more? (I mean until it starts to crackle like crazy?)

    I ask this since the newer CPUs cores need some time to 'activate/speed up' and I pretty sure the Air 3 should be able to run more that 6 instances of Model D.

    For example in AUM's the CPU meter can be at 90% and then when you add a few more instances it drops down to 50% when the high efficiency cores get activated on the more modern CPUs like the one in iPad Air 3.
    (I mean my ancient Air 2 pushes 4-5 instances before it 'cracks up' using a 512 buffer).

    I'm expecting that when I get my next iPad I should at least be able to run 10 instances om Model D without the iPad cracking up.

    Cheers!

    Sorry I forgot to answer, yes I tried to push over 6 it instances, but it started crackling. I think Model D is so resource hungry that the CPU immediately switches into panic mode when you add the first instance :smiley:

    Anyway I think I saw many times what you're saying in AUM, the CPU is around 50% and it stays 50ish even if I add further (not model :smile: ) instruments.

Sign In or Register to comment.