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.

Real world performance difference between iPad 6th gen and 2017 Pro 10.5

I am considering bumping my rig up to a 10.5 Pro from my 2 month old iPad 2018. Has anyone compared these two on real world music applications? The benches are super close on single core, but of course Multi Core is quite a disparity. Can we even make use of Multicore processing in AUM or BM3 or Quantiloop? It will be a pretty penny to make the jump as I refuse to buy used, and ill be losing 100 bucks or so selling the 2018 iPad.

Comments

  • FWIW, I don't think anything has the gen6's bang for the buck. It has a lot of horsepower per dollar. Dumping the gen 6 will and replacing it with a new pro will cost a pretty penny, and I am not sure how much more performance you will get out of it.

  • @espiegel123 said:
    FWIW, I don't think anything has the gen6's bang for the buck. It has a lot of horsepower per dollar. Dumping the gen 6 will and replacing it with a new pro will cost a pretty penny, and I am not sure how much more performance you will get out of it.

    Yeah I figured as much. I’m going to upgrade to a new Pro in 2021 or so, but in the interim might need just a hair more horsepower to float my projects. We will see. Damn sure considering sticking with what I’ve got but that niggling voice in the mind won’t shut up

  • gen6 is amazing for the money. i havent experienced any real lag or power issue yet. from my air1 it was a massive step up. i think ill keep it as my main ipad for another year.

  • The only real reason to go for a Pro vs a 2018 iPad is for more storage. Performance is pretty much on a par. The 2018 iPad is by far the best value in the lineup.

  • If second hand Pro is out of question then I don't think the 10-15% performance increase is justified vs price. The additional 2G ram could be useful in some specific cases but hardly noticeable in everyday use.
    I'm waiting till March, if new budget iPad appears it could possibly beat 2GenPro, if not I'm replacing my Air2 with 6th gen.

  • Thanks for the replies guys! Sticking with my 6th gen for now. It hasn’t failed me yet I’m just surprised by the DSP meter in Quantiloop creeping into spooky territory, although it hasn’t crashed or crackled on me at all under any load I can throw at it. (Get the gen6 folks)

  • edited February 3

    When the gen3 Pros came out at the end of last year, the Beatmaker community (over at Intua forum) started a B3nchmark project to try to accurately compare BM3 performance across the whole range of iPads (precisely so users could make informed choices about bang-for-buck and upgrade plans).

    You can view the results below, but I'd like to make 2 important points:

    1. When we benchmarked the gen3 2018 iPad Pro models they were suffering from the throttling CPU bug (which has been greatly improved in the latest iOS firmware beta). We are still waiting to re-run benchmarks for these devices when new iOS firmware releases.
    2. Over on the Intua forum we are only a small group compared to this forum, and unfortunately none of our users had the gen5 or gen6 standard iPad (2018 model) to test with. If anyone would like to run the test for us, you simply have to download a couple of small BeatMaker 3 projects, then run them at different latency settings inside BM3 and record the peak/average DSP load for each. It only takes 10-15 minutes to complete (depending on your familiarity with BM3), and instructions are available on the site below.

    B3nchmark website:

    http://5pin.link/b3nchmark/

    Preview:

    Using the results you can see, for example, that an iPad Air2 is approx 1.8x faster than a Mini2, and a 1st gen iPad Pro is 1.4x faster than a Air2 ...and 2.5x faster than a Mini2 (interestingly, the results of the gen1 and gen2 pros were very similar).

  • edited February 3

    @tk32 said:
    [Benchmarks]

    I just ran the tests with an 11 Pro - Gen 3, running iOS 12.1.3. I’m assuming someone has to update the charts manually, so I’m posting the results here for those who don’t want to wait:

    1024 samples - 18%
    512 - 14%
    256 - 7%
    128 - 5%

    That aside, some feedback on the benchmarking setup (I wasn’t able to locate the corresponding thread over at the Intua forum, so I’ll post it here. Please pass it on).

    • It took way longer than 5 minutes. No big deal for me (I didn’t believe it in the first place :p ) but please, don’t be misleading.
    • The instructions are terrible. Seriously. Why not post them as a simple text list with clear steps? Instead we have to watch a sh***y YouTube video, listening to a rambling guy who talks about iOS/BM3 bugs which have _absolutely nothing_ to do with the instructions. And, of course, the video cuts off suddenly. Oh god, I hate YouTube videos. /rant
    • The instructions are incomplete. Do we test with 16 or 24 bits? Measurement Mode on or off? Sure, some of those setting are visible in the video, but it would be far better to mention those essential settings explicitly. As text. And still: do we have to use the built in speaker? Apples 3,5 mm adapter (remember, you want people to test their Pro devices without a headphone jack)? An USB audio interface? IDAM?
    • The results site isn’t really working on an iPad ¯_(ツ)_/¯ and the color coding is confusing.
    • Probably 5 more things I’ve already forgotten.
  • I appreciate you taking time to do the benchmark.

    You make several good points, but it's a shame you choose to do it an angry and petulant way. This is all community-driven, and the aim is to help people get accurate data about how iPads preform when used to make music.

    I wasn't trying to mislead with my 5-min estimate, but it's been several months since I last ran the test. How long did it take you?

  • I don't get it... from my experience reducing buffer puts more stress on CPU -> higher %,
    increasing it helps your CPU cope with the demand -> lower %
    What am I reading wrong from this diagram?

  • Nevermind, found the answer at linked site.
    A bit confusing to put these values together on the same chart, as you measure completely different things. When making comparisons it makes sense to compare A to A and B to B.

  • @tk32 said:
    I appreciate you taking time to do the benchmark.

    You make several good points, but it's a shame you choose to do it an angry and petulant way. This is all community-driven, and the aim is to help people get accurate data about how iPads preform when used to make music.

    Yeah, sorry, it turned into a rant while I was writing (and testing). I didn’t mean it that way and I tried to stay constructive, even while ranting. Again, I’m sorry. The thing is, you won’t get accurate and comparable data when the instructions are unclear. That’s what ticked me off a bit.

    I wasn't trying to mislead with my 5-min estimate, but it's been several months since I last ran the test. How long did it take you?

    It took me way longer because I wasn’t aware you can import zip files directly into BM. First I downloaded the zip file with Dropbox, exported it into another app, unzipped it there and imported it into BM. Then I wanted to load the project in BM, but had to locate the sample on my own because BM couldn’t find it initially. Just to name the most time consuming stumbling block. Again, this could’ve been clarified with one simple sentence.

    There were other things as well, just because the instructions were unclear. All in all I had to reboot my iPad 3 times to ensure a clean setup.

    And to your point re „community-driven“: it’s a two-way-street. The easier it is to run the tests, the more users will participate. I was this close to abort it several times, but in the end I wanted to help. Peace <3

  • edited February 3

    Nvm

  • edited February 3

    Geekbench CPU benchmark gives very good overview about performance of individual models... At least it’s good for relative comparision (how much power has one model compared to other). In my experience Geekbench numbers are very good approximation of reality...

    Regarding using CPU meters inside app - what @brambos said ! CPU meter starts making a bit sense just when you start reaching CPU limit (constantly, not just in peaks) - as indicator that you need freeze or resample :)))

    Probably good way is to choose some reference plugins which do no take too much CPU so they can run in many instances also on older devices and then do test by adding them until you start getting crackles... Ideally at least 5 different plugins to avoid issue with shared memory when using too many instances (or to use just build in host plugins)

    Then you can compare how much instances can handle given device.

  • edited February 3

    First things first, I have joined here to answer this thread, I have no idea how long this account will be active because the last account i had, I specifically asked @Michael to delete entirely and i have no idea what his policy is on that, so this account could be removed immediately, I know I am not particularly welcome here and that is fine.

    @whiteout said:
    I just ran the tests with an 11 Pro - Gen 3, running iOS 12.1.3. I’m assuming someone has to update the charts manually, so I’m posting the results here for those who don’t want to wait:

    1024 samples - 18%
    512 - 14%
    256 - 7%
    128 - 5%

    I did receive your results and want to thank you very much, it is appreciated very very much so.

    @whiteout said:

    • The instructions are terrible. Seriously. Why not post them as a simple text list with clear steps? Instead we have to watch a sh***y YouTube video, listening to a rambling guy who talks about iOS/BM3 bugs which have _absolutely nothing_ to do with the instructions. And, of course, the video cuts off suddenly. Oh god, I hate YouTube videos. /rant

    I am very sorry about this, It has been made clear to me in the last few weeks that i should not do videos at all, first i started getting messages about how terrible my videos are, then i was told by @Shazamm that i only make videos to get my ass kissed, and now a sh***y video where i am rambling, needless to say that the videos i do have obviously been a vast failing on my part and i am evaluating carrying on with them, I never got these complaints on my desktop videos, so i should probably just stick to tutorials for desktop software, please note though, i never activated adverts or anything on my videos, I don't think I ever pretended to be something i am not, I was just trying to help others, again, something i failed at in this instance.

    @whiteout said:

    • The instructions are incomplete. Do we test with 16 or 24 bits? Measurement Mode on or off? Sure, some of those setting are visible in the video, but it would be far better to mention those essential settings explicitly. As text. And still: do we have to use the built in speaker? Apples 3,5 mm adapter (remember, you want people to test their Pro devices without a headphone jack)? An USB audio interface? IDAM?

    The cut off video was not my fault, Youtube had a brain fart, as soon as i was informed this morning, i replaced it with a new video, but on reflection i will remove that too and just replace it with text instructions, I am really sorry that this video wasted your time, that is not my intent with any of my videos, again, sorry for any wasted time.

    @whiteout said:

    • The results site isn’t really working on an iPad ¯_(ツ)_/¯ and the color coding is confusing.

    I don't know what to say about this ?
    I test heavily on phone and iPad and have actually removed some really clever things that worked well on desktop, but were poor on mobile devices, but I am clearly not a web developer, I just do what i can to try and help get some stuff out there for a community, I am really sorry you had issues, and I have included a video to show that it works fine on my iPad, I am sorry I don't have the funds to cover every type of iPad available for testing the site, if you can email me, I can try my best to fix any issues you are having, but again, I am not a web developer.
    As you can see from this video though, on my iPads specifically it is designed to fit and works perfectly.

    Color coding is my fault entirely, I thought that having Minis be all shades of red and Standards be all shades of yellow and Pros all shades of green would be useful, without doing these things and getting feedback Its the only way to see if they work.

    @whiteout said:

    • Probably 5 more things I’ve already forgotten.

    If you could remember them, that would be amazing for me, I am no web developer as i keep saying, so everything I do on the site is done from feedback, for instance the day before yesterday i added individual charts for the Mini/Standard/Pro at @tk32 request and it is a great idea.

    @recccp said:
    A bit confusing to put these values together on the same chart, as you measure completely different things. When making comparisons it makes sense to compare A to A and B to B.

    I should do the four buffers on separate charts, I will try this as soon as i get chance, thank you for the idea, much appreciated.

    @whiteout said:
    And to your point re „community-driven“: it’s a two-way-street. The easier it is to run the tests, the more users will participate. I was this close to abort it several times, but in the end I wanted to help. Peace <3

    This is a valid point, when i set these tests up, all the initial results were done by Beatmaker power users in the Discord chat, so they were done in no time at all and I was on hand to answer any questions, it is very much my fault that from this i assumed everybody would be able to carry out the tests in the same fashion, extremely poor instructions on my behalf, again, sorry for any wasted time, that was not my intent at all.

    Some ponts in general..
    First off, the site was not live, was not mean't to go live yet and as such is not even the same site as when the links were posted to this thread, as @tk32 can testify, because he had to update his post.
    @tk32 is extremely scene oriented like myself and was trying to be helpful a user here, as he always is trying to help a user somewhere.
    Geekbench is just that, a benchmark for geeks where somebody can claim an iPad Pro performs betters than a Macbook, this is real use case testing.

    @brambos has some serious concerns about the validity of the tests and I would happily have taken the page down if I was not asked to let it remain by a bunch of users that pointed out that these test are the reason why the Gen 3 pro was found to have serious issues very quickly, the first reports to Apple came after these test where run, so they had at least one use.
    @brambos concerns lie in the fact that unless you are at certain points of stress with every device, you can't tell what is happening internally, be it less cores running etc etc, however these tests were not designed with that in mind, the point of this part of the suite was to give somebody who had an iPad, to look at real world use of a particular project and be able top see how much more resource headroom an upgrade would give them, in this regard it works well.
    It was always intended that the next steps of the benchmarking would be max polyphony tests for real world peak to peak CPU performance comparison, and max reverbs for real world RAM comparison, I am designing those tests right now, but after reading this thread, I really have some work to do on the instructions for the first set of tests, so hopefully they will be available tonight, but may take a few days while i work out how to create pointed clear instructions.

    @brambos has also kindly offered to speak to others in his developer circle and try to work out best practices for undertaking these tests, not only in Beatmaker 3, but also in other software like AB and AUM, I will happily work with anybody who wants to set up benchmarking for their individual software of choice, again i am no web developer, but we can get the charts up there, no matter how poor my design choices are, we can always change them.

  • edited February 4

    I think it's a great idea @winconway - and even if not yet perfect, the dataset you have already compiled seems to prove that multiple devices of the same kind report very similar readings. To me, this indicates the test has merit in helping people determine the capabilities of each device.

    Obviously having guys like @brambos offering advice is massive, but I think the best way to test the validity of this benchmark is to encourage more people to take part. The more data we compile, the more likely we are to see how reliable this method is at giving an objective (ball park) figure for the capabilities of each iPad.

    As for your role in uncovering the gen3 pro issues... hats off. The initial alert really did come from your benchmark test, as far as I can tell (there was no reference to the throttling before the benchmark started to generate ripples in the community).

    (ps. hope you don't get re-removed)

  • Advice from @brambos needs to be taken at face value at all times, he knows more than most and has helped a lot of developers behind the scenes, getting his input on this is extremely important and is the only way for us to get valid real world tests that are useful to users and maybe even developers too.

  • edited February 4

    http://5pin.link/iab/
    I completely overhauled this trying to take on board the feedback given here and the feedback given by @brambos
    It now has much more explanation and written instructions and the charts themselves are separated by buffer and easier to read.

  • edited February 5

    Great, thanks

Sign In or Register to comment.