2020 DAW shootout

Audio Plugin Hosts and other audio software applications discussion
RELATED
PRODUCTS

Post

OK because claims are made, but data is data. I highly encourage other people with multiple DAWs to try this. A good test of a DAWs ability CPU wise is a "to failure" test, you load a CPU hungry plug in into tracks until you get crackling or noticeable audio glitches. IMO if the DAW can play the audio back for a minute without any artifacts, then that's an acceptable level of audio violence. The test is basically measuring when you're going to need to start rendering VSTs and bouncing, IMO this is a great indicator, since at this point in time audio files are not an issue, it's about how many plug ins you can run etc. There are of course other mitigating factors, if you're running Repro-5 in it's heaviest CPU mode and for some reason need to use 12 of them, you're probably going to get different results. Ultra pigs like that strain CPU differently. Then there's teamwork, it's not any mystery that Bitwig and U-He have a great relationship, U-he plug in presets show up in Bitwig's native preset browser, and before when testing Bitwig performed much much better than it does with Reaktor here, (I'm about to do the test again maybe a bit differently ) Anyway...


The test is simple, absolutely no need to download a file etc.

Reaktor 6 Player with Blocks is free, ensemble Lumikko, patch Beetroot.
The first patch it calls up.

8 eighth notes going up a D minor scale looped over two measures, simple.

128 buffer 44100khz

I think it's fair and reasonable to let a DAW use whatever method besides the main buffer, So Reapers "Anticipative FX Processing" is on, Logic's Processor "Buffer Range" is set to high, and DP10 is using PreGen (you can't really turn it off anyway).

I'm posting results below with computer specs and any peculiarities noted.

Post

As stated above test done with Reaktor Player Blocks.
Test Machine - 09 Mac Pro (modified) 2x6 core Xeon 3.33ghz, 24GBRAM, M.2 SSDs for OS etc. Fireface 800 audio card.

Logic Pro 10.48 - 56 AUi instances. I could get more, but there were noticeable artifacts, oddly the most 'gentle' sounding of the artifacts. Logic CPU meter at 100%

Digital Performer 10.11 - 60 AUi - 60 VSTi I could get 61 but doing anything caused glitches. The Performance meter is pegged at 61 and it's no good.
Reaper v6.70- 63 AUi, 58 VSTi This one is just, weird? Reaper seems to be optimizing for whatever changes AU has been doing, or Reapers MIDI abilities are more fully realized in VSTi, MIDI out from VSTi's VS AUi etc.

Bitwig 3.1.3 32 VSTi, 31 (with sandboxing), not too bad of a CPU hit there, but much worse than Reapers performance using Diva, where it skunks Live.

Live 10.19 31 VSTi 31 AUi

Like I've often thought and tested, the "performance oriented" DAWs tend to underperform in stress tests, but I'm pretty shocked by how neck and neck Reaper, DP10 and Logic are currently on my system? I've noted version numbers because the differences are small enough to change in updates IMO.

Post

machinesworking wrote: Thu Apr 02, 2020 2:12 am...it's not any mystery that Bitwig and U-He have a great relationship, U-he plug in presets show up in Bitwig's native preset browser...
That's a weird statement. U-He's presets show up in Bitwig's browser simply because they use industry-standard .fxp and .fxb file-format for their presets. The same for example is true with Brainworx' channel strip emulations that ...also show up in Bitwig's browser.

Are we to expect a channel strip in Bitwig, with selectable SSL, Neve and Focusrite modelling? :)
Music tech enthusiast
DAW, VST & hardware hoarder
My "music": https://soundcloud.com/antic604

Post

antic604 wrote: Thu Apr 02, 2020 6:18 am
machinesworking wrote: Thu Apr 02, 2020 2:12 am...it's not any mystery that Bitwig and U-He have a great relationship, U-he plug in presets show up in Bitwig's native preset browser...
That's a weird statement. U-He's presets show up in Bitwig's browser simply because they use industry-standard .fxp and .fxb file-format for their presets. The same for example is true with Brainworx' channel strip emulations that ...also show up in Bitwig's browser.

Are we to expect a channel strip in Bitwig, with selectable SSL, Neve and Focusrite modelling? :)
Not really that weird. Urs Heckman I met in 2004 in Berlin with one of the guys from Ableton. Years before the Bitwig team split off. He knows all those people, and Bitwig often does U-He related deals. His products should run smooth in Bitwig by proxy! :)

Point taken though about .fxb. His later U-He specific plug in format does not show up. It's late here, but I need to do that Diva test again, it was pretty crazy seeing how well Bitwig held up there, and it's disappointing seeing it at Lives level in the latest round of tests. :?

Post

machinesworking wrote: Thu Apr 02, 2020 2:35 am As stated above test done with Reaktor Player Blocks.
Test Machine - 09 Mac Pro (modified) 2x6 core Xeon 3.33ghz, 24GBRAM, M.2 SSDs for OS etc. Fireface 800 audio card.

Logic Pro 10.48 - 56 AUi instances. I could get more, but there were noticeable artifacts, oddly the most 'gentle' sounding of the artifacts. Logic CPU meter at 100%

Digital Performer 10.11 - 60 AUi - 60 VSTi I could get 61 but doing anything caused glitches. The Performance meter is pegged at 61 and it's no good.
Reaper v6.70- 63 AUi, 58 VSTi This one is just, weird? Reaper seems to be optimizing for whatever changes AU has been doing, or Reapers MIDI abilities are more fully realized in VSTi, MIDI out from VSTi's VS AUi etc.

Bitwig 3.1.3 32 VSTi, 31 (with sandboxing), not too bad of a CPU hit there, but much worse than Reapers performance using Diva, where it skunks Live.

Live 10.19 31 VSTi 31 AUi

Like I've often thought and tested, the "performance oriented" DAWs tend to underperform in stress tests, but I'm pretty shocked by how neck and neck Reaper, DP10 and Logic are currently on my system? I've noted version numbers because the differences are small enough to change in updates IMO.
Okay. Now we have real numbers. It would be nice to get more data so we can get a cross section to do a graph or something. I will do some testing on two machines, maybe 3 if time permits (though I'm not going anywhere since I'm in NYC).
Studio One // Bitwig // Logic Pro X // Ableton 11 // Reason 11 // FLStudio // MPC // Force // Maschine

Post

i5 K2500 / 16Gb / SSDs / ESI Juli@

Ableton Live 10 - 20 VST2
Reaper 6 - 5 VST2 (multicore enabled: 18)

Full Reaktor 6 used, don't have the player installed

Post

.jon wrote: Thu Apr 02, 2020 8:37 pm i5 K2500 / 16Gb / SSDs / ESI Juli@

Ableton Live 10 - 20 VST2
Reaper 6 - 5 VST2 (multicore enabled: 18)

Full Reaktor 6 used, don't have the player installed
your numbers don't make any sense. Do you have the same audio card settings for both Live and Reaper? In no test I've ever done does Live beat any DAW besides Bitwig, and that's a matter of version numbers who wins.

In Reaper the only way I could get anywhere near as bad of a result would be to turn audio processing threads to 2, and turn off Anticipative FX processing. In Fact I can get only 6 on a 12 core machine with Anticipative FX processing turned off, but 63 with it on.

Post

apoclypse wrote: Thu Apr 02, 2020 2:21 pm Okay. Now we have real numbers. It would be nice to get more data so we can get a cross section to do a graph or something. I will do some testing on two machines, maybe 3 if time permits (though I'm not going anywhere since I'm in NYC).

Generalized graphs IMO are kind of useless. It would definitely have to be mentioned that it's for VSTi stress testing in a composing environment. Mostly because certain use situations will not be represented. This test specifically goes for non record enabled tracks where DAWs like Logic, DP, Reaper etc. have the ability to apply behind the scenes buffering to tracks that aren't being currently played. It's the adage that live performance is going to require different things than composing, which requires different things than live band, or orchestra recording does.

In that sense I would bet that Live and Bitwig do at least as well if not better than Cubase and DP etc. in terms of multiple MIDI controllers being used to control multiple VSTs etc. at once.

Post

I'll try this tomorrow, with Studio One and Reaper.

BTW, I don't think you're right about Bitwig and u-he... I don't see any reason why their plugins would run better in Bitwig than in any other host. There's really no underlying magic which would make them work better. :)

Post

chk071 wrote: Fri Apr 03, 2020 12:06 am BTW, I don't think you're right about Bitwig and u-he... I don't see any reason why their plugins would run better in Bitwig than in any other host. There's really no underlying magic which would make them work better. :)
viewtopic.php?f=259&t=537525

Nobody mentioned magic, but actual coding, Diva probably is coded to utilize multi threads in Bitwg efficiently, some DAW's can't utilize all the cores efficiently, actual synths even less, good coding goes long way.

Post

All I'm reading there is that Bitwig performs very differently depending on the used hardware. Don't see anything which would indicate that u-he's plugins run better in it.

I mean, seriously, why would they? VST is VST. If then, there's a difference in general VST performance between hosts, but not in regards of specific plugins.

Post

You can build VST that can crash FL or have memory leaks or whatever bug you can overlook, but that same build works like charm in Cubase, but Cubase (example) can't utilize multi cores to your advantage and your plugin overloads single core all the time and than your friend is programmer of Bitwig and you code your plugin to utilize last bit of that DAW, because on your luck Bitwig can utilize mutli cores efficiently and your plugin is coded to do just that to maximum, than throw Xeon CPU in the equation and you got yourself some amazing CPU+DAW+plugin combo. Actually my theory is that Xeon+macOS is where is at the most, you can take all this with grain of salt, it's just theory based on some DAW audio multi core testings and benchmarks that I observed, where Xeon's on macOS with Logic (10.4.5+ builds that are made having new Xeon Mac Pro in mind) just had unbelievable higher results than all other combination of CPU, OS's and DAW's.

Worth pointing out is that Bitwig test machinesworking did is after some macOS optimizations that Bitwig team did, so who knows really what's the real culprit, maybe it's sum of everything, but it's rather interesting.

Post

Passing Bye wrote: Fri Apr 03, 2020 12:35 am You can build VST that can crash FL or have memory leaks or whatever bug you can overlook, but that same build works like charm in Cubase, but Cubase (example) can't utilize multi cores to your advantage and your plugin overloads single core all the time and than your friend is programmer of Bitwig and you code your plugin to utilize last bit of that DAW, because on your luck Bitwig can utilize mutli cores efficiently and your plugin is coded to do just that to maximum, than throw Xeon CPU in the equation and you got yourself some amazing CPU+DAW+plugin combo. Actually my theory is that Xeon+macOS is where is at the most, you can take all this with grain of salt, it's just theory based on some DAW audio multi core testings and benchmarks that I observed, where Xeon's on macOS with Logic just had unbelievable higher results than all other combination of CPU, OS's and DAW's.
I need to go back to that test again, because as you can see above, with a non U-He plug in like Reaktor, Bitwig scores more like it always did, about the same bad performance as Live, nowhere near DP, Reaper, Logic etc.

Pure track count wise these old Xeons are great, but if I was to use a plug that was AVX compatible like the Legend it would get crushed. Modern Xeons do AVX though.

Post

Please do, very interested in further results!

Post

There are many variables to consider when comparing multiple DAWs against each other... and depending on how you set them, the differences can be expanded or narrowed. You have two choices, a fair comparison and a practical comparison. Given the choice, you go for the practical comparison, because there is little point in knee-capping your tools for the sake of comparison to another. In essence though... what one creates in the four walls of a big container or small container can sound the same, but what matters really is whether it's comfortable in doing so...and that's what ultimately matters most.

I feel there's very little to be gained, because to cover all these DAWs one for one would take, weeks or months and countless hours....because of the number of variables between them. If you're gonna do DAW comparison tests though, at least have a good reason for doing them and make them at least entertaining.... I did Props DAW Reason last year... because up until that point, it was heavily criticised for its poor CPU performance.

Both Ableton & Bitwig, being less CPU efficient in comparison to the other DAWs, is pretty well known by most people I would think... They are tweaked or balanced in a different direction to other larger DAWs... and that's probably to keep the cogs running the best they can.

https://youtu.be/qEwho5W6qeI
KVR S1-Thread | The Intrancersonic-Design Source > Program Resource | Studio One Resource | Music Gallery | 2D / 3D Sci-fi Art | GUI Projects | Animations | Photography | Film Docs | 80's Cartoons | Games | Music Hardware |

Post Reply

Return to “Hosts & Applications (Sequencers, DAWs, Audio Editors, etc.)”