Performance on M1 Macs

[antic604]

Yeah I get it kind of, but I think they're missing out. Running DP in Live Performance mode drops me down to 24 instances in my single note run and 20 in Extrups version with a single chord. Compare that to less than 20 instances in Bitwig on that same quarter note run, and 43 instances in DP11 with Live Performance Mode turned off.

Does DP's "performance mode" gives you all of Bitwig's features, e.g. random audio-rate modulators, incl. in the Grid that's 4x oversampled? I don't think you're comparing like to like, even if it's close on paper.

[machinesworking]

Yeah I get it kind of, but I think they're missing out. Running DP in Live Performance mode drops me down to 24 instances in my single note run and 20 in Extrups version with a single chord. Compare that to less than 20 instances in Bitwig on that same quarter note run, and 43 instances in DP11 with Live Performance Mode turned off.

Does DP's "performance mode" gives you all of Bitwig's features, e.g. random audio-rate modulators, incl. in the Grid that's 4x oversampled? I don't think you're comparing like to like, even if it's close on paper.

I think you picked and chose what to respond to. Read everything I wrote and tell me I'm not acknowledging all of Bitwigs improvising or experimenting.

I just thing some CPU saving would be cool, but honestly that was part of my rational anyway, that even in the composing arena, DP etc. present you with a template to build a song you wrote, down to score editors even. It's set up for adding in a part you might have improvised on an instrument, into a real song etc. Bitwig and it's daddy Live are set up more for jumping into a loop or Grid, Max patch etc. and generating ideas for songs. I used to call Live more of a sampler on steroids than a traditional DAW, and now that the Grid is here, you can say Bitwig is more of an instrument than a DAW, and I think that shows with it's creative use of comping VS complex grouping choices for comping things like multi tracked drums etc.

[takaperry]

Been taking Bitwig through its paces on my new MacBook Pro 14" M1 Pro and Bitwig is absolutely flying through it. All the native devices and presets load up instantly, as do M1 Native plugins such as Valhalla, Overloud TH-U, and Serum. Coming from a MacBook Pro 16" i9, the difference is very apparent. Large projects (80-100 tracks) that would glitch out at a 512 buffer on the Intel MacBook now run at about half the DSP meter on a 256 buffer size.

Thanks for that!

Do you happen to have any experience of Bitwig running on new 14'' M1 Pro vs. the "old" 13'' M1?

I've never owned a 13" M1 MacBook Pro so can't make any comparisons there.

If you want any specific comparisons between the 16" i9 and 13" M1 Pro let me know and I can test some things out. Still in the process of re-activating all my software and plugins on the new MacBook, and then hoping to put both machines side by side and run some comparisons.

[antic604]

Yeah I get it kind of, but I think they're missing out. Running DP in Live Performance mode drops me down to 24 instances in my single note run and 20 in Extrups version with a single chord. Compare that to less than 20 instances in Bitwig on that same quarter note run, and 43 instances in DP11 with Live Performance Mode turned off.

Does DP's "performance mode" gives you all of Bitwig's features, e.g. random audio-rate modulators, incl. in the Grid that's 4x oversampled? I don't think you're comparing like to like, even if it's close on paper.

I think you picked and chose what to respond to. Read everything I wrote and tell me I'm not acknowledging all of Bitwigs improvising or experimenting.

I just thing some CPU saving would be cool, but honestly that was part of my rational anyway, that even in the composing arena, DP etc. present you with a template to build a song you wrote, down to score editors even. It's set up for adding in a part you might have improvised on an instrument, into a real song etc. Bitwig and it's daddy Live are set up more for jumping into a loop or Grid, Max patch etc. and generating ideas for songs. I used to call Live more of a sampler on steroids than a traditional DAW, and now that the Grid is here, you can say Bitwig is more of an instrument than a DAW, and I think that shows with it's creative use of comping VS complex grouping choices for comping things like multi tracked drums etc.

I sort of am (picking & choosing), but I don't think it's practical for Bitwig devs to spend possibly a lot of time trying to bulid a mechanism that will analyse - again, in realtime - if given project / track / device doesn't use any of those Bitwig-unique features and engage some form of pre-buffering accordingly, to improve performance.

BUT if one's using Bitwig "the correct way", i.e. as a sound-design and/or live performance platform, the savings would likely be very small vs. the effort they'd have to go through. If you're not using Bitwig for things that make it special, why are you using it at all? I know this statement is controversial and offends some people, but this is how I feel about it. For some reason, Bitwig almost perfectly fits into how I like to do music, therefore for me it's important only that it delivers adequate performance for what I want to do. I don't care how it compares to Reaper or DP or Cubase, because I can't do there what I can do in Bitwig and even if I could, I can't as easily and fast which is important for the creative flow.

It actually hit me yesterday watching a recent tutorial from Polarity, where he'd build the project basically never interrupting the playback and I work in very much the same way. In other DAWs, people very rarely work like that - they stop it all the time, work on isolated segments, amlost never edit MIDI or audio when it's running. That "flow" makes Bitwig special and it was BUILT from the ground up for that. Perhaps that's why I never feel at home anymore anywhere else? On a very conscious level I can appreciate the depth and breadth or features e.g. Cubase or Reaper offers, but I could never find that "flow" there. For me, it's like buying Ferrari or Lamborghini to be stuck in traffic all day. I'm sure many people can take advantage of them - and their performance (both Reaper and Lamborghini ) - but that's not my reality.

[antic604]

If you want any specific comparisons between the 16" i9 and 13" M1 Pro let me know and I can test some things out. Still in the process of re-activating all my software and plugins on the new MacBook, and then hoping to put both machines side by side and run some comparisons.

Thanks for the offer, but that won't be very helpful in the decision I'm trying to make. But I appreciate it!

[docbot]

OK because I'm an idiot and hate sleep apparently, I ran the test again.
I'm getting 20 Diva instances in VST2 and VST3 natively. This is not bad, I just foolishly figured it would beat the 12 core Xeon here. What I see in Activity Monitor points to it using the Performance Cores and stacking 5 per core before crackling. Compare this to the Mac Pro with 12 cores getting only two instances per core, but eking out 24 by sheer core count.

What's interesting and weird is other DAWs for the most part performed quite worse with the 12 core, so DP11 for instance just smokes on the M1 so far. Again this is comparing a laptop to a workstation, the 12 core 3.33ghz Xeons are still good chips. DP11 has a "live mode" which is supposed to make it more like Bitwig and Live, tomorrow I'll see how it adds up without the rendering and buffering that it does to boost track count.

BTW the other reason I got the Air is it's never twice as powerful. The number quoted was 1.7 times and that looks accurate when you see the Geekbench scores.

It's quiet insane that the M1 Macbook Air is almost as fast as a 12 core Xeon!

[Tj Shredder]

From 9 years ago…

[machinesworking]

It's quiet insane that the M1 Macbook Air is almost as fast as a 12 core Xeon!

What's insane is it's faster/loads more instances in some cases. Like I mentioned I wasn't getting anything like 43 instances of Diva in DP on the Mac Pro. Single core performance is so much better it's insane. Even though in Bitwig I was getting 24 Diva tracks VS maybe 20 at most in that same test think about that in terms of what each core is doing. Each core in the 12 core Xeon is loading only 2 instances VS Bitwig using only the Performance cores of the M1 so each core is loading 5! Theoretically on my same test Bitwig will load between 34 and 40 instances in the M1 Mac and Pro.

[antic604]

Regarding Bitwig not using the efficiency cores to their fullest.

Do we know how much performance those efficient cores generate compared to performance ones? I watched 4 YT videos today comparing 13'' M1 MBP with new 14'' M1 Pro MBP and knowing that the latter have 2x more performance CPU cores I was expecting 2x CPU performance, but it looks more 1,6-1,7x so that would suggest the performance cores aren't really that much weaker? By my count, the efficiency core offers 30% of perf. core. If that's the case, it's not ideal Bitwig doesn't saturate them - that's like 4-5 Diva instances fewer!

[docbot]

It's quiet insane that the M1 Macbook Air is almost as fast as a 12 core Xeon!

What's insane is it's faster/loads more instances in some cases. Like I mentioned I wasn't getting anything like 43 instances of Diva in DP on the Mac Pro. Single core performance is so much better it's insane. Even though in Bitwig I was getting 24 Diva tracks VS maybe 20 at most in that same test think about that in terms of what each core is doing. Each core in the 12 core Xeon is loading only 2 instances VS Bitwig using only the Performance cores of the M1 so each core is loading 5! Theoretically on my same test Bitwig will load between 34 and 40 instances in the M1 Mac and Pro.

Now imagine how the M1 Max will perform... think it will be pretty hard to max them out with traditional production duties. I'll make the jump once the M2 hits the marked

[machinesworking]

Regarding Bitwig not using the efficiency cores to their fullest.

Do we know how much performance those efficient cores generate compared to performance ones? I watched 4 YT videos today comparing 13'' M1 MBP with new 14'' M1 Pro MBP and knowing that the latter have 2x more performance CPU cores I was expecting 2x CPU performance, but it looks more 1,6-1,7x so that would suggest the performance cores aren't really that much weaker? By my count, the efficiency core offers 30% of perf. core. If that's the case, it's not ideal Bitwig doesn't saturate them - that's like 4-5 Diva instances fewer!

I searched the web over and 25-30% is about what I've heard, this also holds up in regard to the 1.7x performance of the M1 Max and Pro compared. 8 performance cores plus two efficiency = roughly 8.5 cores to the M1's 5. Now in Bitwig and other real time operation DAWs you're going to see a lot more performance out of the Max and Pro since it seems that for real time operation the OS and DAW rightly backs off using the efficiency cores. DP11 looks just like Bitwig in Extups jpg in this thread, it backs off using the efficiency cores when put in "Live Performance mode". Yes it's kind of a waste, but there's probably good reasons for it, and at least you probably won't have to be cautious about using other apps when Bitwig is running since most things like Safaris etc. are just going to load on the efficiency cores.

[machinesworking]

Oh, and I'm about to test out using this hub I bought and Display Link to get a second monitor for the Air here. The one big drawback is only one display is supported by the Air and 13" MacBook, but there's a workaround. It will be interesting to see if using it affects CPU performance since it uses USB3 to connect another display.

I know everyone is salivating about the 14-16" MBPs, but I have this dumb idea I'll do shows sometime in the non Covid future, and the Air just seemed a logical choice there.

[antic604]

Yes it's kind of a waste, but there's probably good reasons for it, and at least you probably won't have to be cautious about using other apps when Bitwig is running since most things like Safaris etc. are just going to load on the efficiency cores.

I'm not sure how I feel about this. IMO the efficiency cores should be treated as a buffer that gets tapped into when performance cores are saturated. DAW is a realtime application and sound engine should have the highest priority, so running Safari or whatever else in the beckground should be put at the end of the queue and even ignored in edge cases.

However, since Bitwig still doesn't use hardware GPU acceleration, perhaps they decided to dedicate M1's efficiency cores to the GUI? But if that was the case, paradoxically new M1 Pro/Max could have - theoretically, at least - choppier GUI in Bitwig, because they have half the efficiency cores.

All of the above is unsubstantiated speculation, obviously.

[machinesworking]

Yes it's kind of a waste, but there's probably good reasons for it, and at least you probably won't have to be cautious about using other apps when Bitwig is running since most things like Safaris etc. are just going to load on the efficiency cores.

I'm not sure how I feel about this. IMO the efficiency cores should be treated as a buffer that gets tapped into when performance cores are saturated. DAW is a realtime application and sound engine should have the highest priority, so running Safari or whatever else in the beckground should be put at the end of the queue and even ignored in edge cases.

However, since Bitwig still doesn't use hardware GPU acceleration, perhaps they decided to dedicate M1's efficiency cores to the GUI? But if that was the case, paradoxically new M1 Pro/Max could have - theoretically, at least - choppier GUI in Bitwig, because they have half the efficiency cores.

All of the above is unsubstantiated speculation, obviously.

The fact that DP11 uses all the cores in Pregen (buffer for higher track counts), but also does not use the efficiency cores when using it's "live performance mode", points to either an Apple default, or a known issue with using Efficiency cores for real time operation. In other words if only one of them did not use the Efficiency cores like I first thought, then I would be upset, but since when DP11 is emulating Bitwig and Lives audio engine approach it also does not use the Efficiency cores, that points to a conscious programming decision.

In other words a physical limitation of using lower power cores along with high power ones in real time situations. Or, a decision to keep them for OS tasks so they don't interrupt the "unbuffered" real time audio.

[antic604]

The fact that DP11 uses all the cores in Pregen (buffer for higher track counts), but also does not use the efficiency cores when using it's "live performance mode", points to either an Apple default, or a known issue with using Efficiency cores for real time operation. In other words if only one of them did not use the Efficiency cores like I first thought, then I would be upset, but since when DP11 is emulating Bitwig and Lives audio engine approach it also does not use the Efficiency cores, that points to a conscious programming decision.

In other words a physical limitation of using lower power cores along with high power ones in real time situations. Or, a decision to keep them for OS tasks so they don't interrupt the "unbuffered" real time audio.

That's a good point and something that crossed my mind as well, i.e. that there's some higher latency, lower clock or something when using those Efficiency cores? I just wonder if load allocation like you're seeing (all cores in pre-buffered mode, mostly high-performance cores in realtime) is done on the OS level, or is that something that the DAWs themselves decide? Most likely something in the middle, i.e. DAWs attach priority flags to jobs & the OS takes that into account when scheduling them?