Live 12 with Apple M1 8 Core Vs M4 12 Core Speed Performance Comparison

[spanktron]

Hi everyone.

So as we all know M2+M3 were not the best regarding their Multicore implementation with Live.
Or better said, Live was not making the best use. Resulting in the M2+3 somehow not beeing worth an upgrade back then at least for me.

Has anyone here compared the M1 Base model to the M4 Base model Macbook pro or mini with Ableton both in Cpu+ Render-Export speed difference?

I have found only one video on YT and the guy had the M1 in 48khz and compared it to a M4 running the same project in 44.1 so,,,, bummer

I have a MBP Pro 14 Inch m1 base model with 8 cores of which 6 are Performance.
I cannot for the love of God predict if i would have any significant gain by buying a M4 base model which only has 4 Performance cores.
Yes they are faster but also 2 less than mine. So i cant imagine it being amazingly faster than mine? Unless anyone here knows more

Also would love to see a comparison between the 14 Mbp m1 compared to the m1 max version when using Ableton

[mr.adl]

This guy has extensively tested M1-M4 using different DAWs, inlcuding Ableton.
https://youtu.be/hccy19Hm6M8?si=r9P9ipnHVlMFCFHc

[Machinesworking]

I have an M1 Air and an M2 Studio Ultra and there are not any weird issues with Live. The main and only real issue with Live and these new chips that have efficiency cores and performance cores is they really do not use the efficiency cores for any sort of plugin processing.

IMO in the bigger picture the issue is how many cores, and how many plugins can be stacked on those cores before failure. I think it's entirely possible that Live in particular would behave best with an Ultra with more performance cores rather than M1-M4 On thing that sucks for Live and DAWs that don't really use e cores is M4 is a more powerful chip with less p cores and more e cores. The stats look similar, but until we get real life tests it's very possible than an M2 or 3 with more p cores is going to perform better than an m4 with less.

The difference in performance is not helped by the extra e cores in the M4 with Live. Live gets 90 tracks VS 140 in Reaper and Digital Performer here, and that's due to Live not really doing any work with performance cores, this is with an M2 Ultra 16 p and 8 e cores.

[spanktron]

This guy has extensively tested M1-M4 using different DAWs, inlcuding Ableton.
https://youtu.be/hccy19Hm6M8?si=r9P9ipnHVlMFCFHc

Hi there.

I`ve seen this, thank you.

The problem is that he is comparing the Pro chips not the Base models. I did not clarify it but added it now to the title of the Thread.
Also he is comparing a 10 core m1 with a 12 core m1 so i don`t know what conclusion to draw to be honest.

I would be satisfied with much simpler data, ideally someone here who had an m1 and then bought an m4 and did a comparison
„same project on both machines Cpu load and Render Speed test“ kind of thing.

All videos on YT are with Pro and Max chips

[spanktron]

I have an M1 Air and an M2 Studio Ultra and there are not any weird issues with Live. The main and only real issue with Live and these new chips that have efficiency cores and performance cores is they really do not use the efficiency cores for any sort of plugin processing.

IMO in the bigger picture the issue is how many cores, and how many plugins can be stacked on those cores before failure. I think it's entirely possible that Live in particular would behave best with an Ultra with more performance cores rather than M1-M4 On thing that sucks for Live and DAWs that don't really use e cores is M4 is a more powerful chip with less p cores and more e cores. The stats look similar, but until we get real life tests it's very possible than an M2 or 3 with more p cores is going to perform better than an m4 with less.

The difference in performance is not helped by the extra e cores in the M4 with Live. Live gets 90 tracks VS 140 in Reaper and Digital Performer here, and that's due to Live not really doing any work with performance cores, this is with an M2 Ultra 16 p and 8 e cores.

I hear you. Thanks.

There is a guy on YT that compared them m1 and m4 mini BUT unfortunately he had 44,1k SR on the M4 and 48 on the m1 so this is a huge
deviation in the comparisons objectivity by itself already imho.

[Machinesworking]

.
There is a guy on YT that compared them m1 and m4 mini BUT unfortunately he had 44,1k SR on the M4 and 48 on the m1 so this is a huge
deviation in the comparisons objectivity by itself already imho.

It is for sure, because is he setting the same buffer? For most of us a setting of 128 makes sense, if a DAW is particularly good at getting extra CPU from a setting of 1024 and above that's cool and all, but most of us write in Live, not just mix and master etc.

The elephant with testing is how many of a plugin will fit on a single core? and how good is a DAW at splitting a plugin between cores? In my case 90 instances of Diva divided by 16 cores (again the 8 ecores do almost nothing here) = 5.625 instances per core, So 5 full instances or 80 Diva plus 10 more divided among the 16 cores. I'm sure I'm just coming up with nonsense to a real developer, but the math is always even and it does say that Live is capable of using multiple cores for one instance, otherwise it would be 80 instances. Live does not pre-render or double buffer plugins like Reaper and DP do (Cubase and other also do this), so it's pure CPU force when all plugins are firing.

In other words DP and Reaper can get 50 more instances because they commit to RAM some of the CPU cycles by using a larger buffer or recording the audio to RAM. As soon as you put the record enable button on with DP or Reaper they crumble to less plugins than Live.

All told the m4 is a more modern chip, just don't get fooled by e cores, they simple don't matter much in Live and Apple is putting in more of them in each iteration of the M series it seems..

[spanktron]

.
There is a guy on YT that compared them m1 and m4 mini BUT unfortunately he had 44,1k SR on the M4 and 48 on the m1 so this is a huge
deviation in the comparisons objectivity by itself already imho.

It is for sure, because is he setting the same buffer? For most of us a setting of 128 makes sense, if a DAW is particularly good at getting extra CPU from a setting of 1024 and above that's cool and all, but most of us write in Live, not just mix and master etc.

The elephant with testing is how many of a plugin will fit on a single core? and how good is a DAW at splitting a plugin between cores? In my case 90 instances of Diva divided by 16 cores (again the 8 ecores do almost nothing here) = 5.625 instances per core, So 5 full instances or 80 Diva plus 10 more divided among the 16 cores. I'm sure I'm just coming up with nonsense to a real developer, but the math is always even and it does say that Live is capable of using multiple cores for one instance, otherwise it would be 80 instances. Live does not pre-render or double buffer plugins like Reaper and DP do (Cubase and other also do this), so it's pure CPU force when all plugins are firing.

In other words DP and Reaper can get 50 more instances because they commit to RAM some of the CPU cycles by using a larger buffer or recording the audio to RAM. As soon as you put the record enable button on with DP or Reaper they crumble to less plugins than Live.

All told the m4 is a more modern chip, just don't get fooled by e cores, they simple don't matter much in Live and Apple is putting in more of them in each iteration of the M series it seems..

I dont know what he did with the buffer cause when i saw 44khz on one and 48 on the other i stopped watching tbh.

To the other points, i understand what you are saying even though i know very little about processors on a technical level.

However i am confused cause in a video of another YT-er he compares everything from m1max to m4 10core and 12 core and the
m4 10 core base model has like a geekbench of 14000+ while the m1 max has lower than 13000 and in the graph the m1 max is a bit better than the
m4 that has a higher geekbench when using ableton? Why i am mentioning this. I looked at the geekbench tests and realised they are if i am not mistaking a result of multiplying the result of one performance core( only the performance cores).

So if a machine has 4 P cores with 2500 and 2 E cores the geekbench result is 10000 (2500x4) and not 2500x4+ E cores result. Unless i get it wrong.
So even with 4 Performance cores the m4 should outperform the m1 max but it doesnt according to this video. Only the 12core m4 does almost double the performance of the m1 max.

Thats why i would LOVE if someone would post any info of a direkt comparison of CPU stress and Render speed between a m1 max and an m4 Base model. I find this much better than calculating and predicting.

[brianwdowling]

Note that with M1 macs, some only support 1 additional external monitor, and require specific docking stations that support DisplayLink technology if you want to have multiple monitors. So make sure that you consider this in any decision and check the specs.

[b-righteous]

I am an old PC user who never owned a Mac. However, the upgraded 16 core Mac Studio looks interesting to me. 12 P cores means not such a big hit with Live not using the E cores. Single core performance and efficiency seem to be off the charts compared to AMD/Intel too. Would be nice if the portable devices had less E cores. I guess it makes sense the way they configure those because the E cores produce less heat so it's easier to squeeze more multicore performance out of it. Just not good for Live though. Presonus by comparison just added an option to allow Studio One to use E cores. This is nice since it lets the user chose what is optimal based on their CPU.

[cids]

Maybe you've already seen these:

https://www.youtube.com/watch?v=3o5AgsRCofY

https://www.youtube.com/watch?v=JzhW0_B0Z64&t=345s

[spanktron]

Maybe you've already seen these:

https://www.youtube.com/watch?v=3o5AgsRCofY

https://www.youtube.com/watch?v=JzhW0_B0Z64&t=345s

Thank you.

I had missed the second one so this was useful even though he did not go in depth with a render speed comparison etc but it seems
as if the m1 max has more or less similar power to the m4 base model.
This is a good thing. I ordered an m4 12core and will compare and post my findings here soon

The second video link you posted is the one i am mentioning where he messed the sampling rates between the two so it is not a valid comparison for me.

[spanktron]

Just wanted to post my findings comparing my old m1 macbook Pro M1 16gb/512 Base model to my new M4 12Core 24/512 Mbp

I did not go in depth and will not waste your time.

So. Same OS, Same plugins, Same Track, Same Live version


CPU Load: M1 8 Core Average 53%
M4 12 Core Avergage 29%

Export Speed Test for the Track which was 3:08 Long, 24 Bit, 44,100Khz:

M1 8 Core 3min 18,Sec
M4 12 Core. 1min 37,5 Sec