Why is Diva and Repro Performance so Poor on M1 Max MacBook Pro?
-
Phantom Menace Phantom Menace https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=621128
- Banned
- 15 posts since 30 Jul, 2023
Hi,
I recently purchased an M1 Max MacBook Pro with 64GB of RAM and a 2TB internal NVMe drive to replace my ancient 15" 2015 Quad Core 2.8GHz Intel Core i7 MacBook Pro. I mostly use hardware synths so my 2015 MacBook Pro has served me reasonably well. However, I do write music in my hotel rooms on my many business trips. And the 2015 MacBook Pro was just not powerful enough to run many modern plugins.
However, I'm astonished at just how poor the performance of Diva and Repro-5 is on my M1 Max MacBook Pro. I honestly think my 2015 Intel MacBook Pro performs better. It's just a complete joke trying to run Diva and Repro-5 on my M1 Max MacBook Pro. Even single notes come close to maxing out the performance meter in Cubase Pro 12.
I thought maybe I had downloaded old Intel versions of Diva and Repro-5 given that they haven't been updated since 2021. But nope. Those are the latest versions of these plugins. What's up with that, btw? I would have thought u-he would have highly optimized their flagship plugins for Apple Silicon by now.
I of course have the "Multicore" option enabled for both plugins. But they still easily use 10x the processing power of all of my other 100+ synth plugins including Hive 2.
WTF? How can the performance be this bad?
Incidentally, I bought the used 14" M1 Max MacBook Pro to hold me over until the 3 nanometer M3 Max laptops are available. But I can't even use the M1 MacBook Pro with Diva and Repro-5. Something is completely messed up.
Do I need to sell the M1 Max MacBook Pro and buy something like a 13th Gen Core i9 13900H 15" Dell XPS laptop that is loud and hot as f*ck? Is that the tradeoff I need to make in order to get usable performance?
I recently purchased an M1 Max MacBook Pro with 64GB of RAM and a 2TB internal NVMe drive to replace my ancient 15" 2015 Quad Core 2.8GHz Intel Core i7 MacBook Pro. I mostly use hardware synths so my 2015 MacBook Pro has served me reasonably well. However, I do write music in my hotel rooms on my many business trips. And the 2015 MacBook Pro was just not powerful enough to run many modern plugins.
However, I'm astonished at just how poor the performance of Diva and Repro-5 is on my M1 Max MacBook Pro. I honestly think my 2015 Intel MacBook Pro performs better. It's just a complete joke trying to run Diva and Repro-5 on my M1 Max MacBook Pro. Even single notes come close to maxing out the performance meter in Cubase Pro 12.
I thought maybe I had downloaded old Intel versions of Diva and Repro-5 given that they haven't been updated since 2021. But nope. Those are the latest versions of these plugins. What's up with that, btw? I would have thought u-he would have highly optimized their flagship plugins for Apple Silicon by now.
I of course have the "Multicore" option enabled for both plugins. But they still easily use 10x the processing power of all of my other 100+ synth plugins including Hive 2.
WTF? How can the performance be this bad?
Incidentally, I bought the used 14" M1 Max MacBook Pro to hold me over until the 3 nanometer M3 Max laptops are available. But I can't even use the M1 MacBook Pro with Diva and Repro-5. Something is completely messed up.
Do I need to sell the M1 Max MacBook Pro and buy something like a 13th Gen Core i9 13900H 15" Dell XPS laptop that is loud and hot as f*ck? Is that the tradeoff I need to make in order to get usable performance?
- KVRAF
- 14456 posts since 16 Feb, 2005 from Planet Earth, Somewhere
I think I have read they perform much better with multicore off on apple silicon.
I assume you have tried with multicore off and on?
rsp
I assume you have tried with multicore off and on?
rsp
sound sculptist
- KVRAF
- 6209 posts since 25 Dec, 2004
here... Macbook Pro 2021 ( https://support.apple.com/kb/SP858?view ... cale=en_US )
Rosetta mode...
Diva Init preset on Divine mode 3 note chord sitting at 13% on Ableton Live meter at 128 sample buffer. No multicore. Switch multicore on and the meter jumps around at 25% with audio pops and clicks abound. This running through the Macbook speakers. Which always pop in Live anyway, there is definitely an issue with the inbuilt speakers which is another issue altogether.
Native mode...
same settings ^ CPU meter is around 11%, multicore is still popping and meter around the same
I definitely would not recommend switching multicore on.
Rosetta mode...
Diva Init preset on Divine mode 3 note chord sitting at 13% on Ableton Live meter at 128 sample buffer. No multicore. Switch multicore on and the meter jumps around at 25% with audio pops and clicks abound. This running through the Macbook speakers. Which always pop in Live anyway, there is definitely an issue with the inbuilt speakers which is another issue altogether.
Native mode...
same settings ^ CPU meter is around 11%, multicore is still popping and meter around the same
I definitely would not recommend switching multicore on.
sketches... http://soundcloud.com/onesnzeros
some artists i support... https://bandcamp.com/spectraselecta
some artists i support... https://bandcamp.com/spectraselecta
-
tasmaniandevil tasmaniandevil https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=62450
- KVRAF
- 2171 posts since 22 Mar, 2005 from a planet called u-he
Our multicore feature had been developed and optimized for intel CPUs.
Thus it might not work as efficiently on Apple Silicon.
That's why we do not recommend to use this option on Silicon Macs, or use it only on selected instances of the plugin.
We can run hundreds of Diva voices without the multicore option, even on the oldest M1 Silicon MacBook from 2020.
Thus it might not work as efficiently on Apple Silicon.
That's why we do not recommend to use this option on Silicon Macs, or use it only on selected instances of the plugin.
We can run hundreds of Diva voices without the multicore option, even on the oldest M1 Silicon MacBook from 2020.
That QA guy from planet u-he.
- u-he
- 30194 posts since 8 Aug, 2002 from Berlin
Yes, MultiCore does not work well because some voices end up running on the Efficiency Cores of the CPU, which are very slow. This throttles overall performance.
We have not yet found any means to prevent our plug-ins from processing on Efficiency Cores, so for the time being it's advised to keep the MultiCore option turned off.
That said, the voice count of Diva is excessively high on AS, we're very happy with the overall performance.
We have not yet found any means to prevent our plug-ins from processing on Efficiency Cores, so for the time being it's advised to keep the MultiCore option turned off.
That said, the voice count of Diva is excessively high on AS, we're very happy with the overall performance.
-
Phantom Menace Phantom Menace https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=621128
- Banned
- Topic Starter
- 15 posts since 30 Jul, 2023
Thank you all for your very helpful answers. Yes, Diva and Repro perform much better with the "Multicore" option turned off on my M1 Max MacBook Pro. However, the CPU consumption is still at least 5x greater than any of my other recent "CPU hungry" plugins. These "M series" chips from Apple seem great for iPhones and iPads but piss poor for anything but rendering certain types of video files (and even then are mostly left in the dust by the latest nVidia GPUs).
However, I'm just curious if I would run into the same performance problems with an Intel Core i9 15" Dell XPS laptop with an 13900H processor. If I'm not mistaken, this laptop features a CPU with only 6 performance cores and 8 efficiency cores. Of course these cores run at much higher "turbo boost" speeds than similar Apple M1 cores. But would the Diva and Repro "Multicore" performance be just as bad given that there are only 6 performance cores in this processor?
Although I would hate to live with the heat, noise, horrible battery life, and absolute shit trackpad on the 15" Dell XPS laptop (that would need to always be plugged in to achieve optimal performance,) I would be willing to put up with it if the performance is SIGNIFICANTLY better than the Apple M1 Max laptop.
Thank you for your advice and recommendations.
EDIT: I have absolutely no faith that the "upcoming" 3 nanometer Apple "M3 Max" processor that is at least 1 to 2 years away will even match the performance of the current Intel Core i9 13900H processor, let alone what Intel releases 2 years from now.
I'm no Apple fanboy although I do firmly believe Apple's industrial design is so far ahead of any Windows laptop as to be almost laughable. The only Windows laptop that even comes close is the Dell XPS series.
Having said that, I really could care less between macOS Ventura+ and Windows 11. Yes, macOS looks more refined and stylish and has some nice UI features that Windows 11 lacks. However, I have to install just as many paid 3-party utilities (especially around window management) that macOS lacks to achieve the same essential functionality that is built into Windows 11. Plus, IMO, the Windows 11 taskbar puts the clunky, outdated macOS dock to shame (again, unless you install several paid 3-party utilities that break almost every time Apple updates their OS).
Finally, the macOS menu bar is just stupid. It was designed for the OS9 era when no one had more than one monitor. It's just incredibly clunky and borderline unusable on multiple monitors. The way Windows includes the "menu bar" locally in each application makes SO much more sense in a multi-monitor setup.
However, I'm just curious if I would run into the same performance problems with an Intel Core i9 15" Dell XPS laptop with an 13900H processor. If I'm not mistaken, this laptop features a CPU with only 6 performance cores and 8 efficiency cores. Of course these cores run at much higher "turbo boost" speeds than similar Apple M1 cores. But would the Diva and Repro "Multicore" performance be just as bad given that there are only 6 performance cores in this processor?
Although I would hate to live with the heat, noise, horrible battery life, and absolute shit trackpad on the 15" Dell XPS laptop (that would need to always be plugged in to achieve optimal performance,) I would be willing to put up with it if the performance is SIGNIFICANTLY better than the Apple M1 Max laptop.
Thank you for your advice and recommendations.
EDIT: I have absolutely no faith that the "upcoming" 3 nanometer Apple "M3 Max" processor that is at least 1 to 2 years away will even match the performance of the current Intel Core i9 13900H processor, let alone what Intel releases 2 years from now.
I'm no Apple fanboy although I do firmly believe Apple's industrial design is so far ahead of any Windows laptop as to be almost laughable. The only Windows laptop that even comes close is the Dell XPS series.
Having said that, I really could care less between macOS Ventura+ and Windows 11. Yes, macOS looks more refined and stylish and has some nice UI features that Windows 11 lacks. However, I have to install just as many paid 3-party utilities (especially around window management) that macOS lacks to achieve the same essential functionality that is built into Windows 11. Plus, IMO, the Windows 11 taskbar puts the clunky, outdated macOS dock to shame (again, unless you install several paid 3-party utilities that break almost every time Apple updates their OS).
Finally, the macOS menu bar is just stupid. It was designed for the OS9 era when no one had more than one monitor. It's just incredibly clunky and borderline unusable on multiple monitors. The way Windows includes the "menu bar" locally in each application makes SO much more sense in a multi-monitor setup.
Last edited by Phantom Menace on Sun Aug 20, 2023 3:42 am, edited 1 time in total.
- u-he
- 30194 posts since 8 Aug, 2002 from Berlin
We have done extensive tests on Apple Silicon with Diva and Repro and we have concluded that the performance is outstanding.
-
Phantom Menace Phantom Menace https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=621128
- Banned
- Topic Starter
- 15 posts since 30 Jul, 2023
Thank you for your reply. However, my question is whether an Intel Core i9 13900H mobile processor with 6 performance cores and 8 efficiency cores significantly outperforms an Apple M1 Max processor when running Diva and Repro?Urs wrote: Sun Aug 20, 2023 3:31 am We have done extensive tests on Apple Silicon with Diva and Repro and we have concluded that the performance is outstanding.
I'm not asking you to pick company favorites based on personal preferences, OS features, corporate policies or any other irrelevant metrics to the question at hand.
I just want to know if Diva and Repro's performance running on an Intel Core i9 13900H is significantly better than that on an Apple M1 Max.
I believe this information is quite relevant to your customers who rely on your software for optimal performance.
Thank you.
Last edited by Phantom Menace on Sun Aug 20, 2023 4:20 am, edited 2 times in total.
- u-he
- 30194 posts since 8 Aug, 2002 from Berlin
-
Phantom Menace Phantom Menace https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=621128
- Banned
- Topic Starter
- 15 posts since 30 Jul, 2023
If that is the case, then perhaps you can compare the 12th Gen Intel Core i9 12900H processor to the Apple M1 Max in terms of Diva and Repro performance.
Thank you for your transparency and forthrightness.
-
machinesworking machinesworking https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=8505
- KVRAF
- 8018 posts since 15 Aug, 2003 from seattle
I have no idea what weird things you’ve done to your M1, but I have a M1 Air here that has no issues at all with Diva. I do not own Cubase though so that may be your issue. Diva performs fantastically in Live, Logic, DP, Bitwig and Reaper here.
- u-he
- 30194 posts since 8 Aug, 2002 from Berlin
I'll ask around next week if anyone in the office has that particular CPU. We have, hmm, maybe 4 or 5 employees who use Windows or Linux PCs, but I don't recall if anyone got an upgrade as recently as what I think those i9s are.
We usually test things on machines that perform poorly in order to make sure that we can give advice to users who e.g. are on a budget.
We usually test things on machines that perform poorly in order to make sure that we can give advice to users who e.g. are on a budget.
- KVRAF
- 6209 posts since 25 Dec, 2004
this is a sus thread
sketches... http://soundcloud.com/onesnzeros
some artists i support... https://bandcamp.com/spectraselecta
some artists i support... https://bandcamp.com/spectraselecta
-
machinesworking machinesworking https://www.kvraudio.com/forum/memberlist.php?mode=viewprofile&u=8505
- KVRAF
- 8018 posts since 15 Aug, 2003 from seattle
Quick test, M1 Air in Ableton Live, not arguably the most CPU sucking DAW out there, built in audio, init preset, Great mode, 6 voices, 6 note chord, 128 buffer, 16 tracks. [same number in Reaper] There's no way in hell that Cubase performs worse than that. DP11 can do 22 tracks, and seems to allocate instances or resources to efficiency cores just fine. Everything is maxed out in DP, whereas in Live only the Performance cores are near max. This applies with Reaper as well, and the same numbers show.
So Bitwig and Live- 16 tracks no use of Efficiency cores.
Digital Performer and Reaper- 22 tracks, Efficiency cores being used.
Something's weird about OPs set up, that's all I can say, the M1 Air is significantly underpowered compared to his machines.
-
- KVRian
- 1278 posts since 24 May, 2004
Some numbers from my 16" 2021 M1 Pro, Multicore turned off, 44100 / 256 buffer size, in Ableton Live (native Apple Silicon): With the "BS Deep Space Diva" preset, each Diva instance playing 6 notes, I can have up to 51 Diva instances playing at the same time without dropouts. So 306 voices in total. There are other more expensive presets that use Voice Stack, but the performance is fantastic on my end.
Phantom Menace, what are your results (in numbers) using this preset? Or if you have a particular preset where you found the performance to be poor, I can test it on my end.
Phantom Menace, what are your results (in numbers) using this preset? Or if you have a particular preset where you found the performance to be poor, I can test it on my end.
