I got the MS surface with a Snapdragon(R) X 12-core (X1E80100 @ 3.40 GHz; 16 Gb RAM) running windows ARM. I was expecting Cb14 pro (ARM) to run smoothly on this but to be honest, I’m a bit underwhelmed. A 10 track project (mainly audio, 4 instr tracks) with around 6 insert-plugins running and a fx-channel (revelation) already starts glitching (pops, audio-drops) when I open an instance of Chrome at the same time. The asio-guard is almost maxing out. Exporting a project takes around the same time it would take on my 9year-old W10 pc.
This makes me feel that my windowsARM is not fully optimised for running a DAW. Does anyone have any experience with this ?
Thanks!
Cubase 14 runs great on my Samsung Snapdragon X-Elite 16gb ram
ARM options are great, but do not push them on math operations, esp. not on float 32.
Stick to 24bit integer or even 16b when quality is good enough, and don’t go crazy on sampling rates either.
Thanks, useful info!
If you mean the recording bit depth in project settings with “24 or 16 bit”, that is pretty much completely irrelevant. Cubase’s mix engine always works with 32bit (or 64) float, so as soon as you move a fader or insert a plugin, the integer value from your audio file gets converted to a float value anyway. In a way, if you set the bit depth to 32bit float, you skip that conversation step, thus actually saving cycles…
1 Like
This.
Makes sens, apparently at least.
Then I would recommend to avoid the 64b float option at least.
I wonder how many 3rd party plugins however chose also to convert everything to 32b float internally?
If CB (and plugins?) converts everything internally to 32b float anyway, then it’s a strike for most ARM options, since they are great for moving data or simple integer maths, but not so good at float maths. Some don’t even have any FPU, and those with FPUs are still lagging behind. (note Apple is in a class of its own here)
Not a complete showstopper to use a Snapdragon for CB or other DAWs, but just to say GHz and core-count does not tell the whole story when selecting a processor, especially when complex float maths are involved. Memory bandwidth is also often slower on ARM.
Lots of reasons to love ARM, but it’s just not top of its class for every workloads.
Float maths and memory bandwidth are probably two reasons why you experienced more audio pops and drops than you expected.
Make sure also you have a clean OS installation and avoid all unnecessary background processes, that can also be a significant culprit…
All of them, or more precisely, the DAW does it, if necessary. It is part of the VST specification. Plugins get a buffer of 32f values , or 64f ( if they and the DAW support it), and also return floats. No DAW nowadays works with fixed point audio engine, the last one that did was Protools TDM.
Regarding the OP issue:
First thing I would do is exclude Cubase and the project folders from windows defender realtime scan. Then make sure that at least the Steinberg Power Scheme is enabled, somewhere in Studio setup, to make sure it is not an energy saving issue.
Interesting - thanks for the info.
For my curiosity, are you referring to a VST3 requirement, or was it already part of the VST2 spec.?
VST3 and VST2 and probably also VST1, if there ever was such a thing 
afaik VST2.4 added support for double processing, i.e. 64bit float.