Oh, ok. I guess I shouldn't just assume everyone has a multi-monitor setup.
If you're only using one monitor, then yea the easiest thing will be not passing through a card and just trying to reduce the various related latencies as much as you can.
The latency would be prohibitive for things like some games or video processing work but you could probably have an good experience with Ableton. And of course it's an iterative process, down the line you could decide you do want that extra monitor.
What kind of processor do you have?
> What I'd like ideally is to have my cake and eat it too - to be able to run Ubuntu, but have Ableton running just like a native window - so I can move it to any monitor, minimize it, etc.
Spice unfortunately doesn't have Seamless/Rootless window mode implemented yet (it's a pretty big undertaking) though it is planned eventually.
Even with Xen, where you could have paravirtualized windows that do exactly what you're describing (it's fantastic), it's only available for Linux VMs.
That said, I've heard of people wrapping apps as Windows RemoteApps and achieving seamless windows that way. There's a detailed guide explaining the process but expect to get your hands slightly dirty [0]. I haven't tried it myself so I can't offer any assistance there. However, it seems easy enough and the result looks very satisfying.
> Another problem I'm thinking about is that I won't be able to passthrough the soundcard to Windows and still have sound coming out of it from Ubuntu - so I'd need to use a second soundcard and then have a mixer to combine their outputs!
libvirt has emulated audio devices that work quite well. It should for the most part be seamless, but there is definitely a noticeable latency. This latency can be reduced but for me it wasn't enough. Also, this can be a big source of DPC latency which causes audio crackling.
You're right about both points: You can't share one soundcard with Windows and Ubuntu at the same time (if you want minimal latency) and you will need a second soundcard.
If you only have one set of speakers and don't own separate monitors for Ableton (definitely worth investing in), you could use a cheap Y-cable at the cost of reduced max volume, and a slightly higher noise floor. A mixer would be the best longterm investment for the best sound,but depending on your budget/requirements for a mixer, a pair of monitors might not be much more.
Again, this is an incremental process. You can start with what you have and improve the experience slightly each month with a new purchase.
I've got an i7-8700k with a lot of RAM so it should be up to the task.
I do have multiple monitors but I would rather not use Ableton exclusively on one monitor if I can avoid it. Hmm, actually, thinking about it now, I could just use multiple HDMI inputs on the main monitor and switch when I need to. Actually that could work.
Also I do have an old mixer that I could use. At that stage things are becoming messy and complicated, but I guess that's the cost of getting things to work exactly as you want.
> I could just use multiple HDMI inputs on the main monitor and switch when I need to
That's perfect if you have multiple inputs. If you don't, a simple KVM switch would do the trick too. Unfortunately they're a little pricy for HDMI due to low supply/demand.
I have a soundcard primarily for offloading as much audio processing from the CPU as I can. Music playback of high-bitrate files is far from perfect even when you're using a soundcard, with frequent crackling, but using WASAPI mode all but completely clears up crackling when I play music.
It's also worth noting that if your passed-through GPU is decent enough, you can just use the HDMI as an audio source.
I also have a USB audio interface for my mic and instruments, directly connected to a passed-through USB controller instead of using the soundcard inputs, which helps reduce the noise floor.
Having a dedicated USB controller is very important for reducing latency. A lot of computers nowadays come with at least two, but if you have one then you'll need to get a cheap but reliable USB PCI card. And you have to be careful selecting which chipset because not all of them support Function-Level Reset which means you can only pass through once before needing to restart your computer, and not all of them correctly implement Message Signaled Interrupts which means you could have problems passing through certain configurations of PCI devices.
From here, my next plan is to purchase one of these[0] which may mean I don't even need a soundcard for output at all. It might be the best solution for virtualization right now in terms of latency and offloading processing from your CPU.
However the soundcard might still be necessary if this DAC doesn't support processing at different bit-rates on the fly. Currently my setup allows for me to play songs with different bitrates at their natural bitrate when using WASAPI, preventing the need for downsampling or upsampling. Reaaaaally noticeable with things like raw video game MIDI formats where different bitrates sound completely different.
Your processor should be fine, my 6700k and 16GB of RAM are plenty. Though if we weren't still in a NAND shortage I would upgrade to 32GB.
If you're only using one monitor, then yea the easiest thing will be not passing through a card and just trying to reduce the various related latencies as much as you can.
The latency would be prohibitive for things like some games or video processing work but you could probably have an good experience with Ableton. And of course it's an iterative process, down the line you could decide you do want that extra monitor.
What kind of processor do you have?
> What I'd like ideally is to have my cake and eat it too - to be able to run Ubuntu, but have Ableton running just like a native window - so I can move it to any monitor, minimize it, etc.
Spice unfortunately doesn't have Seamless/Rootless window mode implemented yet (it's a pretty big undertaking) though it is planned eventually.
Even with Xen, where you could have paravirtualized windows that do exactly what you're describing (it's fantastic), it's only available for Linux VMs.
That said, I've heard of people wrapping apps as Windows RemoteApps and achieving seamless windows that way. There's a detailed guide explaining the process but expect to get your hands slightly dirty [0]. I haven't tried it myself so I can't offer any assistance there. However, it seems easy enough and the result looks very satisfying.
> Another problem I'm thinking about is that I won't be able to passthrough the soundcard to Windows and still have sound coming out of it from Ubuntu - so I'd need to use a second soundcard and then have a mixer to combine their outputs!
libvirt has emulated audio devices that work quite well. It should for the most part be seamless, but there is definitely a noticeable latency. This latency can be reduced but for me it wasn't enough. Also, this can be a big source of DPC latency which causes audio crackling.
You're right about both points: You can't share one soundcard with Windows and Ubuntu at the same time (if you want minimal latency) and you will need a second soundcard. If you only have one set of speakers and don't own separate monitors for Ableton (definitely worth investing in), you could use a cheap Y-cable at the cost of reduced max volume, and a slightly higher noise floor. A mixer would be the best longterm investment for the best sound,but depending on your budget/requirements for a mixer, a pair of monitors might not be much more.
Again, this is an incremental process. You can start with what you have and improve the experience slightly each month with a new purchase.
[0]https://bmullan.wordpress.com/2012/09/17/howto-integrate-win...