Using virtualization...

[sstillwell]

Just curious if anyone has tried running SAC as a virtual machine with either Windows or macOS as the host operating system, and what your experiences were.

I fired up one of my Windows 10 VMs and passed through my RME Babyface Pro as a dedicated USB device to the VM (not as an emulated generic sound device...when the VM has it it's not available to macOS). Running SAC "as Administrator" and setting the "Force single CPU" and "Realtime Priority" bits lets me run at 48 samples' latency with one buffer. I haven't had the time yet to really run it for extended periods of time, but I haven't had any slipped buffers yet. I wasn't expecting a good result...so I'm confused.

I guess next step is to try it on my MacBook Pro and pass through the RME Digiface USB that I intend to use with the Focusrite Octopre MkII Dynamics that I used previously with my MOTU 2408 mk2.. To be fair, my Mac Pro where I ran the initial test is a beast...six-core 3.33 GHz Xeon with 48 GB RAM and SSD storage. The MacBook Pro is a little more limited...2.2 GHz i7 w/16GB RAM and 1TB SSD storage.

[cgrafx]

Just curious if anyone has tried running SAC as a virtual machine with either Windows or macOS as the host operating system, and what your experiences were.

I fired up one of my Windows 10 VMs and passed through my RME Babyface Pro as a dedicated USB device to the VM (not as an emulated generic sound device...when the VM has it it's not available to macOS). Running SAC "as Administrator" and setting the "Force single CPU" and "Realtime Priority" bits lets me run at 48 samples' latency with one buffer. I haven't had the time yet to really run it for extended periods of time, but I haven't had any slipped buffers yet. I wasn't expecting a good result...so I'm confused.

I guess next step is to try it on my MacBook Pro and pass through the RME Digiface USB that I intend to use with the Focusrite Octopre MkII Dynamics that I used previously with my MOTU 2408 mk2.. To be fair, my Mac Pro where I ran the initial test is a beast...six-core 3.33 GHz Xeon with 48 GB RAM and SSD storage. The MacBook Pro is a little more limited...2.2 GHz i7 w/16GB RAM and 1TB SSD storage.

I sometimes run SAW Studio on my Mac Pro Tower when Im working on my bands videos so I don't have to run back to my Studio Setup. I run a WinXP VM with the audio defaulted through Parallels sharing the Mac default audio hardware. I haven't done any actual recording with this setup, although I do have a MOTU interface connected that I record VO's on with Adobe Audition, directly on the Mac.

I'm actually not surprised that your seeing decent results, VM's are not emulators, your running code directly on the CPU and they have also gotten really good at handling the hardware layer efficiently.

All cloud computing is handled as VMs specifically because of the ability to efficiently make use of the hardware.

[sstillwell]

I sometimes run SAW Studio on my Mac Pro Tower when Im working on my bands videos so I don't have to run back to my Studio Setup. I run a WinXP VM with the audio defaulted through Parallels sharing the Mac default audio hardware. I haven't done any actual recording with this setup, although I do have a MOTU interface connected that I record VO's on with Adobe Audition, directly on the Mac.

I'm actually not surprised that your seeing decent results, VM's are not emulators, your running code directly on the CPU and they have also gotten really good at handling the hardware layer efficiently.

All cloud computing is handled as VMs specifically because of the ability to efficiently make use of the hardware.

I understand how VMs work...that's what I do for a living.

There's still overhead for the virtual-to-physical layer translation (basically an additional level of MMU or address translation), I generally guesstimate it to be about 10-15% overhead when working on business systems, but stopping to think about it, the slowest part of the virtualization is the EMULATION of physical resources like disks...those don't perform like the real thing at all (I use SQL Server extensively, and it doesn't run with the same performance as it does on bare metal). Since SAC is effectively NEVER disk I/O bound, yeah, it's going to pay less "virtualization tax" than most other applications.

Glad to hear I'm not the only one that's ever used it. I'm running Parallels as well, although I have VMware Fusion and Virtual Box "just in case". Thanks for the reply!

[sstillwell]

So far, so good.

Running a Windows 7 VM on my MacBook Pro...the VM has 2 GB RAM and 2 virtual processors.
Passing through the RME Digiface USB to the VM as a dedicated device.
Running SAC 4.5 "as administrator" with Force Single CPU and Force RealTime Priority Class.
Choosing ASIO driver with 1 x 32 buffer.
32 channels assigned to 32 inputs, EQ, Gate, Comp, and 4 monitor busses assigned to all channels, Master and first four Auxes assigned to outputs.
Load showing as 21-26%.
Zero slipped buffers so far (about 20 minutes).

Note: this is with the MBP running a Time Machine backup to a NAS server on my network over WiFi!

Bob, you've been busy while I've been away, methinks...