UTM

UTM is another great example of this, which is basically a GUI for QEMU. They have a Mac App store $9.99 version and the free open source version: https://github.com/utmapp/UTM

> this won't speed up llama.cpp for everyone, just for users running it in this particular kind of Virtualization.framework VM. correct. these figures apply to llama.cpp inside the macOS guest configuration we tested. Lume is the VM frontend we used, while Apple's Virtualization.framework provides the virtual GPU. bare-metal llama.cpp is unaffected. > The fix here works around a problem where the VM was causing llama.cpp to select the wrong kernels. mostly, with one nuance: llama.cpp is selecting the correct kernels for the capability answers it receives. the stock guest reports an older Apple GPU family and a 32 KB threadgroup memory limit, so llama.cpp chooses slower kernels. Our process-scoped layer reports the tested Apple 9 and 64 KB values while allowing llama.cpp to select newer paths that the paravirtual GPU successfully execute t

apple silicon is virtualization capable and the UTM app (on the app store, but open source so you can build it too) wraps Apple's hypervisor framework, allows me to run on my macbook air (m2 earlier, recently updated to m5 just to get more memory) macos as well as arm versions of both fedora and arch, with plasma and gnome (and i've used hyprland etc to toy around). it's important to set UTM to use Apple Silicon _virtualization_, because otherwise it uses QEMU and is thereby emulating. With Apple Silicon virtualization, having macos and arch and fedora all going at once is amazing. pertinent references : https://github.com/utmapp/UTM or search for UTM on the Apple app store, where it's prebuilt (and that's what i use successfully). https://developer.apple.com/documentation/hypervisor

I don't mind using Apple's native Hypervisor framework, it's better then QEMU (speed/overhead), but Apple has no support to passthrough USB ports. https://github.com/utmapp/UTM/issues/3778

Are the primitives not already there?[0] I don't see why we cannot build an app that when connected to an external monitor switches to a "Desktop Environment". Maybe, even a hacked version of UTM[1] that exposes a fully functional OS on the monitor. [0] https://developer.apple.com/documentation/uikit/presenting-c... [1] https://github.com/utmapp/UTM

Standard Debian-with-root Arm package repo in pKVM hardware-nested VM with OpenGL (v)GPU remoting! So many years of upstream work across Linux, Android and ChromeOS to make this possible. Now we need phone/tablet OEMs to support Android Virtualization Framework (AVF) and OpenTitan-derived enclaves, so this feature can move beyond Pixel hardware. https://www.collabora.com/news-and-blog/blog/2025/01/15/the-... > GFX virtualization aims at providing support for hardware accelerated 3D graphics in virtual machines. Unlike GPU-passthrough, with GFX virtualization the host and all VM guests can access the host GPU simultaneously. Vulkan and OpenGL are supported by virglrenderer using various approaches.. vDRM is a much thinner layer.. able to achieve native GPU performance, where VirGL and Venus may struggle to overcome expensive host/guest syn

In case others were similarly confused, I thought that UTM was commercial but it is Apache 2 https://github.com/utmapp/UTM/blob/v4.6.5/LICENSE

> I would love to see an emulator that could run later OS versions like Tiger. UTM. https://github.com/utmapp/UTM

You need to provide details as you're specifically talking about macOS virtual machines and the licensing limit Apple imposes. The answer is no. https://khronokernel.com/macos/2023/08/08/AS-VM.html https://kb.parallels.com/128867 https://github.com/utmapp/UTM/issues/5563

> These sources also pointed to Apple’s new iPad Pro models, which feature OLED screens, as evidence that an OLED screen alone isn’t enough to guarantee more sales. Several analyst forecasts have downgraded new iPad Pro shipments from 10 million units to six or seven million units. Apple OLED iPad Pro sales shortfall of 30% - 40% cannot be blamed on the unmatched screen. iPad Pro sales will underperform until Apple restores the virtualization API that was removed from iPadOS 16. Invest in iPad software instead of niche headsets. https://forums.developer.apple.com/forums/thread/748130 iPad has a powerful chip, but iPadOS limit the power for it. https://github.com/utmapp/UTM/discussions/3230 iPads do launch the kernel in EL2, but they don’t enable the hypervisor.

The VM disk corruption problem has been affecting all open source Apple hvf hypervisors. lima with vfkit or UTM are most promising ones for now: https://github.com/lima-vm/lima/issues/1957#issuecomment-182..., https://github.com/lima-vm/lima/pull/2026, https://github.com/utmapp/UTM/pull/5919

You can have as many as you like, but you have to pass them as arguments in the VM configuration. It's not intuitive but the instructions are here: https://github.com/utmapp/UTM/issues/2722

I also experienced performance problems with Arch Linux ARM through UTM. Parallels seems to perform much better for me. One issue might be that only OpenGL 2.1 is supported: https://github.com/utmapp/UTM/issues/4285