- 80comments
- 6comments
- 124comments
- 52comments
- 34comments
- 595comments
- 139comments
- 12comments
- 237comments
- 84comments
- 170comments
- 73comments
- 6comments
- 234comments
- 42comments
- —discuss
- 4comments
- 121comments
- 3comments
- 159comments
- 13comments
- 25comments
- 62comments
- 29comments
- 45comments
- —discuss
- 374comments
- 40comments
- 120comments
- 49comments
If AMD are at all smart they'll be attempting to contribute AMD-V support to https://github.com/intel/haxm ASAP.
Now that will be a fun pull request to watch.
Already in the issue tracker, although the "have to ask my superiors" comment isn't too encouraging: https://github.com/intel/haxm/issues/5
Looks like the superiors just replied :)
https://github.com/intel/haxm/issues/5#issuecomment-34652021...
Check again - by now it has a (kinda positive) "Not sure how to do it and if it makes sense to do that in the same driver, but we'd be open for discussions" reply.
Ugh, no Hyper-V support kind of defeats the point these days... Especially on Win10... Maybe there'd be a way to do this for Hyper-V too
I don’t understand. Hyper-V is already hardware accelerated and takes full advantage of VT-D.
Hyper-V doesn't have the same emulated devices support by a long shot.
Hyper-V is extremely strict about what can access its hypercalls, so you can't do anything from userspace like with KVM and HAX. Hypercalls can only be made from ring 0 and creating domains can only be done from the root domain.
There's a project that uses QEMU to emulate the original xbox, xqemu: http://xqemu.com/
I wonder if this will help their project too.
That doesn't help the project at all... HLE seems to be more workable for Xbox... Xenia is VERY far for the Xbox 360, it isn't technical but organisational issues instead
HLE sucks for the XBOX, there is many low level stuff which can be replicated well without hacks.
You can't compare Xenia to the 1st XBOX, at all.
I tried it back in the day while talking about the results on reddit with jayfoxrox.
HL2 ran much faster, (the GUI "widgets" and loading bars were nearly as fast as an XBOX) but the graphics were scrambled.
Is there an equivalent to HAXM for Linux? Would it be KVM?
Yes, exactly. All of KVM, HAXM and the OSX Hypervisor.Framework are basically providing APIs to user applications that encapsulate the CPU's virtualization capability so that you can use it to implement VMs without needing kernel privilege. They differ in how high a level of abstraction they provide (eg KVM does a lot more for you in the kernel, H.f punts to userspace for everything), but the principle is the same.
It's on my backlog of experiments that I'll probably never get to, but has anyone had success running QEMU on, say, AWS (virtualized) Windows servers to run Linux VMs? I presume HAXM isn't relevant for attempting this?
I don't think any of the public cloud providers support nested virtualization.
Several do support nested virtualization under Linux; I've tested that. I don't know if they do under Windows, though.
[Disclaimer - I work for Microsoft]
Azure just released nested virtualization VMs. I've tested Hyper-V running in such a VM and it worked well. Have not yet tested HAXM.
Azure and GCP do now.
Can I ask why? Can’t you just spin up any number of smaller instances running Linux AMIs instead?
HAXM works on OSX AFAIK, it would be nice to have it in OSX as well
Didn't Android SDK on Windows use qemu & haxm for years already?
Yes.
HAXM started off as a part of the Android SDK for Windows but is now a standalone component as its wider applicability has been noted and capitalized upon.