kode54

Oops, I don't know how I missed that. Thanks for pointing it out to me.

You can always try setting the network to use virbr0 instead of br0, and then it will be segregated from the external network. I'm not sure if this also includes NAT for outgoing connections, or whether it has no access to br0 at all.

True. They should try ACS override anyway, and if it works, it'll be his solution. The hardware solution was not really a solution, since it implies expensive purchases, which are a crazy thing to do if you can live without the functionality and don't really have money to burn. I just didn't want them to get their hopes up. But at least it's something that can be reversed easily in the event that it doesn't work.

Assuming it works with his configuration. It certainly didn't with mine, but my machine has a lowly i7 3770, a whole generation older than Haswell.

Possible issue for emulator halting: Drive image may not have been fully allocated on the cache, either as a sparse file or as a non-raw format that supports extents, and then the cache was filled up. Qemu automatically pauses VMs if it cannot allocate more space for an image that hasn't been fully allocated. I don't know how much this applies to raw images with sparse allocation.

Delete the Trial.key from your USB drive. You won't even need to reboot.

In all likelihood, both of your video card ports are associated with the PCIe root hub, and without a Core Extreme or Xeon processor, you won't be able to separate them from the root controller(s) they are associated with now. Short of buying a new motherboard and processor, and possibly memory (Xeon will be using ECC RAM), there's not much you can do.

Correcting my post. It worked on first reboot. Then a Windows Update and reboot later, it failed. Safest option for recent and as of this post current NVidia drivers: No need for features -> kvm -> hidden, or at least doesn't appear to be necessary. No veridian or hyperv sections in features. No hypervtimer in clock. Tossing those into the system post-install seems to work for one boot, then fail. Either the HyperV status isn't being enabled immediately, or the OS is installing the drivers for it after the NVidia drivers already start up, causing it to not be detected. Either way, a good idea to leave these options off when running NV drivers, and use either AMD or Intel or VNC when you need a VM with HyperV, such as for Windows Phone simulation.

I'm not sure how much I'd trust a Xonar card. I don't know how it relates to Xonar DGX, but my Xonar DX has horrible crosstalk between the two front ports. Verified to be a problem of the card itself, as the crosstalk does not occur with the same case front port connected to the onboard audio.

GTX 960 4GB And I learned from another topic on this forum that the NVidia "code 43" errors only apply when freshly installing the drivers with HyperV and other such features visible. It's perfectly safe to turn those back on after rebooting the machine post-install. Gotta be careful of automatic driver updates, though.

"Contact us for a price" for OEM and Enterprise for embedding in a locally hosted software product leads me to believe it would be incredibly expensive to license an all-inclusive license for the entire unRAID platform for all existing and future users. I mean, the charts look pretty, but sounds like a lot of trouble for first party inclusion.

Right, I see the problem. http links, and they're Flash embeds. I guess I'll browse those topics with Chrome.

If this applies to all NVidia cards, you'll either have to get a newer video card, or reroll that VM using Seabios instead of OVMF. It basically says that 630 and lower, and older series cards do not support UEFI booting. You're welcome to see if you can find a UEFI firmware for it to add as a parameter in the XML. Assuming there is a Mac version of a 500 series card you can hope someone has dumped the firmware from, if it'll even work.

I simply assumed from searching and finding that ACS functionality only exists in Core Extreme and Xeon parts meant that the ACS override feature would be broken on anything less than those configurations. I, too, experienced this issue attempting to bind my PCIe video cards when I had ACS override enabled, or at least failed to separate my second video card from the first one. If it really matters, I can try rebooting with ACS override enabled, and see if my Windows VM will even boot without manually binding the PCIe root that is normally coupled with it.

I am using Safari 10 on macOS Sierra, with Adblock disabled on this domain, and no other filters running here, and embedded YouTube videos are not rendering. I only see a big empty space where a video should appear. This happens for any post linking to YouTube videos.

They don't have ACS on root ports, yet they are using the ACS override, which is attempting to break up the video card from its root port(s). They need to kill that kernel option, reboot, and plan their IOMMU adventures properly for the hardware they have.

As mentioned by saarg here in the pre-release forum, the HD audio cannot be separated out of its IOMMU grouping on the Skylake platform.

Are you sure you didn't just enable ACS override where there was none before? Maybe your IOMMU groups changed layout. I had this issue when enabling ACS override on a processor with no ACS feature, where it would try to separate the video cards from the PCIe root controller they're normally grouped with. No I have had ACS enables the entire time. Also there is no issues with IOMMU group 16. It is just the NVIDIA video and audio devices in it. I see from one of your posts that you have a Core i7 4790K. Unfortunately, the only non-Xeon processors blessed with ACS support are the Extreme models, which are not even socket compatible with the regular Core processors in the same series. You will need to get either a Xeon or a Core Extreme processor and motherboard to really play with splitting up IOMMU groupings. And if you intend to go with the easiest configuration of booting to a GPU other than the one(s) you wish to bind to VMs, you'll probably want one of the Xeon E3 models with integrated graphics, and a motherboard with connectivity for that. With what you've got, you'll have to disable that ACS overrides option, and live with the groups you get by default. As far as I may guess, it's quite possible that attempting to use ACS override mode will result in VMs that can't boot unless you manually bind all of the devices that are ordinarily in the same IOMMU group as the video card. In the case of my Core i7 3770 system, that would include the PCIe root controller, and whatever video card is also plugged into the middle slot. You can try to bind the minimum of devices that you want in one VM, but you won't be able to bind any of the unbound devices into another VM. Sorry.

Are you sure you didn't just enable ACS override where there was none before? Maybe your IOMMU groups changed layout. I had this issue when enabling ACS override on a processor with no ACS feature, where it would try to separate the video cards from the PCIe root controller they're normally grouped with.

Quick hint, for those using Mac compact keyboards: "4. As it restarts press delete (if using a mac connect a usb keyboard as mac backspace will not work)" Fn-Delete on a Mac keyboard will emit a delete-forward, aka the PC Delete key.

It needs SSH to be open. I'm not sure if it needs any other ports in.

I saw above that you proposed adding an option to connect to the IP address by default. This would also benefit me, as on my Mac systems, the default connection of http://machinename does not work, since Macs expect to use the .local domain. On a side note, this process helped me to recover my unRAID server from some Docker which I accidentally mounted / to, which then changed ownership and permissions on that mount point and its contents, without having to reboot the whole server to restore the initrd. Small stupidity averted without sacrificing my precious uptime, haha. (OpenSSH will not let anyone log in if /var/empty is world accessible, among other things.)

You'll need to use Nomachine v4 or higher if you want audio over the network. VNC does not support audio.

Another alternative is to plan around having integrated graphics, but then you are limited to some Xeon E3 parts and motherboards equipped with IGP ports. I see that as the affordable solution anyway, since the E5 and E7 parts look super expensive to me.

Quick hint, in case anyone hasn't suggested it already. You can change the Clover resolution, then change the OVMF resolution from the firmware menu, and run at higher than the 800x600 default.