May 9, 201412 yr Has anyone been able to install the Intel CPU graphics on Win7 or Win8? I have it set to pass-through and Windows shows the graphics card. I then install the drivers and Windows 7 & 8 both BSOD on me. Has anyone else been more successful than myself?
May 9, 201412 yr Has anyone been able to install the Intel CPU graphics on Win7 or Win8? I have it set to pass-through and Windows shows the graphics card. I then install the drivers and Windows 7 & 8 both BSOD on me. Has anyone else been more successful than myself? You will probably have better luck with Xen 4.4 on the next unRAID beta. I don't believe iGPU pass through works with Xen 4.3.
May 9, 201412 yr sounds good! I found the same with HD4000 and HD4600. For the time being I picked up an ATI6550 which works fine. -Marcus
May 10, 201412 yr FYI, passing through intel hd graphics has been an ongoing challenge, but this isn't an integrated graphics problem. Its an Intel integrated graphics problem specifically. I don't see this being resolved in the next beta. We have been able to do successful onboard graphics pass through with the aspeed integrated graphics adapter on a supermicro server board with no problem. Sent from my Nexus 5 using Tapatalk
May 10, 201412 yr FYI, passing through intel hd graphics has been an ongoing challenge, but this isn't an integrated graphics problem. Its an Intel integrated graphics problem specifically. I don't see this being resolved in the next beta. We have been able to do successful onboard graphics pass through with the aspeed integrated graphics adapter on a supermicro server board with no problem. Sent from my Nexus 5 using Tapatalk Perhaps a nit, but the Aspeed graphics isn't actually "integrated" in the sense that it's part of the chipset (or in the Intel case actually integrated on the CPU chip) ... it's a separate graphics adapter that happens to be located on the motherboard. Yes, it's "integrated" on the same motherboard ... but I suspect it's using the PCIe bus, so the interface is essentially the same as a dedicated card.
May 10, 201412 yr FYI, passing through intel hd graphics has been an ongoing challenge, but this isn't an integrated graphics problem. Its an Intel integrated graphics problem specifically. I don't see this being resolved in the next beta. We have been able to do successful onboard graphics pass through with the aspeed integrated graphics adapter on a supermicro server board with no problem. Sent from my Nexus 5 using Tapatalk Perhaps a nit, but the Aspeed graphics isn't actually "integrated" in the sense that it's part of the chipset (or in the Intel case actually integrated on the CPU chip) ... it's a separate graphics adapter that happens to be located on the motherboard. Yes, it's "integrated" on the same motherboard ... but I suspect it's using the PCIe bus, so the interface is essentially the same as a dedicated card. Its still an igpu :-). The key difference, like you are noting, is that the a speed adapter is most likely not using the CPU or system ram like intel integrated graphics. There are other igpus out there as well that are like the aspeed igpu and work with pass through. Also, older series Intel igpus have potential to work where the HD 4000 / 4600 do not. Igpus, regardless of Intel or aspeed, both are given a PCI bus address and are passed through using the same method, we just haven't had success with the Intel iris series integrated graphics specifically. For right now, I'm not overly concerned with getting the Intel igpu pass through working. Its not a lime tech problem as it is a Xen problem right now. May be fixed in a later update if the Xen team patches it. Sent from my Nexus 5 using Tapatalk
May 10, 201412 yr FYI, passing through intel hd graphics has been an ongoing challenge, but this isn't an integrated graphics problem. Its an Intel integrated graphics problem specifically. I don't see this being resolved in the next beta. For right now, I'm not overly concerned with getting the Intel igpu pass through working. Its not a lime tech problem as it is a Xen problem right now. May be fixed in a later update if the Xen team patches it. I'm saddened to hear this. I hope a solution presents itself soon. I really don't want to purchase another video card for the interim period while this is resolved. My end goal is to use the iGPU (HD4600) in my Windows 7 Virtual Machine. Fingers crossed for resolution
May 10, 201412 yr It would certainly be nice if this is resolved. But it's also no big deal as long as you're aware of it up front. If I was going to build a "v6 box", I'd use a C series chipset with a fairly high-end Xeon E3 series CPU (probably with an integrated GPU, which I'd just disable); and then use an add-in video card. Pop in 16GB of ECC RAM and it would be a very nice v6 box that would support a Win7 VM with the video card passed through and a couple of other VM's to run any add-ons I wanted to use. If the pass-through issue was resolved in a later update, I'd just remove the video card and pass the Intel HD GPU through.
May 10, 201412 yr We've gotten some pretty cheap amd cards to work with the pass through. The integrated graphics pass through not working on the hd4000/4600 was something that significantly disappointed me as well, but there was only so much time we could spend trying to figure it out before we had to move on to other things. We may have another way to circumvent this in a later release, but for now, don't plan on seeing this for 6.0. Sent from my Nexus 5 using Tapatalk
May 10, 201412 yr Author Fair enough, thanks for letting me know. I am planning on purchasing a new CPU in the near future and it's helpfully to know that I cannot count on the integrated graphics to be passed through. I do have a extra Radeon HD 4350 and it looked like other XEN users from different forums have gotten that to work.
May 10, 201412 yr Fair enough, thanks for letting me know. I am planning on purchasing a new CPU in the near future and it's helpfully to know that I cannot count on the integrated graphics to be passed through. I do have a extra Radeon HD 4350 and it looked like other XEN users from different forums have gotten that to work. I'd be shocked if you couldn't get it to work. We've used cards even older than that and succeeded. In fact, the older / lower end cards are easier and more reliable than the newer ones. Sent from my Nexus 5 using Tapatalk
May 11, 201412 yr Something to remember is that if you do passthrough the iGPU to a VM then you no longer have access to it for the host. This means telnet/ssh (or ipmi if you have it) will be the only ways to access the host. You won't be able to just connect a monitor to the server to interact with the host. Sent from my Nexus 4 using Tapatalk
May 11, 201412 yr Something to remember is that if you do passthrough the iGPU to a VM then you no longer have access to it for the host. This means telnet/ssh (or ipmi if you have it) will be the only ways to access the host. You won't be able to just connect a monitor to the server to interact with the host. Sent from my Nexus 4 using Tapatalk This is true but we actually found a way to potentially give the video back to the host after shutting down the VM. Its only a concept right now, but definitely plausible... Sent from my Nexus 5 using Tapatalk
May 11, 201412 yr Something to remember is that if you do passthrough the iGPU to a VM then you no longer have access to it for the host. This means telnet/ssh (or ipmi if you have it) will be the only ways to access the host. You won't be able to just connect a monitor to the server to interact with the host. FWIW, in my case, I have a really, really cheap (crappy) video card installed in addition to the iGPU, to give to unRAID. The iGPU in my case is the only card that can provide sufficient power/performance to my Windows VM.
May 11, 201412 yr FYI, passing through intel hd graphics has been an ongoing challenge, but this isn't an integrated graphics problem. Its an Intel integrated graphics problem specifically. I don't see this being resolved in the next beta. For right now, I'm not overly concerned with getting the Intel igpu pass through working. Its not a lime tech problem as it is a Xen problem right now. May be fixed in a later update if the Xen team patches it. I'm saddened to hear this. I hope a solution presents itself soon. I really don't want to purchase another video card for the interim period while this is resolved. My end goal is to use the iGPU (HD4600) in my Windows 7 Virtual Machine. Fingers crossed for resolution Same here, really do hope some sort of fix is found in the future. Sent from my SM-N9005 using Tapatalk
May 11, 201412 yr FYI, passing through intel hd graphics has been an ongoing challenge, but this isn't an integrated graphics problem. Its an Intel integrated graphics problem specifically. I don't see this being resolved in the next beta. For right now, I'm not overly concerned with getting the Intel igpu pass through working. Its not a lime tech problem as it is a Xen problem right now. May be fixed in a later update if the Xen team patches it. I'm saddened to hear this. I hope a solution presents itself soon. I really don't want to purchase another video card for the interim period while this is resolved. My end goal is to use the iGPU (HD4600) in my Windows 7 Virtual Machine. Fingers crossed for resolution Same here, really do hope some sort of fix is found in the future. Sent from my SM-N9005 using Tapatalk Not a matter of if as much as when and how. There may be more than one way to address this, and we are experimenting further. Trust me when I say this is definitely something we want to try and solve. Just don't expect this necessarily for 6.0. Sent from my Nexus 5 using Tapatalk
February 16, 201511 yr I was wondering if there has been any new developments with regard to supporting Intel HD 4000 graphics passthrough in Xen. I have had similar issues as earlier posters on this thread. Initially installing Windows and using the IGD as a secondary graphics device (gfx_passthru=0), I can boot Windows 7 and interact with it through VNC. (I still don't see anything on my main display) The graphics device shows up in "Device Manager" as a generic graphics device but with issues. Specifically, I get the error "This device cannot start. (Code 10)" After I install the Win7 64-bit driver for this specific graphics card and reboot the system, I get the BSOD. After I reboot again, and enter "Windows Startup Repair", Windows fails to repair the system and reports problems with a bad driver. Under normal circumstances, the Linux kernel uses the driver "i915" to manage this device. To prevent the Dom0 kernel from touching the device during startup, I pass it specific command-line arguments in GRUB. Namely: blacklist=i915 xen-pciback.hide=(00:02.0) However, even though you can keep Linux from touching the graphics device, you still can't prevent the host firmware from accessing the graphics device during boot. I am guessing that this is not an issue since problems only seem to occur with intel IGDs and not other IGDs. I am still wondering if the firmware passes any information to GRUB and the OS that needs to be passed on to Xen.
February 16, 201511 yr I was wondering if there has been any new developments with regard to supporting Intel HD 4000 graphics passthrough in Xen. I have had similar issues as earlier posters on this thread. Initially installing Windows and using the IGD as a secondary graphics device (gfx_passthru=0), I can boot Windows 7 and interact with it through VNC. (I still don't see anything on my main display) The graphics device shows up in "Device Manager" as a generic graphics device but with issues. Specifically, I get the error "This device cannot start. (Code 10)" After I install the Win7 64-bit driver for this specific graphics card and reboot the system, I get the BSOD. After I reboot again, and enter "Windows Startup Repair", Windows fails to repair the system and reports problems with a bad driver. Under normal circumstances, the Linux kernel uses the driver "i915" to manage this device. To prevent the Dom0 kernel from touching the device during startup, I pass it specific command-line arguments in GRUB. Namely: blacklist=i915 xen-pciback.hide=(00:02.0) However, even though you can keep Linux from touching the graphics device, you still can't prevent the host firmware from accessing the graphics device during boot. I am guessing that this is not an issue since problems only seem to occur with intel IGDs and not other IGDs. I am still wondering if the firmware passes any information to GRUB and the OS that needs to be passed on to Xen. IGD pass through is a very complicated task. It is not like a traditional PCI device, and the man who authored the VFIO driver has even said, repeatedly, that IGD pass through is not something most folks can achieve. I realize that VFIO doesn't apply to Xen, but this statement holds true for IGD pass through irrespective of Xen or KVM. That said, there is work on a few projects (KVMgt for one) that aim to improve this, but there is still much to be seen on this front. All I can say is that IGD pass through is something we are paying close attention to and will try to support as best as the upstream technology can do, but this isn't something we are planning to see in version 6.0.
Archived
This topic is now archived and is closed to further replies.