<div dir="ltr"><div class="gmail_extra"><div class="gmail_quote"><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div dir="ltr"><span class="im"></span> First off basically you want your GPUs directly connected to the CPU for best performance and not over the chipset which only has the equivalent of 4 PCIe-Lanes shared for everything you usually connect things to (USB, Sound, PCI, small PCIex1s). Secondly most Xeon E3/Core i have integrated graphics which are a) more powerful than a Geforce 210 and b) sport really good (libre) graphics drivers.<br> Adding to the first point the difference between Intels small sockets (11xx) and big sockets (2011) ist the number of available PCIe-Lanes.<br> The first only has 16 PCIe-Lanes from the CPU which are split in up to 3 slots (and this only with higher end boards) and all other periphery/the chipset uses DMI for CPU-connection (4 PCIe-Lanes equivalent). PCIe-ACS ist not supported on the CPU port, where you put your GPU and the chipset is basically a field of quirks and luck I think (as an enduser) [and since you only have 4PCIe-Lanes (16GBit/s if I remember correctly) for all your periphery you don't really want to put your GPU there].<br> The second has 28 to 40 Lanes from the CPU which are supported with ACS, so that the different physical slots get into separate IOMMU-groups and you can arbitrarily choose which addin-card you want to passtrough to the VM. For the chipset (onboard Sound, USB, SATA) the same things hold true as for the small platform.</div></blockquote><div><br></div><div>That makes perfect sense, thank you, Felix!<br> <br></div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><span class="im"><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div dir="ltr"><div><div><div><div class="gmail_extra"><div class="gmail_quote"><span><div><br></div></span><div>I am not sure I understand this correctly. Does this mean QEMU assigns a UEFI ROM of another 750 Ti on my card when it launches the VM? </div></div></div></div></div></div></div></blockquote><div><br></div></span><div>Yes, it's just like flashing but without the need to flash your hardware.<br></div><div>The libvirt-XML line within the <hostdev> part would be:<br><rom file='path-to-your.rom'/><br></div><div>Techpowerup has a lot of bios versions of GPUs [1], and for your card there seem to be two versions which, according to AW's script both have a valid UEFI [2].<br></div><span class="im"><div> </div><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div dir="ltr"><div><div><div><div class="gmail_extra"><div class="gmail_quote"><div>If so, can it cause any problems because it's not the original ROM or because of the implementation in QEMU?<br></div></div></div></div></div></div></div></blockquote><div> <br></div></span><div>The unlikely danger of breaking the GPU's BIOS in the process of flashing does of course not exist with the romfile-method. But all other dangers of flashing the "wrong" bios do remain, like dangerous voltages, wrong fan speeds that result in overheating etc. But: even among different manufacturers with the same GPU model, this danger is quite low if you watch for the clock and memory frequencies. With your model there should to be even less to worry, as there are bios versions for your exact card. With my older AMD card I blindly tested different BIOS versions of other manufacturers till I found one that was stable. No permanent damages occured, just temporary instabilities.<br></div><span class="im"><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><div dir="ltr"><div><div><div><div class="gmail_extra"><div class="gmail_quote"><span><blockquote class="gmail_quote" style="margin:0px 0px 0px 0.8ex;border-left:1px solid rgb(204,204,204);padding-left:1ex"><span></span> <p dir="ltr">You can also add an emulated GPU and install the guest drivers which eliminate the need of GOP, you'll just have to boot the guest without video output until the driver loads.</p></blockquote></span><div>Is that the same as the romfile method?<br></div></div></div></div></div></div></div></blockquote><div> </div></span><div>No, you simply boot your virtual UEFI system without a compatible UEFI card, so there's no output from it. As soon as Windows and the GPU drivers are loaded, it doesn't matter, you just have no boot information (and miss blue screens or problems when you try to upgrade Windows - then you just have to replace the GPU with an emulated one for the length of this procedure)<br><br><br></div><div>I'm one of the 750Ti-owners that have a working configuration. My KFA²-model luckily came with an UEFI rom, but I believe with the available roms on techpowerup you should get it running quite well, too. I'd recommend using qemu-git and linux-vfio-lts from the Arch AUR (commenting the patches you don't need) because this mostly "just works" and comes with current versions needed to cheat the nvidia driver running on a performant virtual machine.<br><br></div><br>[1] <a href="https://www.techpowerup.com/vgabios/index.php" target="_blank">https://www.techpowerup.com/vgabios/index.php</a><br>[2] <a href="https://www.techpowerup.com/vgabios/index.php?architecture=NVIDIA&manufacturer=Palit&model=GTX+750+Ti&interface=&memType=&memSize=" target="_blank">https://www.techpowerup.com/vgabios/index.php?architecture=NVIDIA&manufacturer=Palit&model=GTX+750+Ti&interface=&memType=&memSize=</a></blockquote><div><br></div><div>Wow, then the rom file option is what I am looking for. Actually I had seen those BIOSes, but I thought they would not work because the site says "UEFI Supported: No". I am really glad that I don't have to buy a new graphics card. :-D<br></div><div>Thanks a lot for the advice and explanation, Marcel!<br><br><br></div><div>I think I know all I need to know now and I have a good feeling going into this. Thank you again for your help and support everybody! I will probably be back in a few weeks asking for more guidance. :-P<br><br></div><div>So long,<br><br></div><div>Jens<br></div></div><br></div></div>