<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 TRANSITIONAL//EN">
<HTML>
<HEAD>
<META HTTP-EQUIV="Content-Type" CONTENT="text/html; CHARSET=UTF-8">
<META NAME="GENERATOR" CONTENT="GtkHTML/4.6.6">
</HEAD>
<BODY>
Hi,<BR>
<BR>
Currently we've only been using oVirt for server virtualization. Using it as a platform for Virtual Desktop Infrastructure is very tempting for us, so I've been reading up about it.<BR>
<BR>
Are there any plans to support Intel GVT-s or GVT-g in the future? Xen already has it already supported and KVM is supposed to be working on it, but there will have to be some modifications in oVirt too I guess.<BR>
<BR>
<A HREF="https://01.org/blogs/skjain/2014/intel%C2%AE-graphics-virtualization-update">https://01.org/blogs/skjain/2014/intel%C2%AE-graphics-virtualization-update</A><BR>
<BR>
Qemu supports GVT-d/VTd already but I don't know if I can use it in oVirt... it doesn't really scale however<BR>
<BR>
<A HREF="http://www.phoronix.com/scan.php?page=news_item&px=MTY4MTc">http://www.phoronix.com/scan.php?page=news_item&px=MTY4MTc</A><BR>
<BR>
<BLOCKQUOTE>
<B><I>Intel GVT-d</I></B><I> for direct GPU access whereby the guest virtual machine has full access to the graphics processor. The guest operating system's drivers are used and there's no limitations or interference by the hypervisor. Intel GVT-d works with QEMU through its VTd support. </I><BR>
<BR>
<B><I>Intel GVT-s</I></B><I> as graphics virtualization at the API level to have one graphics processor exposed to potentially multiple virtual machines. Intel GVT-s is done using an API forwarding technique that interfaces with the graphics hardware. It appears though Intel hasn't done much in this realm for open-source Linux but they mention "many commercial desktop and workstation remoting products in the market use this approach." VirtualBox and VMware are some notable examples. Within the open-source space, <A HREF="http://www.phoronix.com/scan.php?page=search&q=Virgil3D">Virgil3D</A> is aiming for similar functionality. </I><BR>
<BR>
<B><I>Intel GVT-g</I></B><I> as one GPU shared to many virtual machines by exposing a virtual GPU. Each virtual desktop is running Intel's native graphics driver and is part of their XenGT approach. "On a time sliced basis, an agent in the hypervisor directly assigns the full GPU resource to each virtual machine. Thus, during its time slice, while the virtual machine gets a full dedicated GPU, from overall system view point several virtual machines share a single GPU. Intel has been developing GVT-g under the code name 'XenGT' for Xen. Up-streaming of GVT-g to KVM is also in works. More recently, Intel has been disclosing this solution to select partners, and making the source available for variety of processor graphics configurations." </I><BR>
<BR>
<I>Sunil ended his Intel Open-Source Technology Center blog post with, "Major ISVs and OEMs are aligning with Intel to productize Intel GVT based solutions. Open source developers are finding Intel GVT portfolio with Intel processor-graphics products equally enticing. It will be interesting to see some cool innovations emerge from graphics virtualization." Those interested can find more information at <A HREF="https://01.org/blogs/skjain/2014/intel%C2%AE-graphics-virtualization-update">01.org</A>.</I><BR>
<BR>
</BLOCKQUOTE>
Kind regards,<BR>
<BR>
Jorick Astrego
</BODY>
</HTML>