Today I am happy, NVIDIA vgpu for XenApp on XenServer to announce.
Citrix supported GPU sharing for XenApp for a long time now, for NVIDIA (and other vendors) GPUs on vSphere, XenServer and physical servers. This support complements the existing GPU-sharing for XenApp feature is based on GPU pass-through; wherein a GPU is passed through a XenApp Windows Server VM, done sharing by multiple users in the RDS-shift meetings and access apps VM is running on the server.
- Citrix Technology Professional (CTP), Dane Young "Adding vgpu support for XenApp workloads will help GPU acceleration for all users and to allow meetings on by the costs continue driving if it for the workload. the number of Windows applications benefit from basic GPU resources continues to grow. As such, GPU and vgpu should be part of every conversation to Windows applications and desktops, whether it is hosted client or server NVIDIA vgpu for XenApp makes this even more cost effective and a reality for all environments or budgets "
- CTP, Alexander Ervik Johnsen:. ". vgpu support thereafter to achieve as an industry standard for other providers. XenApp is the most widely used Citrix's product, so bring vgpu support in this will certainly speed up the introduction and help customers more applications deliver to end users"
This new support provides an additional option for those looking for an easier image management looking smaller XenApp server with a smaller number of user sessions in a XenApp server VM running to the controlled administration to 3D to support graphic software titles and licenses. This new feature allows a XenApp server VM part of a GPU NVIDIA GRID access vgpu technology.
Our standard recommendations for XenApp GPU sharing remain
For the majority of users, we put the long standard existing XenApp GPU sharing feature recommend. Our general advice for GPU-sharing and XenApp is and remains, that the user has a GPU with a specification at least nearly as powerful as an NVIDIA K5000, if not should use an NVIDIA K00 GPU pass-through. This is based on a few facts:
- The average GPU from a SolidWorks CAD users is used, is a K00, if you are using applications that use the GPU, or CAD, you really need something more if you want to plan it to share user density per server with some applications able to support
- users often 20-30 users on such a map (see here for some sample densities)
- high- maximize end CAD Designer on a single workstation often use a dedicated K5000 or K00 for applications such as Dassault CATIA or NX.
You can learn more about the certified and GPU as the K00 GPU for Read Pass-Through and vgpu, here.
When can I vgpu for XenApp?
Well .... If you have a vgpu enabled version of XenDesktop / XenApp 7.5 or higher, the vgpu support for VDI, it is already working (although please prerequisite for vgpu NVIDIA drivers Note below). This communication is only on formal support. The feature has been tested has always QA (Quality Assurance) and is technically the same as Citrix vgpu VDI support for Window Server operating systems.
This is one of the advantages of unifying XenApp and XenDesktop in a single architecture, features of XenDesktop can be easily transferred in XenApp. In fact, we were recently USB redirection support in XenApp 7.6 can introduce as a result of this change. We constantly are the softest and broadest continuum of GPU, graphical and HDX offers solutions.
We had simply chosen not to be able to formally announce support.
continue to blur the boundaries and mix What vgpu means for XenApp technically mean how it differs from XenApp GPU sharing?
Instead of the Windows Server VM provide an entire GPU pass-through, the XenApp server VM will get a portion of the physical GPU vgpu. Add The RDS-layer on a second sharing layer, by seeing that vgpu (part of a physical) vgpu instead of enforced GPU. Both vgpu and pass-through allows the use direct hardware sharing the GPU via the manufacturer driver. Neither method includes API not intercept the expense and liability of synthetic drivers. to find more information about vgpu and GPU pass-through in these overviews:
- HP Reference Architecture
- The virtualization matrix graphic analysis
- @TeamRGE Graphics Virtual desktops SmackDown! White Paper
- Cisco Reference Architecture
vgpu NVIDIA technology is currently only available for Citrix XenServer. GPU sharing for Citrix XenApp is for NVIDIA and GPUs from other manufacturers; and on other platforms, for example, , Physical and vSphere / ESXi hypervisor from VMware, which supports GPU pass-through (VDGA)
If you have questions about which technology sharing GPU use - please to you questions either the Citrix forums, here; or the NVIDIA forums, here; where you. Your plans with other users, and our support and technical staff
application compatibility
discuss, the application and the server VM will find it both a "GPU" they do not know if it is more a vgpu than one GPU to pass-through. Both vgpu for XenApp and GPU pass-through (GPU sharing) based on direct access to the hardware on the NVIDIA native drivers. participates intercept No software emulation or API. As such, the only limitation is the application compatibility is for both RDS with XenApp vgpu and GPU pass-through. As such, application compatibility from manufacturers about the Citrix Ready program for XenApp certified GPU sharing application, on the Citrix Ready marketplace find many applications here.
NVIDIA driver
Please be sure to use the proper NVIDIA driver. For XenApp GPU sharing on NVIDIA GPUs, you should use the GPU pass-through driver. For XenApp vgpu you should use the NVIDIA vgpu driver. GPU pass-through driver (when this article was written) to support experimentally CUDA and OpenCL during vgpu is not seen here. You should check the current status at the time of reading.
So it works, tested and you did not tell us?
Well, I'm afraid to say yes. As a product manager part dictating my work best practices and always avoid in a jumble that user to help. I've seen a lot of confused users and some half-baked PoC with XenApp. At first I felt support for this feature. I could encourage users to use it and not the best technical option. vgpu has a small overhead (some%) over GPU pass-through. My first was rational that users additional meetings and users to add VM on a server, should increase user density on a GPU instead dismember the GPU the number of users on the server VM to match.
In particular, I have a lot of users with PoC make mistakes seen to recognize, in the absence of such a K1 card has four GPUs is roughly equivalent to a K0. It really is for the support PowerPoint, Windows Aero and not 3D or CAD applications. If you compare the specifications of a K0 compared to a K00, you will recognize this type of card that only rarely serious GPU loads on one physical workstation (let alone an option for sharing between multiple 3D users on XenApp) used.
why you have changed your mind?
I our customers simply underestimated, understand a lot of you the technicalities of the problem and presents very good scenarios where vgpu and XenApp sense. Formerly an engineer, I focused on the technical advantages of increasing user density on increasing density session (to avoid that tiny% overhead of vgpu vs. GPU pass-through) without fully as the cost of managing for softer factors make.
As a product manager you have to be extremely receptive to the fact you understand the needs of your customers wrong. And as such, we are committed regularly via our feedback forums and visiting customers. It was this regular touch base opportunities that my prejudices and assumptions were quickly corrected. A user describes an application scenario in an educational scenario (University) where vgpu made for XenApp perfect sense so:
- Rachel
I agree that a second virtualization layer introduction necessarily not make sense ... when you approach it from an image management perspective. Suppose we have 6 x Dell R720 (256 GB / 2 x E5-2680v2 / 1 x Grid K2), which are used by several graphics-intensive applications engineering department in our university used to support. was to have my plan 1 x XenApp server pinned to a wall outlet per R720, (directly connected to K2) with the K2 pgpu via by ~~ POS = TRUNC. That collection of 6 x XenApp server would most CAD / CAM applications from our students students (SolidWorks, NX, Creo, Catia) used, for example, are used. My first thoughts were 2 x XenApp servers per R720 each have share the other jack and use a K260Q vgpu. That would an additional 12 x XenApp server that I can (...... engineering, aerospace engineering, civil engineering) allocate between the various departments serve me applications that are used by a limited group of users.
For example, the number of students at the same AGI Satellite Tracking used to justify not necessarily mean that I have two full K2 pGPU devoted (two XenApp server) or that we use 16 x Win7 VM.
I could of course the main pool of XenApp servers install STK inside. But there are some said application installed not too many heavy engineering in the same Windows server image.
At the end of the day I'm glad to have options (XenApp + pass-through, vgpu + VDI) and a range of hardware (mixture of R720 + K2 and older R720 + K1 ), is our biggest challenge must learn to how to best use them in order based requires any technology and our environment properly size semester.
Thank
last thought
I'm still a little worried , this support could users to cause the wrong solution. But in this case, I think I got it wrong. The need, a fully supported solution for our existing users, which overrides the limitations understand my caution. One of the nice things about Citrix is that we encouraged to make positive these U-turns and that a well will be read and considered as a user forum post.
If you have any doubt, as I envisage best solution for the majority of users the time-tested GPU sharing XenApp provides a proven history for many years remain - to read the IMSCAD case studies to more the success of AutoCAD find out with many published case studies of XenApp use. If you have any doubts, please call either ask questions on the Citrix forums, here; or the NVIDIA forums, here; where you. Your plans with other users, and our support and technical staff
discuss
0 Komentar