The official version of Windows 10 is about to be released this month, and the next version of Windows Server is still in the tech preview, so what's new in the next version of RDS?
Here I actually pay more attention to the change of Vgpu, because in the architectural design industry, Virtual desktop has a lot of non-verbal advantages, but for designers to really on the virtual desktop design, then the graphics card support is very important.
Let's take a look at the main changes in RDS for next-generation Windows Server, and then vgpu this topic.
Under the premise that both the virtual host and the guest host are the next version of Windows Server and WINDOWS10, the addition of RemoteFX supports OpenGL 4.4 and OpenGL 1.1 APIs such as Adobe Photoshop, Maya and other design software. Of course, DirectX 11.0, WDDM 1.2, etc. are supported;
RemoteFX supported dedicated memory (dedicated settings 1g+ shared memory 1G, up to 2G), the size of the previous memory is dynamically allocated, mainly based on the number of your monitor, the next version of this relationship is deleted.
Performance improvements: Primarily for VM host transport performance VMBus stability and application compatibility bug fixes, such as:
Adobe Photoshop CC
Unigine Heaven Benchmark (DX and OpenGL Benchmark)
Unigine Tropics (DX and OpenGL Benchmark)
Blender (OpenGL Open source 3D Animation Suite)
Google Earth (DX and OpenGL)
Specviewperf (OpenGL Benchmark)
Glview (OpenGL API Extension Viewer with render tests)
Lsprepost (OpenGL post processing application)
Lightwave 3D (OpenGL application)
OK, that's roughly the case, then I'm most concerned about the dedicated video memory this piece.
First talk about Citrix, after all, is the desktop virtualization in the Big Brother, the main use is VGPU and through the transmission (through), Citrix and Nvidia cooperation is indeed very cow, in the 3-dimensional design industry can emerge, see the principle
With GRID? VGPU? Technology, graphical commands for each virtual machine are passed directly to the GPU without Hypervisor decoding. This allows GPU hardware to achieve time fragmentation (time-sliced) to provide top-level shared virtualization graphics performance.
650) this.width=650; "height=" 422 "title=" clip_image001 "style=" border:0px;padding-top:0px;padding-right:0px; Padding-left:0px;background-image:none, "alt=" clip_image001 "src=" http://s3.51cto.com/wyfs02/M00/6F/CB/ Wkiol1wovf6slzzuaahlhcgu0jy573.jpg "border=" 0 "/>
VGPU Manager allows up to eight users to share a single physical GPU, allocating the available GPU graphics resources to the virtual machine in a balanced way. Each NVIDIA? The GRID K1 card has up to four GPUs, allowing 32 users to share a single video card.
650) this.width=650; "height=" 153 "title=" image "style=" margin:0px;border:0px;padding-top:0px;padding-right:0px; Padding-left:0px;background-image:none; "alt=" image "src=" http://s3.51cto.com/wyfs02/M01/6F/CB/ Wkiol1wovf6dsxcsaadpgw69b34756.jpg "border=" 0 "/>
If using Passthrough mode, then a graphics card 2 core can support 2 3D virtual desktop, dedicated video memory can reach 4G.
And Microsoft's next-generation Windows Server:
In my opinion, I think I can not jump hypervisor layer, may be Microsoft and graphics card manufacturers in-depth research and cooperation did not get a better combination, it may be technical preview version so do not speculate. For the time being, it is not possible to pass through (through), or to achieve the dedicated memory on the hypervisor layer. But that's a step forward. (Citrix is done with dedicated memory Vgpu allocation cap 2G)
650) this.width=650; "height=" 364 "title=" GPU Virtual technology "style=" border:0px;padding-top:0px;padding-right:0px; Padding-left:0px;background-image:none; "alt=" GPU Virtual technology "src=" http://s3.51cto.com/wyfs02/M01/6F/CE/ Wkiom1wovckj-6vyaag2lhvc1oc143.jpg "border=" 0 "/>
What about the settings for dedicated memory for the next generation of Windows Server? Look at the picture
650) this.width=650; "height=" 364 "title=" Storm 20157179079703 "style=" border:0px;padding-top:0px;padding-right:0px; Padding-left:0px;background-image:none, "alt=" Storm 20157179079703 "src=" http://s3.51cto.com/wyfs02/M02/6F/CE/ Wkiom1wovcptwhlzaaeaff3kdr0435.jpg "border=" 0 "/>
650) this.width=650; "height=" 364 "title=" Storm 20157179089187 "style=" border:0px;padding-top:0px;padding-right:0px; Padding-left:0px;background-image:none, "alt=" Storm 20157179089187 "src=" http://s3.51cto.com/wyfs02/M00/6F/CE/ Wkiom1wovcobdnwjaaebsey9ly0116.jpg "border=" 0 "/>
So let's look at the effect of OpenGL support:
650) this.width=650; "height=" 364 "title=" Storm 20157179108765 "style=" border:0px;padding-top:0px;padding-right:0px; Padding-left:0px;background-image:none, "alt=" Storm 20157179108765 "src=" http://s3.51cto.com/wyfs02/M01/6F/CE/ Wkiom1wovcsii1t1aaheqq5yohy298.jpg "border=" 0 "/>
I also agree with a friend's point of view, "Microsoft in the pure technical level, has always been PK but rivals, Microsoft's advantage is the application-oriented, can integrate his multiple products to form a solution."
Well, the next generation of Windows Server RDS will share these, and what's next, I'll continue to share with you.
This article is from the "Zjs Microsoft Desktop Virtualization" blog, reproduced please contact the author!
Talking about RDS of next generation Windows Server