I assume that many of you with exictement have read through all news in vSphere 5.1
What cought my eye in View point of view was this:
Improved 3D Graphics Support – (View Only) – hardware acceleration with possiblity to leverage NVIDIA’s hardware cards installed in ESXi server, where those graphics cards are virtualized and used in View Desktops. It’s targeted for graphics intensive workloads, CAD designers, medical imaging etc…
NVIDIA Quadro 4000/5000/6000 and NVIDIA Tesla M2070Q are supported graphics cards. Note that the ESXi Image profile must be installed with the NVIDIA GPU VIB file."
It certainly raises some questions for me:
a) The Nvidia GPU VIB-files is that free of charge or is it licensed from Nvidia or Vmware?
b) Have anyone tested 5.1 with Quadro-cards (and VMware Techie out there?) Pros? Cons?
c) Is the GPU fully accessible from the View guest or is there some kind of abstraction layer in between? After some research it seems like the abstraction layer is utilizing Xorg in a clever way.
I have a Dell R720 with a K1 will try a K2 later, right now im doing a couple of tests
Performace an stability seems to be lacking reminds me of my times of PC Gamer.
I've applied the registry change to remove the devTap cap but i think PCoIP is the issue now...
PCoIP crashed and my session got disconnected when using 3D Mark06
Using the vsphere console it runs and finishes with a score of 5799
Windows Movie Maker 2012 does not run apparently its an issue with the DDI Version of 9Ex, fortunately 2011 does work...
Well, so far.... I'm out of luck. At least I've found information about HOW to enable vSGA...
Unfortunatly it grinds to a halt at the first bullet
Got it to recognize FX5800 on HP DL370G6. FYI: the NVidia cards are detected and drivers installed during ESXi 5.1 install - no download of NVIDIA VIB required.
On the server with the nvidia card installed:
1. In vSphere, highlight host, click Configuration tab, click Advanced Settings.
2. Click either "Configure Passthrough..." or "Edit...", whichever is there.
3. Scroll down list of components looking for "nvidia". Mine was at the very bottom of the list. Check it ON.
4. Restart the ESXi Host.
5. For each VM that will use the nvidia card, go into the VM Settings and add the PCI Device from list. This is the nvidia card.
Hope this helps.
Does anyone have access to the nVidia VIB needed for this? I have a server all set with a Quadro 4000 ready to try this but the card is not being claimed after a reboot and xorg isn't starting. I noticed the KB article for this has been pulled...frustrating to say the least.
FWIW (last poster) I don't believe we're looking to do DirectPath, which is what you've descibed. That allows only a single VM to "claim" the GPU. What you described is the way to get the video card set up for DirectPath.
I am using Dell T3500 workstation for testing conf as follows
when I bypass Quadro 4000 GPU and reboot the esx 5.1. It comes up to cnic_register loaded successfull and hungs.
Kindly give suggest what is the problem how to over come this.
Thanks in advance
Are you sure this is supported?? I'm assuing you are trying to do some high graphical applications on your virtual desktop? The only supported method I am aware of is to use NVidia VGX graphics card. http://www.nvidia.com/object/vgx-hypervisor.html
It's supposed to be supported for a very limited subset of GPUs and VMware software. The new VMware/nVidia piece is called vSGA (yes, I know about VGX). It was announced with 5.1 at VMworld, but there's been very little information forthcoming. A VMware KB article was briefly published and then withdrawn, etc.
vSGA is the term that VMWare uses for their own technology. I found this today:
"This is all made possible though the joint collaboration and support of our partner NVIDIA. We have been working closely with NVIDIA to enable this capability across their Quadro and VGX lines of graphics adapters. This also would not be possible without the support of NVIDIA’s native driver for VMware ESX working in tandem to combine the power of the market-leading virtualization platform with the power of the leader in computer graphics."
From the blog post ti says that Quadro and VGX cards should work eventually (or even now).
All Thanks for all your response,
Actually I had tried with vSGA and its working fine, but I want to achive GPU by pass so that I can have dedicated GPU for individual VM and there will be robust performance.
I have doubt we wanted to go for dell R720 with Tesla 2090 with esx 5.1 and view 5.1 will it work?
This configuration is required because we will be running graphic hungry application like Auto CAD and Revit. Kindly give all your inputs.
It seems somebody needs to evaluate all this speculation!
I'm working on getting a Dell 720 with NVIDIA TESLA M2075 for a lab environment with simulated variable WAN to test dual monitor workstations for running AutoCAD Design Suite 2013-- specifically Inventor. On top of this we need to support USB 3D Connexion Spaceballs at the workstations. Will post my findings.
You were able to source the nVidia VIB? Would you be willing to shoot me a copy? I'm dying to try this.
FWIW, there's another thread here in the communities that I've contributed to...search on DirectPath and ATI, you'll find it. I've been very successful in using DirectPath and Quadro cards...ultimately the key was using both the PCIhole.start and PCIhole.end variables in the guest configuration, in conjunction with the standard DirectPath requirements like memory reservations.
I’m not 100% sure the VIB is needed. When I was working with the Quadro card, it was detected, I just had no luck passing it thru in my setup. I was unable to find them on NVIDA’s site, just checked recently too. Still working on getting the VM software together… Maybe I’m jumping the gun for VMWare, but I do have a Citrix configuration together to test—but we aren’t talking about that.
My card (Quadro 4000) is detected during PCI enumeration and I can successfully do DirectPath with it. But from the few snippets I've been able to gather about vSGA it relys on an nVidia driver (delievered via the VIB) that basically allows xorg to run. Right now if I try to start xorg it fails, period.
This is very frustrating given that VMware/nVidia announced this almost 2 months ago and there's basically nothing of any value out there about implementing it. I just want to see the KB and the VIB.
I'd guess that you may have your host misconfigured. The T3500 workstation may not even be a good candidate for this...I've gotten it to work on an HP Z800 and several models of HP DL servers (all with Quadro 3800/2000/4000) but haven't tried Dell. The T3500 is EOL and may not have a BIOS that's compatible and/or processors that support VT-D (on the other hand that may not be an issue...I don't know Dell well enough to tell you).
That said...you *MUST* have more than one video card in the box. The one you're trying to pass through is going to be claimed by and dedicated to the guest VM you're trying to tie it to. But your host needs a video card too...I think that might be why you're seeing the error you're getting, you've taken away the host's video card by dedicating it to the VM.
You need to keep in mind that doing DirectPath with video cards isn't supported. At this point it's 100% experimental and those of us who've gotten it to work have done so by taking our time, reading logs and trying different things. No one is going to hand you a solution to this...that's not because people don't like you or don't want to be helpful, it's because there really aren't any definitive/straightforward answers. I happened to see you were rather rude about this in another thread and I debated whether I was even going to answer your plea here.
I'd strongly suggest that you go back and read the requirements for DirectPath (for DirectPath of anything)...things like processor features, BIOS levels, RAM reservations, etc. etc. etc. Search the DirectPath/ATI thread for posts I made regarding Quadro cards (yes, they do work despite what some of the folks there posted recently). You should also remember that this is not suitable in any way for a production environment as it is NOT supported.
Does anyone know how this compares to RemoteFX? This sounds very similar to RemoteFX since it appears to carve up the GPU in a similar way (software-based API intercept method of GPU virtualization). After some research it appears to limit the VM's GPU vRAM to 128 MB, however there are some sources that hint this may be increased to 512 MB. From this link http://www.virtuallygeeky.com/2012/10/virtualized-3d-gaming-on-vmware-view.html (in the last few Barcelona videos) it appears that the performance is better than RemoteFX even though it seems to use a similar "abstraction layer" in between the VM and the physical GPU. I have never been able to get this kind of performance using RemoteFX and a Fermi Quadro GPU so I'm curious if VMware is doing things better/different? Any thoughts?
Also this doesnt appear, on the first pass at least, to be as capable as Citrix's "yet to be released" solution using nVidia's new Kepler based VGX boards => http://www.nvidia.com/object/vgx-partners.html. It looks like nVidia's new GPUs are needed to fully utilize the latest tech that nVidia showcased back in May at GTC 2012. Since VMware's vSGA will also work with the older Fermi Quadro cards its leading me to believe (which at this point I'm wanting to be wrong about this) that its not as capable even when using it with the VGX boards. What do you guys think?