Splitting up a nVidia GRID K2's GPUs?

[LMUISHuntsville]

What I'm talking about is designating one GPU for vSGA and pass-through the second GPU to a VM for high performance 3D work.  In theory this should work, but I'm curious to know if anyone has actually done this or could contribute to this line of thinking.  Right now I'm not chomping at the bit to buy a 3K+ card to test this theory out.

Regards,

Nathan

[mikeleahy1234]

mine says 80.4.60.0.21

[LMUISHuntsville]

Thats interesting..  what nvidia driver version are you using?  I've tried 320 and 311.5; both exhibit the screen blanking issue when connecting through view.  I havent checked for any VMware patches lately...

[mikeleahy1234]

320

[LMUISHuntsville]

When you bring up your nVidia Control Panel what does your "View System Topology" look like?  Are there more display connection points shown on yours vs my one "VGA" connection?

[mikeleahy1234]

looks the same as yours

[LMUISHuntsville]

Oh ok; I know you are only using one monitor right now, but have you tried doing what you mentioned earlier (about going to screen resolution - detect and two more monitors will appear....)?  Meaning that while you are currently using the K2's emulated "VGA" connection that you can hit detect and two more "emulated" connections will appear?  Since no more than one VGA connection is showing in the topology section I'm having a hard time understanding how anymore connections would pop up.  I was really expecting at the very least something like four "DVI" connections or "Display port" connections showing up on the topology area since Kepler can drive up to four montiors.  When I pass through the Q4000 quite a bit shows up..  (i.e. one DVI, and two Display ports; but only two can be used at once).

[mikeleahy1234]

multi monitors dont seem to work the vdga

have any of you gotten this working ? I just half a screen on each monitor

[Linjo]

user I have a colleague of yours that have some issues to get vDGA working on a R720, could you give the details of your setup and steps you took to get vDGA working?

Thanks

Linjo

[LMUISHuntsville]

I got verification from nVidia that only one "emulated" vga monitor is currently supported with the current firmware.  They are however working to implement multi monitor (more than two monitors) support.  They currently have a beta firmware that supports two monitors, but I do not have it yet to verify.

[abeleski]

I can also confirm this happens in our environment.

Environment is

Horizon View 5.2.0 build 987719.

ESX 5.1.0 1157734 on Dell R720

LG Zero Client 23CAV42K (Tera 2 chip)

I have setup GPU passthrough for a VM. I followed recommendations of the following VMware guide.

http://www.vmware.com/files/pdf/techpaper/vmware-horizon-view-graphics-acceleration-deployment.pdf

The VM comes up fine with both the nVidia and SVGA card listed in device manager. However it always defaults to using the SVGA card i.e the zero client monitor attaches to the SVGA card and all 3d rendering is done by the SVGA card. If I try to open the nvidia control panel, it complains that I am not currently using a display that is attached to an NVIDIA GPU.

The doco I linked above states that as long as I have configured the cards fine, and I have set the zero client to use PCoIP and native resolution, when I boot up the VM,  it should attach to the nvidia k1, but it doesn’t.

The only way I can get the nvidia card activated is to uninstall the SVGA card from device manager. Once I have done that it activates the nvidia K1 and all is fine but only until you reboot. Upon reboot, windows detects the hardware for SVGA again and problem arises again.

I have found some clues on the web about this. See page 21 of this article “Configuring Virtual Monitors”

http://www.muycomputer.com/wp-content/uploads/2011/11/HP-Hardware-Accelerated-Graphics-for-VDI.pdf

and also

https://communities.vmware.com/thread/398755?start=0&tstart=0

The clues are that it has something to do with EDID. However I cannot do what is suggested which is plug in a monitor to the nvidia card on the host to save the EDID config as the nvidia K1’s don’t have any Display IO ports .

[abeleski]

"

Gunnarb,

We now have a K2 (to go along with our Q4000) and I'm having the exact opposite problem you had.  The Q4000 works fine in a dual monitor passthrough mode (using the monterey command) with view, etc... but when I do the same with the K2 card (single monitor) as soon as I log in using the view client (even a zero client) the screen blanks in/out about every second and the mouse cursor resets to the center of the monitor after each screen blank (basically rendering the desktop unusable).  I've tried both nVidia driver versions 320 and 311.5 with no success.  Do you have any ideas or advice on what you did with the K2 to have it work successfully (like what you show in your youtube video)?  Thanks for any tips you can provide.

Regards,

Nathan"

I can confirm that this also happens with K1 cards. Only when using dual monitor. Observed on a zero client with tera 2 chipset.

[Linjo]

You need a newer firmware on the cards, ask your OEM vendor (Dell, IBM, HP etc) for an updated firmware.

// Linjo

[0zzY]

Hi all!

i have a K1 card on R720 and a lot of this problems configuring vDGA, have you resolved? Do i need a new firmware for the K1 card? Have I to ask to dell or to  nvidia?

[cooladude]

yep, all working fine. Multi monitor also

Firmware needs to come from nvidia - drivers for vdga from nvidia also

[0zzY]

hi cooladude, how can i ask to nvidia for a firmware uppgrade? the drivers for vdga that i'm using are in the official download page, right?

[cooladude]

320.86 is the latest vdga driver

Do you have a support contract with nvidia or did you buy the cards from dell ?

[0zzY]

included in dell servers

[cooladude]

Open a case with dell tech support so and ask them for the firmware.

Its a tricky job to update it.

Drop me an email if you have any issues

mike_leahy@dell.com

[0zzY]

Thanks cooladude, my bios is 80.07.be.00.02, yours?

I have 2 problems using vdga i'm not using linked clone)::

1. unable to change resolution from nvida control panel
2. unable to start the client vm: if i power off the vm i can power it on only if video settings for the virtual machine are: enable 3D support and 3D render software. After the boot process the vm is auto reconfigured using the view poll settings (it reconfigure 3D render hardware). If i shut down the vm and i try to power on i receive the error Device 'Video card ' is not supported. This is a general limitation of the virtual machine's compatibility with the ESXi version of the selected host..

Had you got the same problems?

[Linjo]

Ozzy, it sounds like you have both vSGA and vDGA configured. Could you try to disable xorg just to make sure you are not running vSGA?

Would be helpful if you posted the vmx-file and some screen-shots of your configuration.