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.
Svga and vSGA has nothing todo with each other and you will not be able to tell from inside the VM that its using a GPU except for much improved 3D.
One tool that I use to verify is this one:
http://www.ozone3d.net/gpu_caps_viewer/
Run the OpenGL 2.1 benchmark and 600x600 and you should get about 30fps, without vSGA you would only get 1-3fps.
// Linjo
Goodmorning guys, hope u had a good weekend.
Week has started again. i just ran some of your commands to see what's installed and whats not.
here's the update:
esxcli software vib list | grep nvidia says nvidia vmware esxi 5.1 host driver vmware accepted
Esxcli system module load –module nvidia
unable to load module, status : busy. seems to me its loaded? correct me if im wrong ofcourse.
Esxcli hardware pci list –c 0x300 –m 0xff
I both see my quadro 600 and my quadro 4000 if i do that command.
when i do the nvidia smi command it gives me my quadro 4000 status, temps and gpu fan usage.
display 4MB / off 2047 mb
xorg service is still running.
same this happens when i try to boot up the VM using the hardware option,
device videocard is not supported by host!
and yes it is running on the right host!
Thanks in advance,
drxz
Could you post the VMX-file and the vmware.log from the VM?
// Linjo
Hi,
Thanks for the update
Kindly see the table below
~ # nvidia-smi
Mon Mar 25 09:44:56 2013
+------------------------------------------------------+
| NVIDIA-SMI 4.304.76 Driver Version: 304.76 |
|-------------------------------+----------------------+----------------------+
| GPU Name | Bus-Id Disp. | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
|===============================+======================+======================|
| 0 Quadro 4000 | 0000:05:00.0 Off | N/A |
| 36% 82C P1 N/A / N/A | 29% 594MB / 2047MB | 7% Default |
+-------------------------------+----------------------+----------------------+
| 1 Quadro 4000 | 0000:42:00.0 Off | N/A |
| 36% 52C P12 N/A / N/A | 15% 306MB / 2047MB | 12% Default |
+-------------------------------+----------------------+----------------------+
+-----------------------------------------------------------------------------+
| Compute processes: GPU Memory |
| GPU PID Process name Usage |
|=============================================================================|
| No running compute processes found |
+-----------------------------------------------------------------------------+
GPU utilization is is just 15% to 30%. How can I utiliz to its max.
Here's the vmx log. trying to find the vmware.log file. its all new to me!
This thread is becoming a bit messy with a few different issues being discussed.
I would recommend to post a new thread if its not directly related to what have been previously discussed.
The GPU will not be used 100% unless there is enough work for it todo, same as any CPU...
I think it looks healthy with a range from 15% to 30% it means that you have some headroom and have not maxed out the CPU.
I am sure you can get it to 90%+ if you are running some GPU benchmark apps in the VM:s
// Linjo
Hi, do you know if vSGA works with nVIDIA's latest K4000 or K5000?
It seems only Fermi architecture is supported at the moment.
Not sure if there is any Kepler compatible ESXi driver available right now.
Any experience on that please?
One more question: would vMotion work on a VM that has video hardware acceleration enabled?
Cheers.
Can't see anything that stands out in the vmx-file except for the size of the svga.vramsize, try to set it to 512Mb in the vSphere GUI and enable 3D.
The vmware.log is usually in the same folder as the vmx-file on the datastore.
Is this vm a newly built machine or has it been converted or used for something else?
// Linjo
it has been on 512 for the last few days to try. didnt help.
Its a standard template so there are more vm's that use the same template.
Will try to make a clean install and see what that does.
let me check again for the log file.
does it help if i configure the gpu for passtrough or isnt that neseccary?
Thanks.
edit,
Logfile included
Hi,
Anyone tried with inferior cards, for example quadro 3000 or 2000?
Unfortunately only the Quadro 4000-6000; VGX K1 and K2 (GRID), and a one other Tesla (cant remember the model) are currently supported with the 304.79 driver. The Keplar Quadro cards (K4000/K5000) are not supported; hopefully that will only be a "yet".
The GPU should not be marked for passthrough.
Could you post the output from the "gpuvm" cli-command?
// Linjo
gpuvm command gives nothing.
just goes to the next input line without any feedback!
"would vMotion work on a VM that has video hardware acceleration enabled?"
Yes vMotion will work with vSGA enabled VMs. Also if you are going between a vSGA enabled host and a non-vSGA enabled host if you have 3D rendering set to "Automatic" then it will also support vMotion to hosts without GPUs using Soft3D.
Is your xorg service running?
issue "/etc/init.d/xorg status" to check. If it is not start it.
xorg is running yes!
allthough i started it in vsphere, where it says running. when i entered the command it said not running.
started it, but still same problem that the vm wont start with the same device videocard is not supported.
okay, now i do get input from the gpuvm command.
xserver unix:0, gpu maximum memory 2076672KB
gpu memory left 2076672kb
thats it.
Have you tried to start the vm again now that "gpuvm" gives you some output? Thats basically just saying how much GPU VRAM is avaliable to divy out to vSGA enabled VMs, which currently you dont have any powered on (or unable too in your case).
It seems to be working guys,
10 vm's working on this gpu now. 5 of them are now playing a 1080p movie without a problem.
I dont know how to thank you guys, but really thanks!
also got our teradici server offload card working.
Thanks, have a great week.
DrxzEEh
Great to hear! Happy it now works for you.
Please give us feedback on how you think it performs and the usecases that you see.
// Linjo