Artificial Intelligence Computing Leadership from NVIDIA
XenServer6.2 with XenDesktop7.6 with Grid K2
Hi, I have an issue with Grid K2 card installed in Fujitsu workstation R940 (drivers 346.68 with 348.27). Everything works great except when I select GPU from Pass-Through to vGPU (from K200 to K280Q), VM (Win7 64 bit) becomes very slow - slow boot, very bad interaction, no mouse arrow only pointer. I have somehow installed drivers for nvidia (for VM) but nothing changes. VM with 4 vCPU, 16Gb ram, installed on SSD. Can someone point me to the right direction, where to look to solve the problem. [url]https://www.imageupload.co.uk/images/2015/10/06/display1.png[/url]
Hi,

I have an issue with Grid K2 card installed in Fujitsu workstation R940 (drivers 346.68 with 348.27). Everything works great except when I select GPU from Pass-Through to vGPU (from K200 to K280Q), VM (Win7 64 bit) becomes very slow - slow boot, very bad interaction, no mouse arrow only pointer. I have somehow installed drivers for nvidia (for VM) but nothing changes.
VM with 4 vCPU, 16Gb ram, installed on SSD.
Can someone point me to the right direction, where to look to solve the problem.

https://www.imageupload.co.uk/images/2015/10/06/display1.png

#1
Posted 10/06/2015 06:14 PM   
Did you install the corresponding NVIDIA driver on your to match the server version? What does nvidia_smi.exe report on that VM as far as resource usage goes?
Did you install the corresponding NVIDIA driver on your to match the server version? What does nvidia_smi.exe report on that VM as far as resource usage goes?

-=Tobias

#2
Posted 10/07/2015 02:40 AM   
[quote="piper"] Everything works great except when I select GPU from Pass-Through to vGPU (from K200 to K280Q) [/quote] Does this mean that you have problems when you select the passthrough option? Or does it mean that you have a problem whenever you attach a GPU (passthrough or vGPU) Where do you get the problem? In the XenCenter console or via an ICA connection?
piper said: Everything works great except when I select GPU from Pass-Through to vGPU (from K200 to K280Q)


Does this mean that you have problems when you select the passthrough option?

Or does it mean that you have a problem whenever you attach a GPU (passthrough or vGPU)

Where do you get the problem? In the XenCenter console or via an ICA connection?

Jason Southern, Regional Lead for ProVis Sales - EMEA: NVIDIA Ltd.

#3
Posted 10/07/2015 09:34 AM   
If I select vGPU (K200 to K280Q) in VM properties then VM becomes slow (slow win7 install, slow boot,...) and if VM GPU is set to none or passthrough everything is running smoothly. Hope it is clear now. I am checking this with XenCenter console. It looks to me that problem is in XenServer side, something is using CPU and I dont now what. I am going to reinstall XenServer from USB to SSD and will see if anything changes... Thanks
If I select vGPU (K200 to K280Q) in VM properties then VM becomes slow (slow win7 install, slow boot,...) and if VM GPU is set to none or passthrough everything is running smoothly.
Hope it is clear now.
I am checking this with XenCenter console. It looks to me that problem is in XenServer side, something is using CPU and I dont now what. I am going to reinstall XenServer from USB to SSD and will see if anything changes...

Thanks

#4
Posted 10/07/2015 10:40 AM   
I had no luck. Everything works the same. Please check CPU usage when VM (with K280Q) is booted and no app is running. [url]https://www.imageupload.co.uk/images/2015/10/07/CPU_usage.png[/url] Bios is up to date. Patches for Xenserver 6.2 installed #lsmod | grep nvidia nvidia 7547799 50 i2c_core 20294 1 nvidia #dmesg | grep -E "NVRM" [ 23.982887] NVRM: loading NVIDIA UNIX x86 Kernel Module 346.68 Sun Apr 26 20:17:20 PDT 2015 [ 37.166450] NVRM: Your system is not currently configured to drive a VGA console [ 37.166452] NVRM: on the primary VGA device. The NVIDIA Linux graphics driver [ 37.166453] NVRM: requires the use of a text-mode VGA console. Use of other console [ 37.166454] NVRM: drivers including, but not limited to, vesafb, may result in [ 37.166455] NVRM: corruption and stability problems, and is not supported. #dmesg | grep -E "nvidia" [ 23.537816] nvidia: module license 'NVIDIA' taints kernel. [ 23.982172] nvidia 0000:04:00.0: enabling device (0000 -> 0003) [ 23.982180] nvidia 0000:04:00.0: PCI INT A -> GSI 32 (level, low) -> IRQ 32 [ 23.982263] nvidia 0000:04:00.0: setting latency timer to 64 [ 23.982456] nvidia 0000:05:00.0: enabling device (0000 -> 0003) [ 23.982461] nvidia 0000:05:00.0: PCI INT A -> GSI 32 (level, low) -> IRQ 32 [ 23.982477] nvidia 0000:05:00.0: setting latency timer to 64 [ 23.982702] nvidia 0000:06:00.0: PCI INT A -> GSI 40 (level, low) -> IRQ 40 [ 23.982714] nvidia 0000:06:00.0: setting latency timer to 64 [ 35.089524] nvidia 0000:04:00.0: get owner: 7ff0 [ 35.089599] nvidia 0000:04:00.0: irq 835 (291) for MSI/MSI-X [ 35.910299] nvidia 0000:05:00.0: get owner: 7ff0 [ 35.910341] nvidia 0000:05:00.0: irq 834 (290) for MSI/MSI-X [ 36.587080] nvidia 0000:06:00.0: get owner: 7ff0 [ 36.587107] nvidia 0000:06:00.0: irq 833 (289) for MSI/MSI-X [ 66.857972] nvidia 0000:06:00.0: get owner: 7ff0 [ 69.746824] nvidia 0000:06:00.0: get owner: 7ff0 [ 69.746877] nvidia 0000:06:00.0: irq 833 (289) for MSI/MSI-X [ 1977.700175] nvidia 0000:06:00.0: get owner: 7ff0 [ 1980.545937] nvidia 0000:06:00.0: get owner: 7ff0 [ 1980.545987] nvidia 0000:06:00.0: irq 833 (289) for MSI/MSI-X [ 4408.935784] nvidia 0000:06:00.0: get owner: 7ff0 [ 4411.846006] nvidia 0000:06:00.0: get owner: 7ff0 [ 4411.846059] nvidia 0000:06:00.0: irq 833 (289) for MSI/MSI-X [ 6489.225152] nvidia 0000:06:00.0: get owner: 7ff0 [ 6492.103800] nvidia 0000:06:00.0: get owner: 7ff0 #nvidia-smi Wed Oct 7 21:23:56 2015 +------------------------------------------------------+ | NVIDIA-SMI 346.68 Driver Version: 346.68 | |-------------------------------+----------------------+----------------------+ | GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC | | Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. | |===============================+======================+======================| | 0 GRID K2 On | 0000:04:00.0 Off | Off | | N/A 35C P8 29W / 117W | 11MiB / 4095MiB | 0% Default | +-------------------------------+----------------------+----------------------+ | 1 GRID K2 On | 0000:05:00.0 Off | Off | | N/A 32C P8 28W / 117W | 3787MiB / 4095MiB | 0% Default | +-------------------------------+----------------------+----------------------+ | 2 Quadro 2000 Off | 0000:06:00.0 Off | N/A | | 30% 58C P8 N/A / N/A | 3MiB / 1023MiB | 0% Default | +-------------------------------+----------------------+----------------------+ +-----------------------------------------------------------------------------+ | Processes: GPU Memory | | GPU PID Type Process name Usage | |=============================================================================| | No running processes found | +-----------------------------------------------------------------------------+
I had no luck. Everything works the same. Please check CPU usage when VM (with K280Q) is booted and no app is running.

https://www.imageupload.co.uk/images/2015/10/07/CPU_usage.png

Bios is up to date.
Patches for Xenserver 6.2 installed


#lsmod | grep nvidia
nvidia 7547799 50
i2c_core 20294 1 nvidia

#dmesg | grep -E "NVRM"
[ 23.982887] NVRM: loading NVIDIA UNIX x86 Kernel Module 346.68 Sun Apr 26 20:17:20 PDT 2015
[ 37.166450] NVRM: Your system is not currently configured to drive a VGA console
[ 37.166452] NVRM: on the primary VGA device. The NVIDIA Linux graphics driver
[ 37.166453] NVRM: requires the use of a text-mode VGA console. Use of other console
[ 37.166454] NVRM: drivers including, but not limited to, vesafb, may result in
[ 37.166455] NVRM: corruption and stability problems, and is not supported.

#dmesg | grep -E "nvidia"
[ 23.537816] nvidia: module license 'NVIDIA' taints kernel.
[ 23.982172] nvidia 0000:04:00.0: enabling device (0000 -> 0003)
[ 23.982180] nvidia 0000:04:00.0: PCI INT A -> GSI 32 (level, low) -> IRQ 32
[ 23.982263] nvidia 0000:04:00.0: setting latency timer to 64
[ 23.982456] nvidia 0000:05:00.0: enabling device (0000 -> 0003)
[ 23.982461] nvidia 0000:05:00.0: PCI INT A -> GSI 32 (level, low) -> IRQ 32
[ 23.982477] nvidia 0000:05:00.0: setting latency timer to 64
[ 23.982702] nvidia 0000:06:00.0: PCI INT A -> GSI 40 (level, low) -> IRQ 40
[ 23.982714] nvidia 0000:06:00.0: setting latency timer to 64
[ 35.089524] nvidia 0000:04:00.0: get owner: 7ff0
[ 35.089599] nvidia 0000:04:00.0: irq 835 (291) for MSI/MSI-X
[ 35.910299] nvidia 0000:05:00.0: get owner: 7ff0
[ 35.910341] nvidia 0000:05:00.0: irq 834 (290) for MSI/MSI-X
[ 36.587080] nvidia 0000:06:00.0: get owner: 7ff0
[ 36.587107] nvidia 0000:06:00.0: irq 833 (289) for MSI/MSI-X
[ 66.857972] nvidia 0000:06:00.0: get owner: 7ff0
[ 69.746824] nvidia 0000:06:00.0: get owner: 7ff0
[ 69.746877] nvidia 0000:06:00.0: irq 833 (289) for MSI/MSI-X
[ 1977.700175] nvidia 0000:06:00.0: get owner: 7ff0
[ 1980.545937] nvidia 0000:06:00.0: get owner: 7ff0
[ 1980.545987] nvidia 0000:06:00.0: irq 833 (289) for MSI/MSI-X
[ 4408.935784] nvidia 0000:06:00.0: get owner: 7ff0
[ 4411.846006] nvidia 0000:06:00.0: get owner: 7ff0
[ 4411.846059] nvidia 0000:06:00.0: irq 833 (289) for MSI/MSI-X
[ 6489.225152] nvidia 0000:06:00.0: get owner: 7ff0
[ 6492.103800] nvidia 0000:06:00.0: get owner: 7ff0

#nvidia-smi
Wed Oct 7 21:23:56 2015
+------------------------------------------------------+
| NVIDIA-SMI 346.68 Driver Version: 346.68 |
|-------------------------------+----------------------+----------------------+
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
|===============================+======================+======================|
| 0 GRID K2 On | 0000:04:00.0 Off | Off |
| N/A 35C P8 29W / 117W | 11MiB / 4095MiB | 0% Default |
+-------------------------------+----------------------+----------------------+
| 1 GRID K2 On | 0000:05:00.0 Off | Off |
| N/A 32C P8 28W / 117W | 3787MiB / 4095MiB | 0% Default |
+-------------------------------+----------------------+----------------------+
| 2 Quadro 2000 Off | 0000:06:00.0 Off | N/A |
| 30% 58C P8 N/A / N/A | 3MiB / 1023MiB | 0% Default |
+-------------------------------+----------------------+----------------------+

+-----------------------------------------------------------------------------+
| Processes: GPU Memory |
| GPU PID Type Process name Usage |
|=============================================================================|
| No running processes found |
+-----------------------------------------------------------------------------+

#5
Posted 10/07/2015 07:26 PM   
Hi, here [url]http://www.nvidia.com/object/grid-certified-servers.html[/url] it is stated that Fujitsu workstation R940power is certified for Nvidia GRID. Can please somebody tell me with which hypervisor and version it is certified (XenServer, vSpere Esxi....) for GRID and vGPU.
Hi,

here http://www.nvidia.com/object/grid-certified-servers.html it is stated that Fujitsu workstation R940power is certified for Nvidia GRID.
Can please somebody tell me with which hypervisor and version it is certified (XenServer, vSpere Esxi....) for GRID and vGPU.

#6
Posted 10/08/2015 09:52 AM   
http://hcl.xenserver.org/ Does not list the R940 only R920 & R930 I would check with Fujitsu on their certification as I can't find anything regarding using that server with XenServer.

http://hcl.xenserver.org/


Does not list the R940 only R920 & R930

I would check with Fujitsu on their certification as I can't find anything regarding using that server with XenServer.

Jason Southern, Regional Lead for ProVis Sales - EMEA: NVIDIA Ltd.

#7
Posted 10/08/2015 07:37 PM   
Scroll To Top

Add Reply