Artificial Intelligence Computing Leadership from NVIDIA
Not all supported vGPU types are creatable in multi-GPU Hypervisor
Dear Forum, I have two new Dell PowerEdge R740 Servers with 2 V100 PCIe 32GB cards installed and I am running RedHat Enterprise Linux 7.7 with the NVIDIA-vGPU-rhel-7.7-440.53 rpm package installed. This is running RedHat Openstack 13 (Queens) with the latest Zstream patches. [root@srv-p23-30 ~]# nvidia-smi Wed Feb 12 14:31:27 2020 +-----------------------------------------------------------------------------+ | NVIDIA-SMI 440.53 Driver Version: 440.53 CUDA Version: N/A | |-------------------------------+----------------------+----------------------+ | GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC | | Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. | |===============================+======================+======================| | 0 Tesla V100-PCIE... On | 00000000:3B:00.0 Off | 0 | | N/A 27C P0 24W / 250W | 51MiB / 32767MiB | 0% Default | +-------------------------------+----------------------+----------------------+ | 1 Tesla V100-PCIE... On | 00000000:D8:00.0 Off | 0 | | N/A 26C P0 24W / 250W | 39MiB / 32767MiB | 0% Default | +-------------------------------+----------------------+----------------------+ +-----------------------------------------------------------------------------+ | Processes: GPU Memory | | GPU PID Type Process name Usage | |=============================================================================| | No running processes found | +-----------------------------------------------------------------------------+ Without any guest VMs running, If I query the GPU cards for CGPU supported types I get full list as expected: [root@srv-p23-30 ~]# nvidia-smi vgpu -s GPU 00000000:3B:00.0 GRID V100D-1Q GRID V100D-2Q GRID V100D-4Q GRID V100D-8Q GRID V100D-16Q GRID V100D-32Q GRID V100D-1A GRID V100D-2A GRID V100D-4A GRID V100D-8A GRID V100D-16A GRID V100D-32A GRID V100D-1B GRID V100D-1B4 GRID V100D-2B GRID V100D-2B4 GRID V100D-4C GRID V100D-8C GRID V100D-16C GRID V100D-32C GPU 00000000:D8:00.0 GRID V100D-1Q GRID V100D-2Q GRID V100D-4Q GRID V100D-8Q GRID V100D-16Q GRID V100D-32Q GRID V100D-1A GRID V100D-2A GRID V100D-4A GRID V100D-8A GRID V100D-16A GRID V100D-32A GRID V100D-1B GRID V100D-1B4 GRID V100D-2B GRID V100D-2B4 GRID V100D-4C GRID V100D-8C GRID V100D-16C GRID V100D-32C However when I check for the capabilities, the first GPU lists only one type but the second GPU is what I expect to see: [root@srv-p23-30 ~]# nvidia-smi vgpu -c GPU 00000000:3B:00.0 GRID V100D-2Q GPU 00000000:D8:00.0 GRID V100D-1Q GRID V100D-2Q GRID V100D-4Q GRID V100D-8Q GRID V100D-16Q GRID V100D-32Q GRID V100D-1A GRID V100D-2A GRID V100D-4A GRID V100D-8A GRID V100D-16A GRID V100D-32A GRID V100D-1B GRID V100D-1B4 GRID V100D-2B GRID V100D-2B4 GRID V100D-4C GRID V100D-8C GRID V100D-16C GRID V100D-32C So what is happening is that no matter what GPU type I try to create my VM with it ends up just being GRID V100D-2Q within the VM. Is there something I am missing here? I apologize if this was already asked. I looked through a bunch of forum articles already and did not see anything. Thank You, Salvatore
Dear Forum,


I have two new Dell PowerEdge R740 Servers with 2 V100 PCIe 32GB cards installed and I am running RedHat Enterprise Linux 7.7 with the NVIDIA-vGPU-rhel-7.7-440.53 rpm package installed. This is running RedHat Openstack 13 (Queens) with the latest Zstream patches.


[root@srv-p23-30 ~]# nvidia-smi
Wed Feb 12 14:31:27 2020
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 440.53 Driver Version: 440.53 CUDA Version: N/A |
|-------------------------------+----------------------+----------------------+
| GPU Name Persistence-M| Bus-Id Disp.A | Volatile Uncorr. ECC |
| Fan Temp Perf Pwr:Usage/Cap| Memory-Usage | GPU-Util Compute M. |
|===============================+======================+======================|
| 0 Tesla V100-PCIE... On | 00000000:3B:00.0 Off | 0 |
| N/A 27C P0 24W / 250W | 51MiB / 32767MiB | 0% Default |
+-------------------------------+----------------------+----------------------+
| 1 Tesla V100-PCIE... On | 00000000:D8:00.0 Off | 0 |
| N/A 26C P0 24W / 250W | 39MiB / 32767MiB | 0% Default |
+-------------------------------+----------------------+----------------------+

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


Without any guest VMs running, If I query the GPU cards for CGPU supported types I get full list as expected:

[root@srv-p23-30 ~]# nvidia-smi vgpu -s
GPU 00000000:3B:00.0
GRID V100D-1Q
GRID V100D-2Q
GRID V100D-4Q
GRID V100D-8Q
GRID V100D-16Q
GRID V100D-32Q
GRID V100D-1A
GRID V100D-2A
GRID V100D-4A
GRID V100D-8A
GRID V100D-16A
GRID V100D-32A
GRID V100D-1B
GRID V100D-1B4
GRID V100D-2B
GRID V100D-2B4
GRID V100D-4C
GRID V100D-8C
GRID V100D-16C
GRID V100D-32C

GPU 00000000:D8:00.0
GRID V100D-1Q
GRID V100D-2Q
GRID V100D-4Q
GRID V100D-8Q
GRID V100D-16Q
GRID V100D-32Q
GRID V100D-1A
GRID V100D-2A
GRID V100D-4A
GRID V100D-8A
GRID V100D-16A
GRID V100D-32A
GRID V100D-1B
GRID V100D-1B4
GRID V100D-2B
GRID V100D-2B4
GRID V100D-4C
GRID V100D-8C
GRID V100D-16C
GRID V100D-32C


However when I check for the capabilities, the first GPU lists only one type but the second GPU is what I expect to see:

[root@srv-p23-30 ~]# nvidia-smi vgpu -c
GPU 00000000:3B:00.0
GRID V100D-2Q

GPU 00000000:D8:00.0
GRID V100D-1Q
GRID V100D-2Q
GRID V100D-4Q
GRID V100D-8Q
GRID V100D-16Q
GRID V100D-32Q
GRID V100D-1A
GRID V100D-2A
GRID V100D-4A
GRID V100D-8A
GRID V100D-16A
GRID V100D-32A
GRID V100D-1B
GRID V100D-1B4
GRID V100D-2B
GRID V100D-2B4
GRID V100D-4C
GRID V100D-8C
GRID V100D-16C
GRID V100D-32C


So what is happening is that no matter what GPU type I try to create my VM with it ends up just being GRID V100D-2Q within the VM.


Is there something I am missing here? I apologize if this was already asked. I looked through a bunch of forum articles already and did not see anything.

Thank You,

Salvatore

#1
Posted 02/12/2020 07:55 PM   
Scroll To Top

Add Reply