p106-100 Limited to 7 GPUs, limited nvidia-smi

Graphics / Linux Linux
1

I am trying to run 8 * P106-100s in a CentOS 7.3 system. The drivers 384.47 and 384.59 only detect 7 GPUs.

My hardware detects and supports 8 GPUs.

[root@localhost ~]# lspci | grep NVIDIA
01:00.0 3D controller: NVIDIA Corporation Device 1c07 (rev a1)
02:00.0 3D controller: NVIDIA Corporation Device 1c07 (rev a1)
03:00.0 3D controller: NVIDIA Corporation Device 1c07 (rev a1)
05:00.0 3D controller: NVIDIA Corporation Device 1c07 (rev a1)
08:00.0 3D controller: NVIDIA Corporation Device 1c07 (rev a1)
09:00.0 3D controller: NVIDIA Corporation Device 1c07 (rev a1)
0a:00.0 3D controller: NVIDIA Corporation Device 1c07 (rev a1)
0b:00.0 3D controller: NVIDIA Corporation Device 1c07 (rev a1)

The drivers 384.47 and 384.59 only detect 7 GPUs.

+-----------------------------------------------------------------------------+
| NVIDIA-SMI 384.59                 Driver Version: 384.59                    |
|-------------------------------+----------------------+----------------------+
| GPU  Name        Persistence-M| Bus-Id        Disp.A | Volatile Uncorr. ECC |
| Fan  Temp  Perf  Pwr:Usage/Cap|         Memory-Usage | GPU-Util  Compute M. |
|===============================+======================+======================|
|   0  P106-100            Off  | 00000000:01:00.0 Off |                  N/A |
| 29%   35C    P0    28W / 120W |      0MiB /  6075MiB |      0%      Default |
+-------------------------------+----------------------+----------------------+
|   1  P106-100            Off  | 00000000:02:00.0 Off |                  N/A |
| 29%   39C    P0    28W / 120W |      0MiB /  6075MiB |      0%      Default |
+-------------------------------+----------------------+----------------------+
|   2  P106-100            Off  | 00000000:03:00.0 Off |                  N/A |
| 29%   39C    P0    28W / 120W |      0MiB /  6075MiB |      0%      Default |
+-------------------------------+----------------------+----------------------+
|   3  P106-100            Off  | 00000000:05:00.0 Off |                  N/A |
| 29%   39C    P0    28W / 120W |      0MiB /  6075MiB |      0%      Default |
+-------------------------------+----------------------+----------------------+
|   4  P106-100            Off  | 00000000:08:00.0 Off |                  N/A |
| 29%   40C    P0    28W / 120W |      0MiB /  6075MiB |      0%      Default |
+-------------------------------+----------------------+----------------------+
|   5  P106-100            Off  | 00000000:09:00.0 Off |                  N/A |
| 31%   41C    P0    28W / 120W |      0MiB /  6075MiB |      0%      Default |
+-------------------------------+----------------------+----------------------+
|   6  P106-100            Off  | 00000000:0B:00.0 Off |                  N/A |
| 29%   40C    P0    28W / 120W |      0MiB /  6075MiB |      3%      Default |
+-------------------------------+----------------------+----------------------+
                                                                               
+-----------------------------------------------------------------------------+
| Processes:                                                       GPU Memory |
|  GPU       PID  Type  Process name                               Usage      |
|=============================================================================|
|  No running processes found                                                 |
+-----------------------------------------------------------------------------+

Additionally setting clock and memory speed is not supported for the P106-100. This is a severe limitation for GPUs purpose built for mining.

nvidia-smi -ac 1500,9000
Setting applications clocks is not supported for GPU 00000000:01:00.0.
Treating as warning and moving on.
Setting applications clocks is not supported for GPU 00000000:02:00.0.
Treating as warning and moving on.
Setting applications clocks is not supported for GPU 00000000:03:00.0.
Treating as warning and moving on.
Setting applications clocks is not supported for GPU 00000000:05:00.0.
Treating as warning and moving on.
Setting applications clocks is not supported for GPU 00000000:08:00.0.
Treating as warning and moving on.
Setting applications clocks is not supported for GPU 00000000:09:00.0.
Treating as warning and moving on.
Setting applications clocks is not supported for GPU 00000000:0B:00.0.
Treating as warning and moving on.
All done.

I purchased 496 of these GPUs and currently 1/8 are unusable. Additionally the other 7/8 are running 20% below desired and tested stable speeds in a test windows environment.

Please acknowledge and advise.

2

For over clocking, you can assign offset clock with nvidia-settings (after u have enabled it with nvidia-xconfig). Work fine for me with gtx1070, gtx1060, and p106. I’m still looking for a way to over voltage the P106s. Since it can’t be resold in the future, I want a setting that will fry the cards after 2-3 years when the 7nm Navi come

3

NVIDIA does not want your business. Go bankrupt miner!

In the past during gold rushes, most money was made by companies selling mining equipment.
NVIDIA is determined not to repeat that. That’s why NVIDIA has decided that new drivers will allow max 2 cards per machine and will underclock your memory if used for more than 4 hours per day.

End trolling. Or was it trolling? :)

@Morgiana - please provide list of commands you entered to do that.
Do I need to use nvidia driver for x for that approach to work?

4

Run nvidia-bug-report.sh, attach output. The application clock setting (-ac) is not for setting clocks but for limiting them. Only supported on specific gpus, also depending on vendor. Proceed as Morgiana said to set clocks.

5

>>The drivers 384.47 and 384.59 only detect 7 GPUs.
Please run nvidia-bug-report.sh, attach generated log file.

Please try cool-bits xorg.conf option to overclock gpu and memory clock via nvidia-settings graphical tool