我在EC2 p2.xlarge共享实例上运行Ubuntu 16.04。
P2.xlarge实例可以访问单个GPU(NVidia K80 GPU的1/2)到我的共享实例。
我的印象是100%的GPU已分配给我的实例。但事实显然并非如此。当我的实例在GPU上运行 nothing 时:
$ nvidia-smi
Tue Feb 21 00:11:16 2017
+-----------------------------------------------------------------------------+
| NVIDIA-SMI 375.39 Driver Version: 375.39 |
|-------------------------------+----------------------+----------------------+
| 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 K80 Off | 0000:00:1E.0 Off | 0 |
| N/A 39C P0 55W / 149W | 0MiB / 11439MiB | 63% Default |
+-------------------------------+----------------------+----------------------+
+-----------------------------------------------------------------------------+
| Processes: GPU Memory |
| GPU PID Type Process name Usage |
|=============================================================================|
| No running processes found |
+-----------------------------------------------------------------------------+
片刻之后:
| 0 Tesla K80 Off | 0000:00:1E.0 Off | 0 |
| N/A 40C P0 57W / 149W | 0MiB / 11439MiB | 97% Default |
下一刻:
| 0 Tesla K80 Off | 0000:00:1E.0 Off | 0 |
| N/A 40C P0 56W / 149W | 0MiB / 11439MiB | 100% Default |
并且几乎待在那里......
| 0 Tesla K80 Off | 0000:00:1E.0 Off | 0 |
| N/A 41C P0 56W / 149W | 0MiB / 11439MiB | 100% Default |
| 0 Tesla K80 Off | 0000:00:1E.0 Off | 0 |
| N/A 41C P0 56W / 149W | 0MiB / 11439MiB | 99% Default |