| ... | @@ -50,7 +50,7 @@ Alternatively, you can leverage [NGC](https://github.com/NVIDIA/nvidia-docker/wi | 
... | @@ -50,7 +50,7 @@ Alternatively, you can leverage [NGC](https://github.com/NVIDIA/nvidia-docker/wi | 
| 
 | 
## Container Runtime
 | 
 | 
## Container Runtime
 | 
| 
 | 
 | 
 | 
 | 
| 
 | 
#### Does it have a performance impact on my GPU workload? 
 | 
 | 
#### Does it have a performance impact on my GPU workload? 
 | 
| 
 | 
No, usually the impact should be in the order of magnitude less than 1% and hardly noticeable.\
 | 
 | 
No, usually the impact should be in the order of less than 1% and hardly noticeable.\
 | 
| 
 | 
However be aware of the following (list non exhaustive):
 | 
 | 
However be aware of the following (list non exhaustive):
 | 
| 
 | 
* _GPU topology and CPU affinity_ \
 | 
 | 
* _GPU topology and CPU affinity_ \
 | 
| 
 | 
You can query it using `nvidia-smi topo` and use [Docker CPU sets](https://docs.docker.com/engine/admin/resource_constraints/#cpu) to pin CPU cores.
 | 
 | 
You can query it using `nvidia-smi topo` and use [Docker CPU sets](https://docs.docker.com/engine/admin/resource_constraints/#cpu) to pin CPU cores.
 | 
| ... |  | 
... |  | 
| ... |  | ... |  |