8b2485c441
Update scheduling-gpus.md
...
modify "nvidia drivers" to "Nvidia drivers"
2017-08-18 09:19:32 -07:00
58fa6bf42c
Update scheduling-gpus.md ( #4850 )
...
* Update scheduling-gpus.md
It sounds like that kubelet will not automatically load the nvidia uvm driver,I have recurrent it when my cluster reboot nvml driver and nvidia-uvm driver not loaded.
To fix it simply,make the docker-plugin autostart.
I think it's important to tell other developers to prevent this situation
* Update scheduling-gpus.md
2017-08-17 17:00:10 -07:00
d76fbb0b27
Update scheduling-gpus.md
...
privileged mode is not needed on my kubernetes cluster, and it will make other developers confuse.
e.g:
I try to use this example to test whether the keyword "resources->limit" will work.when I set limit:1 with privileged,I can reach all gpus instead of single gpu.
```
kind: Pod
apiVersion: v1
metadata:
name: gpu-pod-1
spec:
containers:
- name: gpu-container-1
image: tensorflow/tensorflow:latest-gpu
securityContext:
privileged: true
env:
- name: LD_LIBRARY_PATH
value: "$LD_LIBRARY_PATH:/usr/lib/nvidia"
- name: CUDA_HOME
value: "/usr/local/nvidia"
# securityContext:
# privileged: true
resources:
limits:
alpha.kubernetes.io/nvidia-gpu: 1
volumeMounts:
- mountPath: /usr/local/nvidia/bin
name: bin
- mountPath: /usr/lib/nvidia
name: lib
volumes:
- hostPath:
path: /usr/local/cuda/bin
name: bin
- hostPath:
path: /var/lib/nvidia-docker/volumes/nvidia_driver/375.26/lib64/
name: lib
```
it show all gpu devices because of privileged mode
```
[name: "/cpu:0"
device_type: "CPU"
memory_limit: 268435456
locality {
}
incarnation: 3223127178054071976
, name: "/gpu:0"
device_type: "GPU"
memory_limit: 103809024
locality {
bus_id: 1
}
incarnation: 6184347283432924133
physical_device_desc: "device: 0, name: Quadro K4200, pci bus id: 0000:03:00.0"
, name: "/gpu:1"
device_type: "GPU"
memory_limit: 103809024
locality {
bus_id: 1
}
incarnation: 1058630844567460797
physical_device_desc: "device: 1, name: Quadro K4200, pci bus id: 0000:04:00.0"
]
```
2017-08-07 16:49:35 -07:00
bda690e547
Update scheduling-gpus.md
2017-08-04 14:36:53 -07:00
18e9aa59d1
Rename OWNERS assignees: to approvers:
...
They are effectively the same, assignees is deprecated
2017-08-03 16:22:52 -07:00
ec0c650e6a
Fixed apostrophe
2017-08-02 14:06:20 -07:00
4d9320bb40
yaml file errors on scheduling-gpus.md ( #4436 )
...
* Update scheduling-gpus.md
* Update scheduling-gpus.md
2017-08-01 14:33:04 -07:00
2e257d9707
Changes for move to Netlify ( #4464 )
...
* disable jekyll-redirect-from gem
* add _redirects file
* disable 404 redirect script
* add 301 redirect to test
* retain _redirects file
* Convert redirect_from's to _redirect file. (#4409 )
* Remove redirect_from's. (#4424 )
* Add 301's to _redirects. (#4427 )
* add whitespace before 301
* move redirects in /js/redirects/js to _redirects
* add disabled option for cn redirect
* convert include to array in _config.yml
* enable redirects.js script for legacy support
2017-07-28 08:23:11 -07:00
b6cc04ed74
Merge pull request #4116 from feiskyer/gpu
...
Update GPU docs
2017-06-22 10:35:56 -07:00
a0ffa2569c
Update GPU docs
2017-06-16 21:15:28 +08:00
6c99bbb509
Change Task titles to imperative: Daemons, GPUs. ( #4032 )
2017-06-08 14:48:46 -07:00
ae5df6944d
Add GPU mountpath warning
...
The GPU path is not permissive by default and requires a bit of
additional setup if the operator does not allow for privileged
containers.
Related kubernetes/kubernetes#46007
Signed-off-by: Christopher M. Luciano <cmluciano@us.ibm.com>
2017-05-23 22:39:56 -07:00
80a2fff4dc
Deprecation complete: gpus ... ( #3537 )
2017-04-25 20:09:38 -07:00
fbfafc9738
Fix broken link:scheduling-gpus.md ( #3423 )
...
* Fix broken link:scheduling-gpus.md
Fix broken link:scheduling-gpus.md
* add link anchor
add link anchor
2017-04-17 09:17:39 -07:00
d2ff41b6a5
Move GPU Support to Tasks ( #3212 )
...
* Move User Guides/GPU Support to Tasks/Managing GPUs/Scheduling GPUs
* fix typo in tasks.yml
* change discussion to steps
2017-04-03 17:09:43 -07:00
Anthony
dengyi1996
Aaron Crickenberger
cliffburdick
WanLinghao
Andrew Chen
Pengfei Ni
Steve Perry
Christopher M. Luciano
Xiaoyu Zhang