Desktop app not utilizing GPU


#1

The desktop app is not utilizing GPU for training. Is there any specific way to get this done?


#2


#3

First, install MSI Afterburner, it will show more and better statistics. also, in training tab of browser DLS you can check utilizing GPU memory.

Model may not fit in GPU memory, how much GPU memory do you have onboard?


#4

@Harveen_chadha

do you see GPU dropdown in the training tab ? If not, please share the log from log tab to check if GPU was enabled or not.

Did you complete the CUDA 9.0 installation during deep learning studio install?


#5

I am quiet aware of the memory problem. I have 4GB graphic card, the problem is current utilization of GPU is zero.


#6

Full Logs
Starting Deep Learning Studio…
Checking GPU support…
GPU supported
skipping cifar dataset download!
skipping IMDB dataset download!
skipping MNIST dataset download!
skipping reuters dataset download!
Using MXNet backend.
C:\Users\kanwal\AppData\Local\Programs\DeepLearningStudio\conda3\lib\site-packages\allauth\account\templatetags\account_tags.py:4: DeprecationWarning: {% load account_tags %} is deprecated, use {% load account %}
DeprecationWarning)
C:\Users\kanwal\AppData\Local\Programs\DeepLearningStudio\conda3\lib\site-packages\allauth\socialaccount\templatetags\socialaccount_tags.py:4: DeprecationWarning: {% load socialaccount_tags %} is deprecated, use {% load socialaccount %}
" {% load socialaccount %}", DeprecationWarning)
No changes detected
Operations to perform:
Apply all migrations: account, admin, auth, authtoken, automl, contenttypes, environments, project, projects, reversion, sessions, sites, socialaccount
Running migrations:
No migrations to apply.
Using MXNet backend.
C:\Users\kanwal\AppData\Local\Programs\DeepLearningStudio\conda3\lib\site-packages\allauth\account\templatetags\account_tags.py:4: DeprecationWarning: {% load account_tags %} is deprecated, use {% load account %}
DeprecationWarning)
C:\Users\kanwal\AppData\Local\Programs\DeepLearningStudio\conda3\lib\site-packages\allauth\socialaccount\templatetags\socialaccount_tags.py:4: DeprecationWarning: {% load socialaccount_tags %} is deprecated, use {% load socialaccount %}
" {% load socialaccount %}", DeprecationWarning)
Using MXNet backend.

  • Running on http://127.0.0.1:6666/ (Press CTRL+C to quit)
    Using MXNet backend.
    [I 19:18:18.279 NotebookApp] Serving notebooks from local directory: C:\Users\kanwal\AppData\Local\Programs\DeepLearningStudio\data\1
    [I 19:18:18.279 NotebookApp] 0 active kernels
    [I 19:18:18.280 NotebookApp] The Jupyter Notebook is running at:
    [I 19:18:18.280 NotebookApp] http://127.0.0.1:8886/?token=
    [I 19:18:18.280 NotebookApp] Use Control-C to stop this server and shut down all kernels (twice to skip confirmation).
    [02/Jul/2018:19:18:19] ENGINE Bus STARTING
    [02/Jul/2018:19:18:19] ENGINE Serving on http://0.0.0.0:8000
    [02/Jul/2018:19:18:19] ENGINE Bus STARTED
    Starting server worker…

Server worker has been started. [ID: 1 | PID: 2008]

Server worker is ready to accept messages. [ID: 1 | PID: 2008]

[2018-07-02T13:48:42.967Z] INFO: Theia/2008 on Kanwal: Theia app listening on http://0.0.0.0:8889. []

Received message which is neither a response nor a notification message:

“8889”

[2018-07-02T13:48:43.265Z] INFO: Theia/2008 on Kanwal:
[nsfw-watcher: 5608] Started watching: c:\Users\kanwal\AppData\Local\Programs\DeepLearningStudio\home\theia\examples\browser\package.json
[]

C:\Users\kanwal\AppData\Local\Programs\DeepLearningStudio\conda3\lib\site-packages\allauth\account\templatetags\account_tags.py:4: DeprecationWarning: {% load account_tags %} is deprecated, use {% load account %}
DeprecationWarning)
C:\Users\kanwal\AppData\Local\Programs\DeepLearningStudio\conda3\lib\site-packages\allauth\socialaccount\templatetags\socialaccount_tags.py:4: DeprecationWarning: {% load socialaccount_tags %} is deprecated, use {% load socialaccount %}
" {% load socialaccount %}", DeprecationWarning)


#7


#8

As per log, GPU is supported on system. Which GPU you have?

It is possible that DLS is not able to extract GPU usage information.

You should try using some utility which shows GPU usage in details.

Can you share full screenshot of the training page?


#9


#10


#11

i Think it is quite visible that the app is not utlizing GPU from task manager! I have Nvidia 940MX


#12

i am having same problem please give some solution


#13

Do you have GPU dropdown in the training tab ? If not, try refreshing the page. If GPU dropdown is not getting enabled, training will not run on GPU.

if you dont see the GPU dropdown, please type http://127.0.0.1:8881/status in browser and let me know what output you see.


#14

I am having the same issue
There is no GPU dropdown in the training tab and this is the output of http://127.0.0.1:8881/status

{
“command”: “train”,
“gpus”: 0,
“inuse”: 1,
“project_id”: “0”,
“run_id”: “1”,
“uptime”: “2753.769702543177”
}

What should I do


#15

Hi, rajendra. I ran into a similar issue and haven’t found a solution yet. that’s my output:

{
“gpus”: 0,
“inuse”: 0,
“uptime”: “161445.1848081077”
}

Thanks


#16

Same problem here. I am on Windows 10. I was trying out a simple MNIST classification problem and I was surprised to see that no GPU activity on my Task Manager. CPU is heavily used.

Logs

Starting Deep Learning Studio...

Checking GPU support…
GPU not supported


#17

Outcome of http://127.0.0.1:8881/status

{
gpus: 0,
inuse: 0,
uptime: "186.51027155379577"
}

#18

same here

{
“gpus”: 0,
“inuse”: 0,
“uptime”: “555.5794758”
}


#19

Same issue here.

  1. Installed DLS 2.5.0 with GPU support+CUDA 9.0+Visual Studio.
  2. Log File display:
    Starting Deep Learning Studio…
    Checking GPU support…
    GPU supported
  3. No GPU load on training on simple MNIST classification problem. CPU usage 100%
  4. DSL Status:
    {
    “command”: “train”,
    “gpus”: 0,
    “inuse”: 1,
    “project_id”: “0”,
    “run_id”: “1”,
    “uptime”: “1153.4313307929663”
    }

Any help would be appreciated.


#20

Hi
which graphic card you are using?