System Crashes while using eIQ portal

cancel
Showing results for 
Search instead for 
Did you mean: 

System Crashes while using eIQ portal

Jump to solution
1,672 Views
Ramson
Contributor IV

Hello Team,

When using eIQ portal, the CPU utilization is reaching 100% which causes system crash regularly. So I couldn't run the training continuously.

I do have a GeForce RTX 2060 GPU with CUDA 11.2.66 drivers installed as well. , but as you can see in the below image, the GPU utilization is almost 0%.

eiq_potal_cpu_util.jpg

My System Details are as follows:

 

Proccessor - Intel(R) Core(TM) i7-8700 CPU @3.20GHz, 3192 Mhz. 
RAM - 16 GB 
GPU - Nvidia GeForce RTX 2060
OS Name : Microsoft Windows 10 Pro

 

 

I couldn't find any documents explaining the system requirements and enabling GPU utilization for running eIQ portal. Please provide any information regarding this. 

 

Thanks in Advance,

Ramson Jehu K

Labels (1)
Tags (2)
0 Kudos
1 Solution
1,390 Views
david_piskula
NXP Employee
NXP Employee

Hello @Ramson;

just wanted to let you know that a new version of eIQ Toolkit (and eIQ Portal) has been released today: https://www.nxp.com/design/software/development-software/eiq-ml-development-environment/eiq-toolkit-...

This version supports TensorFlow 2.5.0.

Regards,

David

View solution in original post

0 Kudos
10 Replies
1,429 Views
david_piskula
NXP Employee
NXP Employee

Hello @Ramson,

are you still encountering the crashes when using the tool? Could you please provide me with the check log for inspection? You can find it in %AppData% -> Roaming\eIQ Portal\logs\main.log

As for the issue with TensorFlow, have you been able to solve it? I'm afrad I won't be able to help with TensorFlow build errors if even Google didn't understand your issue.

Regards,

David

0 Kudos
1,423 Views
Ramson
Contributor IV

Hi @david_piskula , 

I will update the log soon. Regarding the tensorflow, still I'm not able to build it. But I have doubt whether Its necessary to build tensorflow from source for running eIQ Portal?

Regards,

Ramson Jehu K

0 Kudos
1,418 Views
david_piskula
NXP Employee
NXP Employee

Hi @Ramson,

you don't have to build TF from source. eIQ Portal comes preinstalled with TensorFlow. You can check the version by opening the command line from eIQ Portal and using the following commands:

david_piskula_0-1634652507556.png

 

0 Kudos
1,414 Views
Ramson
Contributor IV

Hi @david_piskula ,

Thanks for clarifying. I will update you with the log as soon as possible.

Regards 

Ramson Jehu K

0 Kudos
1,391 Views
david_piskula
NXP Employee
NXP Employee

Hello @Ramson;

just wanted to let you know that a new version of eIQ Toolkit (and eIQ Portal) has been released today: https://www.nxp.com/design/software/development-software/eiq-ml-development-environment/eiq-toolkit-...

This version supports TensorFlow 2.5.0.

Regards,

David

0 Kudos
1,123 Views
Ramson
Contributor IV

Hi @david_piskula ,

Thanks for the update. we have been working the updated eIQ portal a while now. The system crash is not happening now. But we are facing a different issue which is mentioned in this link  https://community.nxp.com/t5/eIQ-Machine-Learning-Software/Validation-encountered-error/td-p/1372431... 

Thanks

0 Kudos
1,665 Views
david_piskula
NXP Employee
NXP Employee

Hello @Ramson,

thank you for your interest in eIQ Portal. I will need to report this issue to the development team. Could you please also provide more information about your dataset (either what it is exactly, if it's publicly available, or at least image size and dataset size) and the exact software version you are using? You can find the version in Help -> About eIQ Portal.

Best Regards,

David

0 Kudos
1,664 Views
Ramson
Contributor IV

Hi @david_piskula ,

We are using VOC dataset, which is mentioned in this Au-zone's video: https://youtu.be/gbaeK8hUdHA .

The eIQ Portal Version is 2.1.30. 

Can you provide details about system requirements and how to enable GPU utilization?

Thanks 

0 Kudos
1,652 Views
david_piskula
NXP Employee
NXP Employee

Hello @Ramson,

the Cuda driver version depends on the TensorFlow version supported by eIQ Portal. In the currently released eIQ Portal, the TensorFlow version is 2.3.2.

Please try installing the v7.6 cuDNN and CUDA 10.2 and let me know if that helps with enabling the training on your GPU.

david_piskula_0-1631799463811.png

https://www.tensorflow.org/install/source#gpu

Best Regards,

David

0 Kudos
1,605 Views
Ramson
Contributor IV

Hi @david_piskula ,

I have been struggling for the past few days building TF v2.3.2 with CUDA 10.2 and v7.6 cuDNN. I'm facing the following issue. 

external/com_google_absl\absl/time/clock.h(70): error C2065: 'Duration': undeclared identifier 
external/com_google_absl\absl/time/clock.h(70): error C2146: syntax error: missing ')' before identifier 'duration' external/com_google_absl\absl/time/clock.h(70): error C2143: syntax error: missing ';' before '{' 
external/com_google_absl\absl/time/clock.h(70): error C2447: '{': missing function header (old-style formal list?) Target 
//tensorflow/tools/pip_package:build_pip_package failed to build INFO: Elapsed time: 899.084s, Critical Path: 504.33s INFO: 1352 processes: 1352 local. FAILED: Build did NOT complete successfully

I have also posted this in Tensorflow github : https://github.com/tensorflow/tensorflow/issues/52092 .

Please help if have came across this error before while building.

Thanks in Advance 

Ramson Jehu K

0 Kudos