error when using vtune to profile tensorflow file

[lethean]

I tried to use vtune to profile my tensorflow python file:

vtune --collect hotspot python train.py

and got this error:

vtune: Error: Valid setenv symbol is not found in the binary of the analysis target.
vtune: Error: Binary file of the analysis target does not contain symbols required for profiling. See the 'Analyzing Statically Linked Binaries' help topic for more details.

and if I delete code related to tensorflow, the error will not occur.

Any help to solve this problem is apprciated ! 

[JyothisV_Intel]

Hi,

Good day to you.

Thanks for posting in Intel Communities.

We tried profiling a code that contains TensorFlow from our side and was unable to get the above mentioned error.

To assist you better, can you get back to us with the following information so that we can try replicate the exact that you are facing from our side:

1. A sample reproducer code that exhibits this issue

2. The VTune version that you are using. (Type vtune -V in a terminal window in a terminal window to get this information)

3. The Python version that you are using. (Type python -V in a terminal window to get this information)

4. The TensorFlow version that you are using. (Type python -c "import tensorflow as tf; print(tf.__version__)" in a terminal window to get this information)

5. The processor model and the OS name and version in which you are running the Intel VTune profiler

Regards,

Jyothis V James

[lethean]

1. 

import tensorflow as tf

w1=tf.Variable(tf.random_normal(([2,3]),stddev=1.0,seed=1) ) 

2.  Intel(R) VTune(TM) Profiler 2022.2.0 (build 623516) Command Line Tool

3.  Python 3.6.12 :: Anaconda, Inc.

4.  1.15.5

5. Model name: Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GHz 

    OS：Ubuntu 18.04.5 LTS

 Thanks！

[JyothisV_Intel · Accepted solution]

Hi,

Good day to you.

Thanks for getting back to us with all the necessary details. We tried replicating your issue from our side on a system with the following environment configuration and was able to get the same error.

(tf) :~$ vtune -V
Intel(R) VTune(TM) Profiler 2022.3.0 (build 624050) Command Line Tool
Copyright (C) 2009 Intel Corporation. All rights reserved.

(tf) :~$ python -V
Python 3.6.12 :: Anaconda, Inc.

(tf) :~$ python -c "import tensorflow as tf; print(tf.__version__)"
1.15.0

You are getting this error since the environment in which you are trying to perform Intel VTune analysis is using an outdated end-of-life Python version (Refer https://endoflife.date/python)

Kindly try upgrading the Python version of the environment in which you are running the code or analysis using the command below:

conda install python==3.7

The issue is not present when using Python version 3.7 and above as shown below:

(tf) :~$ cat train.py
import tensorflow as tf
w1=tf.Variable(tf.random_normal(([2,3]),stddev=1.0,seed=1))

(tf) :~$ vtune -V
Intel(R) VTune(TM) Profiler 2022.3.0 (build 624050) Command Line Tool
Copyright (C) 2009 Intel Corporation. All rights reserved.

(tf) :~$ python -V
Python 3.7.0

(tf) :~$ python -c "import tensorflow as tf; print(tf.__version__)"
1.15.0

(tf) :~$ vtune --collect hotspots python train.py
vtune: Collection started. To stop the collection, either press CTRL-C or enter from another console window: vtune -r /home/jyothisvjames/r002hs -command stop.
WARNING:tensorflow:From train.py:2: The name tf.random_normal is deprecated. Please use tf.random.normal instead.
vtune: Collection stopped.

If this resolves your issue, make sure to accept this as a solution. This would help others with similar issues.

Thanks and Regards,

Jyothis V James

[JyothisV_Intel]

Hi,

Good day to you.

We have not heard back from you. Has the work-around solution helped you resolve your issue? If yes, kindly accept this as a solution as it will help others with a similar issue; else do let us know so that we can assist you better.

Thanks and Regards,

Jyothis V James

[lethean]

It works! Thanks so much!!!

[JyothisV_Intel]

Hi,

Good day to you.

Thanks for accepting our solution and we are glad to know that your issue is resolved.

If you need any additional information, please post a new question as this thread will no longer be monitored by Intel.

Thanks and Regards,

Jyothis V James