Intel® oneAPI Data Analytics Library
Community support for building compute-intensive applications that run fast on Intel® architecture.
Announcements
Welcome to the Intel Community. If you get an answer you like, please mark it as an Accepted Solution to help others. Thank you!
204 Discussions

Decision forest regression crash when observationsPerTreeFraction is not 1

Jason_Jones
Beginner
287 Views

If I add the line:

algorithm.parameter.observationsPerTreeFraction = 0.8;

to the example file df_reg_dense_batch.cpp which comes with the DAAL installation, the application will crash every time. I usually get an error message like this:

*** Error in `_results/gnu_intel64_parallel_a/df_reg_dense_batch.exe': double free or corruption (!prev): 0x00007f1c240034b0 ***

I have had this problem while working on a custom app and seen it with DAAL 2019 on macOS and also with DAAL 2020 on Linux. It crashes in all cases where the observationsPerTreeFraction is set to a number less than 1.0.

For the sake of a specific repro case:

Compile line: make libintel64 example=df_reg_dense_batch compiler=gnu
Example file: df_reg_dense_batch.cpp
Compiler: gcc (Ubuntu 5.4.0-6ubuntu1~16.04.12) 5.4.0 20160609
OS: Ubuntu 16.04.6 LTS

Thanks for looking into this!

0 Kudos
1 Reply
Gennady_F_Intel
Moderator
287 Views

thanks, Jason. We will check the problem on our side and get back to you asap.

Reply