Segmentation Fault for "tcgen" using the Test Case

Submitted by louiskstse on Wed, 08/04/2021 - 19:58
Forum: GFDL Vortex Tracker Support | General

Hello,

 

I tried to run the Test Case using The GFDL vortex tracker (version 3.9a), it was successful for both multi/ and onebig/ using "tracker" mode.

However, when I followed the steps listed in P.17 of the Users Guide for tracking storms in "tcgen" mode, segmentation fault occurred

If I am not mistaken, I just need to do three steps:

1. Set trkrinfo%type = "tcgen"

2. Set trkrinfo%westbd, trkrinfo%eastbd, trkrinfo%southbd, trkrinfo%northbd

3. Create a blank TCVital file fort.14

I have been trying different settings for a few weeks, but segmentation fault still occurs

 

Best regards,

Louis

input_namelist.pdf , output_segmentation_fault.pdf
Permalink

Submitted by biswas on Wed, 08/04/2021 - 22:40

Hi Louis,

The steps you are following are correct. What kind of files are you using for tcgen? Have you tried with multi files configuration? I tried with you setting with GFS files and they seems to work fine. Do you want your domain to be small as in the namelist?

Thanks

Biswas

Permalink

Submitted by louiskstse on Thu, 08/05/2021 - 01:05

Hi Biswas,

Thank you for your reply.

I tried using multi files configuration as well, and the same error occurred (segmentation fault). I am using the files in the Test Case, i.e. hwrf_onebig.25x25.AL142016.2016100400.pgrb, so it is a little bit surprised to see segmentation fault. Actually I also tried to track the storms using a regridded MPAS file, same situation happens in which the "tracker" mode works fine, but "tcgen" mode fails due to segmentation fault.

If it is confirmed that the steps are correct, I am wondering if the tracker was indeed not completely built.

Thanks,

Louis

Permalink

Submitted by louiskstse on Sun, 08/08/2021 - 22:04

It is found that my vortex tracker, as well as the dependencies (NetCDF, PNetCDF...) were compiled using Intel compilers 2018. It seems to be too new as compared to the version mentioned in the User Guide (Intel v15.0.3), do you think it is a possible reason of getting segmentation fault?

Thanks,

Louis

Permalink

Submitted by lucia.cisco on Mon, 08/23/2021 - 03:59

Hi Louis,

Did you find a solution for your problem? I tried to do the same and I got the same error, even though my compiler is Intel 2021. If you succeeded in solving this issue, could you please tell me what you figured out? Probably I am doing the same error.

Thank you in advance,

Lucia 

Permalink

Submitted by louiskstse on Tue, 08/24/2021 - 01:29

Hi Lucia,

I haven't solve the issue yet. But based on my experience on compiling different applications, it seems that the compilers (e.g. 2021) are too new. As some compilers were not backward compatible, therefore older version should work, but the more recent versions do not work.

The user guide mentions that "the vortex tracker has been tested on Linux PGI v14.3 and Linux Intel v15.0.3". Maybe you can try to build the tracker using these compilers. However, I suspect that you might need to rebuild all libraries, including HDF5, NetCDF, PNetCDF,... using the compilers in older version as well.

Thank you,

Louis

Sitemap

Contact

303.497.8921
dtcenter@ucar.edu

Addresses

mailing:
PO Box 3000
Boulder, CO 80307-3000

Shipping:
3090 Center Green Dr.
Boulder, CO 80301

Follow Us

YouTube Icon
Twitter Icon

Copyright © 2024. All rights reserved.