Closed emorway-usgs closed 5 years ago
Took a few commits to work on this issue: 7a6fbca c9a7368 75aba17 9f020a5 eaa42ca 101d775 23b4a65 8e7f5a3 4bdfa74 a130cd1 3f54657
This problem is now running successfully on Travis with the changes contained in the commits referenced above. Intel was running the model all the way through even though a variable was 'infinite', so it took temporary print statements to finally figure out what was happening.
Vivek started getting an error with a model that works with MT3DMS 5.3. Turns out the model runs when TTSMULT=1.0 in the BTN input file; however, when equal to 1.2 in this particular model, the distributed MT3D-USGS_64.exe bombs.
Vivek and I noted that when run from Visual Studio in "Release" mode, with TTSMULT=1.2, the model will bomb when the following compile flag is set equal to "Underflow gives 0.0; Abort on other IEEE exceptions (/fpe:0)" (more to follow image):
The line of code that is bombing is shown below, DTRANS is getting really large by the time the model gets to transport step ~447 in the code below:
A model for debugging this problem was added with commit a226eee (11/8/16)