Test Failures after Building Dalton 2020

Problems with Dalton installation? Find answers or ask for help here
Post Reply
rlt92
Posts: 15
Joined: 20 Nov 2018, 12:29
First name(s): Richard
Last name(s): Thurston
Affiliation: UC Berkeley
Country: United States

Test Failures after Building Dalton 2020

Post by rlt92 » 10 Apr 2020, 10:14

Hi, I've been attempting to build Dalton on my available cluster but I've been running into test failure issues when I've been trying to build Dalton. I've been able to get a working version of Dalton running on my local machine (Dalton 2018) using serial math and serial gnu compilers but I haven't been able to get a completely working version of Dalton working on the cluster. I've tried a variety of different build configurations using the different available software on the cluster but nothing I've tried has resulted in all of the tests passing for Dalton 2020. Attached are the output files associated with some of my trial builds including a trail using the builtin math and serial gnu compilers. Any advice in fixing this issue would be appreciated.
Attachments
intel-mkl-intel-openmpi.out
(160.05 KiB) Downloaded 78 times
builtin_math-gnu-serial.out
(370.37 KiB) Downloaded 86 times
available_modules.txt
(6.16 KiB) Downloaded 81 times

User avatar
magnus
Posts: 524
Joined: 27 Jun 2013, 16:32
First name(s): Jógvan Magnus
Middle name(s): Haugaard
Last name(s): Olsen
Affiliation: Aarhus University
Country: Denmark

Re: Test Failures after Building Dalton 2020

Post by magnus » 10 Apr 2020, 10:28

Dalton2020 has not been released yet so expect issues like this. I would recommend that you do not use it for production calculations.

rlt92
Posts: 15
Joined: 20 Nov 2018, 12:29
First name(s): Richard
Last name(s): Thurston
Affiliation: UC Berkeley
Country: United States

Re: Test Failures after Building Dalton 2020

Post by rlt92 » 11 Apr 2020, 06:39

Thanks for the reply, unfortunately I've had similar issues with Dalton 2018 and Dalton 2016. I'm concerned that there is a problem with the versions of the compilers or the math libraries that I'm using for building Dalton. Attached is a recent output file for a build of 2018 Dalton using Intel openmpi compilers and an mkl library.
Attachments
dalton2018_intel-mkl-intel-openmpi.out
(150.72 KiB) Downloaded 78 times

taylor
Posts: 599
Joined: 15 Oct 2013, 05:37
First name(s): Peter
Middle name(s): Robert
Last name(s): Taylor
Affiliation: Tianjin University
Country: China

Re: Test Failures after Building Dalton 2020

Post by taylor » 11 Apr 2020, 09:17

(I am commenting here only on your issues with Dalton2018)

I am not sure that tests running out of time is a cause for concern, but out of eight failures in your posted output there are three that are not timeouts. It is difficult to comment on what may be wrong without knowing why those tests failed --- can you post more specific information on these, please?

Best regards
Pete

rlt92
Posts: 15
Joined: 20 Nov 2018, 12:29
First name(s): Richard
Last name(s): Thurston
Affiliation: UC Berkeley
Country: United States

Re: Test Failures after Building Dalton 2020

Post by rlt92 » 11 Apr 2020, 09:59

For the Dalton 2018 build, here is the log for the benchmark_eri_atzs test that failed and the "LastTest" log file summarizing the errors, the other log files for the failed tests are not available. At first glace the benchmark test failed due to an out of memory condition. The other two tests appear to have failed due to a UnicodeDecodeError.
Attachments
benchmark_eri_atzs.log
(104.18 KiB) Downloaded 80 times
LastTest.log
(1.03 MiB) Downloaded 79 times

User avatar
magnus
Posts: 524
Joined: 27 Jun 2013, 16:32
First name(s): Jógvan Magnus
Middle name(s): Haugaard
Last name(s): Olsen
Affiliation: Aarhus University
Country: Denmark

Re: Test Failures after Building Dalton 2020

Post by magnus » 11 Apr 2020, 10:47

I would say that your build is ok.

The benchmark tests time out, and the one that doesn't time out, runs out of memory. Also those tests have not been maintained and therefore may not be up-to-date with recent modifications of the output etc.

The other two tests are probably failing due to Python 2 vs 3 issues. Both tests are the only ones that use the latest version of the runtest library.

hjaaj
Posts: 394
Joined: 27 Jun 2013, 18:44
First name(s): Hans Jørgen
Middle name(s): Aagaard
Last name(s): Jensen
Affiliation: Universith of Southern Denmark
Country: Denmark

Re: Test Failures after Building Dalton 2020

Post by hjaaj » 11 Apr 2020, 10:56

I looked at the LastTest.log, and the two errors that are not timeout were both in the python script testing for errors, not in Dalton. According to the log they aborted because there is a non-ascii character (0xe2). From your build directory you need to check a test directory of one of the failed tests, e.g. (from your build/ diractory) "cd test/dft_b3lyb_cart" and look at the various files there to find the problem.

Something different: to avoid timeouts for sequential tests you can exclude the benchmark tests with "ctest -LE benchmark", you can also increase the timeout limit with e.g. "ctest --timeout 3600" for 1 hour timeout limit.

rlt92
Posts: 15
Joined: 20 Nov 2018, 12:29
First name(s): Richard
Last name(s): Thurston
Affiliation: UC Berkeley
Country: United States

Re: Test Failures after Building Dalton 2020

Post by rlt92 » 11 Apr 2020, 19:17

Thanks for the replies. Initially I built Dalton2018 using python 3.6 so I tried python 3.7 and that solved the UnicodeDecodeError problem. The out of memory issue is still there but since I didn't try anything to fix that problem I guess that's to be expected.

Post Reply

Who is online

Users browsing this forum: No registered users and 0 guests