Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

cime nightly tests should test all supported compilers on the test machines #1770

Closed
billsacks opened this issue Jul 28, 2017 · 4 comments
Closed

Comments

@billsacks
Copy link
Member

My understanding is that the cime nightly tests - and scripts_regression_tests in general - just tests the default compiler on the given machine. This fails to catch problems like #1769 . We should run the cime nightly tests across all supported compilers on the test machines, and possibly extend scripts_regression_tests so that it always runs at least one test from each supported compiler on the current machine.

I personally don't feel we need to run all system tests on all compilers. But ideally we'd have at least one debug test and one non-debug test on each compiler on each tested machine.

cc @jedwards4b @jgfouca @mvertens @rljacob @fischer-ncar

@jedwards4b
Copy link
Contributor

@billsacks I believe that any developer can add and maintain cdash tests - I would suggest that you add any additional testing you are interested in.

@rljacob
Copy link
Member

rljacob commented Aug 1, 2017

I think what @billsacks wants is for jenkins to automatically run tests on different compilers on a given machine. That requires messing with jenkins options and that's not open to this developer.

@jedwards4b
Copy link
Contributor

We don't use jenkins - we have a simple collection of cronjobs that run and submit results to cdash.

@jgfouca
Copy link
Contributor

jgfouca commented Aug 1, 2017

@jedwards4b yes, in that case the selection of compiler should be easy with an additional option to create_test. @billsacks is right though, the testing of non-default compilers is severely lacking in scripts_regression_tests.

rljacob added a commit that referenced this issue Aug 10, 2017
Add a --compiler and --mpilib options to scripts_regression_tests.py
to facilitate more testing with non-default compilers and mpilib's.

Test suite: scripts_regression_tests.py --compiler pgi on yellowstone + scripts_regression_tests.py

scripts_regression_tests.py --compiler gnu --mpilib openmpi on cheyenne
Test baseline:
Test namelist changes:
Test status: bit for bit
Addresses #1770

User interface changes?: Yes scripts_regression_tests.py now has --compiler and --mpilibs option
billsacks added a commit that referenced this issue Aug 11, 2017
Allow --compiler and --mpilib in ctest scripts

This PR allows ctest to use the new --compiler and --mpilib options in scripts_regression_tests.py
by using env variables CIME_COMPILER and CIME_MPILIB

Test suite: CIME_COMPILER=nag ./cime.ctest.hobart
Test baseline: 
Test namelist changes: 
Test status: bit for bit
Fixes #1770 

User interface changes?: 

Update gh-pages html (Y/N)?:

Code review: Bill Sacks
jgfouca added a commit that referenced this issue Oct 17, 2017
Add mach-specific throughput checking to test machines

Throughput comparison was previously done at 25% tolerance/deviation
from baseline. This PR adds machine-specific throughput tolerance of
10% for all test machines. This should enable detection of performance
regression of greater than 10% on machines and tests that have
baselines (e.g. melvin, skybridge, chama).

[BFB]

* origin/azamat/tests/check-tput:
  Add mach-specific throughput checking to test machines
jgfouca added a commit that referenced this issue Feb 23, 2018
Add mach-specific throughput checking to test machines

Throughput comparison was previously done at 25% tolerance/deviation
from baseline. This PR adds machine-specific throughput tolerance of
10% for all test machines. This should enable detection of performance
regression of greater than 10% on machines and tests that have
baselines (e.g. melvin, skybridge, chama).

[BFB]

* origin/azamat/tests/check-tput:
  Add mach-specific throughput checking to test machines
jgfouca added a commit that referenced this issue Mar 13, 2018
Add mach-specific throughput checking to test machines

Throughput comparison was previously done at 25% tolerance/deviation
from baseline. This PR adds machine-specific throughput tolerance of
10% for all test machines. This should enable detection of performance
regression of greater than 10% on machines and tests that have
baselines (e.g. melvin, skybridge, chama).

[BFB]

* origin/azamat/tests/check-tput:
  Add mach-specific throughput checking to test machines
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

No branches or pull requests

4 participants