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.

Sign up for GitHub

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

Jump to bottom

For several outfld(), use a temporary array to avoid computing on values not needed #1501

Merged
merged 2 commits into from
May 9, 2017
Merged

For several outfld(), use a temporary array to avoid computing on values not needed #1501

merged 2 commits into from
May 9, 2017

Conversation

ndkeen
Copy link
Contributor

@ndkeen ndkeen commented May 7, 2017

For several outfld() calls that do array operations in the function call,
instead create a temporary array, and only do the operation on ncol values.

For example:
!call outfld( 'RTP2_CLUBB', rtp2*1000._r8, pcols, lchnk )
tmp_array = rtp2(:ncol,:)*1000._r8
call outfld( 'RTP2_CLUBB', tmp_array, ncol, lchnk )

We hit issues using a couple of different PE layouts in these outfld calls. Additionally, for cori-knl, the SMS_D_Ld1.ne16_ne16.FC5ATMMOD test was failing, but now passes.

Fixes #1263

@ndkeen
For several outfld() calls that do array operations in the function c…
4dee5ee 
…all,

instead create a temporary array, and only do the operation on ncol values.

For example:
!call outfld( 'RTP2_CLUBB',       rtp2*1000._r8,           pcols, lchnk )
tmp_array = rtp2(:ncol,:)*1000._r8
call outfld( 'RTP2_CLUBB',       tmp_array,               ncol,  lchnk )
@ndkeen ndkeen requested a review from singhbalwinder May 7, 2017 23:14
@ndkeen ndkeen added Atmosphere BFB PR leaves answers BFB labels May 7, 2017
singhbalwinder
singhbalwinder requested changes May 8, 2017
View reviewed changes
components/cam/src/physics/cam/clubb_intr.F90 Outdated
call outfld( 'WPTHVP_CLUBB', wpthvp*cpair, pcols, lchnk )
call outfld( 'ZT_CLUBB', 1._r8*zt_out, pcols, lchnk )
call outfld( 'ZM_CLUBB', 1._r8*zi_out, pcols, lchnk )
tmp_array = wprcp(:ncol,:)*cpair
Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Should it be tmp_array = wpthvp(:ncol,:)*cpair?

Copy link
Contributor Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yes, it should be. Really sorry about that.

Copy link
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

no worries, most likely a copy-paste error.

@ndkeen
Correct a copy-paste error in first attempt.
67ed2c0 
Should be tmp_array = wpthvp(:ncol,:)*cpair
ndkeen added a commit that referenced this pull request May 8, 2017
@ndkeen
Merge branch 'ndk/cam/add-tmp_arrays-for-outfld3' into next (PR #1501)
1efed93 
For several outfld() calls that do array operations in the function call,
instead create a temporary array, and only do the operation on ncol values.

For example:
!call outfld( 'RTP2_CLUBB', rtp2*1000._r8, pcols, lchnk )
tmp_array = rtp2(:ncol,:)*1000._r8
call outfld( 'RTP2_CLUBB', tmp_array, ncol, lchnk )

We hit issues using a couple of different PE layouts in these outfld calls. Additionally, for cori-knl, the SMS_D_Ld1.ne16_ne16.FC5ATMMOD test was failing, but now passes.

Fixes #1263
@ndkeen ndkeen merged commit dd54f30 into master May 9, 2017
ndkeen added a commit that referenced this pull request May 10, 2017
@ndkeen
This is a fix to a PR #1501 (github issue #1263).
cf9e47a 
I allocated the tmp_array to be too large and GNU compiler stopped with an array mismatch.
Change this to be state%ncol in size, not pcol
@ndkeen ndkeen mentioned this pull request May 10, 2017
Merged
singhbalwinder added a commit that referenced this pull request May 10, 2017
@singhbalwinder
Merge branch 'ndk/cam/fix-add-tmp-arrays-use-ncol' into next (PR #1521)
cee0ac4 
Fixes tmp_array size to avoid array mismatch found by GNU

This is a fix to a PR #1501 (github issue #1263).
 I allocated the tmp_array to be too large and GNU compiler stopped
 with  an array mismatch.
 Change this to be state%ncol in size, not pcol

 Fixes #1263

[BFB]

* ndk/cam/fix-add-tmp-arrays-use-ncol:
  This is a fix to a PR #1501 (github issue #1263). I allocated the
  tmp_array to be too large and GNU compiler stopped with an array
  mismatch. Change this to be state%ncol in size, not pcol Add a
  more reasonable pe layout for the a%ne30np4 grid for cori-knl.
  May not be optimal, but allows for testing.
singhbalwinder added a commit that referenced this pull request May 11, 2017
@singhbalwinder
Merge branch 'ndk/cam/fix-add-tmp-arrays-use-ncol' (PR #1521)
c5d6848 
Fixes tmp_array size to avoid array mismatch found by GNU

This is a fix to a PR #1501 (github issue #1263).
 I allocated the tmp_array to be too large and GNU compiler stopped
 with  an array mismatch.
 Change this to be state%ncol in size, not pcol

 Fixes #1263

[BFB]

* ndk/cam/fix-add-tmp-arrays-use-ncol:
  This is a fix to a PR #1501 (github issue #1263). I allocated the
  tmp_array to be too large and GNU compiler stopped with an array
  mismatch. Change this to be state%ncol in size, not pcol Add a
  more reasonable pe layout for the a%ne30np4 grid for cori-knl.
  May not be optimal, but allows for testing.
jgfouca added a commit that referenced this pull request May 15, 2017
@jgfouca
Merge branch 'master' into agsalin/update-to-cime5.3.1
8593b23 
Upstream merge to resolve conflicts.

* master: (93 commits)
  Adds the --force-move option and implies --copy-only when --last-date is specified without --force-move
  added spunup A_WCYCL options (1850S and 2000S) for v0atm
  Updates land initial conditions for ne120_oRRS18v3
  Fix jenkins_generic_job when user selects specific machines.
  Fix small problems with MPAS components from debug tests
  Adds a warning when using the --last-date option and to its help
  Implement the copy_only option for short term archiving. This copies files rather than moving them
  Implemented most of the machinery for testing with "incomplete" log files
  Fix code format issue - replace unused variable with _
  Update template.st_archive
  Adds options to the st_archive to specify the last date (--last-date) to archive, and whether to disable archiving incomplete log files (--no-incomplete-logs)
  Turn off salinity restoring by default until it passes exact restart
  This is a fix to a PR #1501 (github issue #1263). I allocated the tmp_array to be too large and GNU compiler stopped with an array mismatch. Change this to be state%ncol in size, not pcol
  Fix a floating invalid that we found with certain PE's (and with intel v17.02)
  bug fix
  Revert mpas-o commit with new threaded vector reconstruction consistency changes
  Changing jobmin to 1 for batch q on blues
  Modifies SMS_D_Ln5_P8x4 to SMS_Ln5 to avoid cetus and melvin issues
  Point at corrected salinity restoring files for oEC60to30v3 and oRRS18to6v3
  Make sure clean clm cleans up obj dir
  ...
jgfouca pushed a commit that referenced this pull request Jun 2, 2017
@jedwards4b
Merge pull request #1501 from ESMCI/mvertens/somfix
10fbc43 
Fix for SOM bug introduced in PR #1451 This PR fixes the SOM bug introduced in PR#1451

Test suite: scripts_regression_tests and

verified that the following two tests passed functionality and were bfb with d4dfbf1 (the CIME PR
before the aquaplanet SOM changes were introduced)
ERS_D_Ld5.f19_g16.ETEST.yellowstone_intel.cice-default
ERS_Ld5.f19_g16.ETEST.yellowstone_intel.cice-default
verified that the following two tests passed functionality (the QSC4 compset is a new CAM compset that has aquaplanet SOM functionality)
ERS_D_Ld5.f09_f09.QSC4.yellowstone_intel
ERS_Ld5.f09_f09.QSC4.yellowstone_intel
Test baseline: d4dbf1
Test namelist changes: None
Test status: bit for bit
Fixes #1497
User interface changes?: None
Code review:jedwards
jgfouca added a commit that referenced this pull request Jun 2, 2017
@jgfouca
Merge branch 'master' into agsalin/update-to-cime5.3.1
0bb7c3a 
Upstream merge to resolve conflicts.

* master: (93 commits)
  Adds the --force-move option and implies --copy-only when --last-date is specified without --force-move
  added spunup A_WCYCL options (1850S and 2000S) for v0atm
  Updates land initial conditions for ne120_oRRS18v3
  Fix jenkins_generic_job when user selects specific machines.
  Fix small problems with MPAS components from debug tests
  Adds a warning when using the --last-date option and to its help
  Implement the copy_only option for short term archiving. This copies files rather than moving them
  Implemented most of the machinery for testing with "incomplete" log files
  Fix code format issue - replace unused variable with _
  Update template.st_archive
  Adds options to the st_archive to specify the last date (--last-date) to archive, and whether to disable archiving incomplete log files (--no-incomplete-logs)
  Turn off salinity restoring by default until it passes exact restart
  This is a fix to a PR #1501 (github issue #1263). I allocated the tmp_array to be too large and GNU compiler stopped with an array mismatch. Change this to be state%ncol in size, not pcol
  Fix a floating invalid that we found with certain PE's (and with intel v17.02)
  bug fix
  Revert mpas-o commit with new threaded vector reconstruction consistency changes
  Changing jobmin to 1 for batch q on blues
  Modifies SMS_D_Ln5_P8x4 to SMS_Ln5 to avoid cetus and melvin issues
  Point at corrected salinity restoring files for oEC60to30v3 and oRRS18to6v3
  Make sure clean clm cleans up obj dir
  ...
@ndkeen ndkeen deleted the ndk/cam/add-tmp_arrays-for-outfld3 branch February 5, 2018 20:51
jgfouca added a commit that referenced this pull request Feb 27, 2018
@jgfouca
Merge branch 'master' into agsalin/update-to-cime5.3.1
ab82990 
Upstream merge to resolve conflicts.

* master: (93 commits)
  Adds the --force-move option and implies --copy-only when --last-date is specified without --force-move
  added spunup A_WCYCL options (1850S and 2000S) for v0atm
  Updates land initial conditions for ne120_oRRS18v3
  Fix jenkins_generic_job when user selects specific machines.
  Fix small problems with MPAS components from debug tests
  Adds a warning when using the --last-date option and to its help
  Implement the copy_only option for short term archiving. This copies files rather than moving them
  Implemented most of the machinery for testing with "incomplete" log files
  Fix code format issue - replace unused variable with _
  Update template.st_archive
  Adds options to the st_archive to specify the last date (--last-date) to archive, and whether to disable archiving incomplete log files (--no-incomplete-logs)
  Turn off salinity restoring by default until it passes exact restart
  This is a fix to a PR #1501 (github issue #1263). I allocated the tmp_array to be too large and GNU compiler stopped with an array mismatch. Change this to be state%ncol in size, not pcol
  Fix a floating invalid that we found with certain PE's (and with intel v17.02)
  bug fix
  Revert mpas-o commit with new threaded vector reconstruction consistency changes
  Changing jobmin to 1 for batch q on blues
  Modifies SMS_D_Ln5_P8x4 to SMS_Ln5 to avoid cetus and melvin issues
  Point at corrected salinity restoring files for oEC60to30v3 and oRRS18to6v3
  Make sure clean clm cleans up obj dir
  ...
jgfouca added a commit that referenced this pull request Mar 14, 2018
@jgfouca
Merge branch 'master' into agsalin/update-to-cime5.3.1
e026242 
Upstream merge to resolve conflicts.

* master: (93 commits)
  Adds the --force-move option and implies --copy-only when --last-date is specified without --force-move
  added spunup A_WCYCL options (1850S and 2000S) for v0atm
  Updates land initial conditions for ne120_oRRS18v3
  Fix jenkins_generic_job when user selects specific machines.
  Fix small problems with MPAS components from debug tests
  Adds a warning when using the --last-date option and to its help
  Implement the copy_only option for short term archiving. This copies files rather than moving them
  Implemented most of the machinery for testing with "incomplete" log files
  Fix code format issue - replace unused variable with _
  Update template.st_archive
  Adds options to the st_archive to specify the last date (--last-date) to archive, and whether to disable archiving incomplete log files (--no-incomplete-logs)
  Turn off salinity restoring by default until it passes exact restart
  This is a fix to a PR #1501 (github issue #1263). I allocated the tmp_array to be too large and GNU compiler stopped with an array mismatch. Change this to be state%ncol in size, not pcol
  Fix a floating invalid that we found with certain PE's (and with intel v17.02)
  bug fix
  Revert mpas-o commit with new threaded vector reconstruction consistency changes
  Changing jobmin to 1 for batch q on blues
  Modifies SMS_D_Ln5_P8x4 to SMS_Ln5 to avoid cetus and melvin issues
  Point at corrected salinity restoring files for oEC60to30v3 and oRRS18to6v3
  Make sure clean clm cleans up obj dir
  ...
rljacob pushed a commit that referenced this pull request Apr 12, 2021
@jgfouca
Merge branch 'master' into agsalin/update-to-cime5.3.1
2e84b48 
Upstream merge to resolve conflicts.

* master: (93 commits)
  Adds the --force-move option and implies --copy-only when --last-date is specified without --force-move
  added spunup A_WCYCL options (1850S and 2000S) for v0atm
  Updates land initial conditions for ne120_oRRS18v3
  Fix jenkins_generic_job when user selects specific machines.
  Fix small problems with MPAS components from debug tests
  Adds a warning when using the --last-date option and to its help
  Implement the copy_only option for short term archiving. This copies files rather than moving them
  Implemented most of the machinery for testing with "incomplete" log files
  Fix code format issue - replace unused variable with _
  Update template.st_archive
  Adds options to the st_archive to specify the last date (--last-date) to archive, and whether to disable archiving incomplete log files (--no-incomplete-logs)
  Turn off salinity restoring by default until it passes exact restart
  This is a fix to a PR #1501 (github issue #1263). I allocated the tmp_array to be too large and GNU compiler stopped with an array mismatch. Change this to be state%ncol in size, not pcol
  Fix a floating invalid that we found with certain PE's (and with intel v17.02)
  bug fix
  Revert mpas-o commit with new threaded vector reconstruction consistency changes
  Changing jobmin to 1 for batch q on blues
  Modifies SMS_D_Ln5_P8x4 to SMS_Ln5 to avoid cetus and melvin issues
  Point at corrected salinity restoring files for oEC60to30v3 and oRRS18to6v3
  Make sure clean clm cleans up obj dir
  ...
rljacob pushed a commit that referenced this pull request May 6, 2021
@jgfouca
Merge branch 'master' into agsalin/update-to-cime5.3.1
b0d5e6c 
Upstream merge to resolve conflicts.

* master: (93 commits)
  Adds the --force-move option and implies --copy-only when --last-date is specified without --force-move
  added spunup A_WCYCL options (1850S and 2000S) for v0atm
  Updates land initial conditions for ne120_oRRS18v3
  Fix jenkins_generic_job when user selects specific machines.
  Fix small problems with MPAS components from debug tests
  Adds a warning when using the --last-date option and to its help
  Implement the copy_only option for short term archiving. This copies files rather than moving them
  Implemented most of the machinery for testing with "incomplete" log files
  Fix code format issue - replace unused variable with _
  Update template.st_archive
  Adds options to the st_archive to specify the last date (--last-date) to archive, and whether to disable archiving incomplete log files (--no-incomplete-logs)
  Turn off salinity restoring by default until it passes exact restart
  This is a fix to a PR #1501 (github issue #1263). I allocated the tmp_array to be too large and GNU compiler stopped with an array mismatch. Change this to be state%ncol in size, not pcol
  Fix a floating invalid that we found with certain PE's (and with intel v17.02)
  bug fix
  Revert mpas-o commit with new threaded vector reconstruction consistency changes
  Changing jobmin to 1 for batch q on blues
  Modifies SMS_D_Ln5_P8x4 to SMS_Ln5 to avoid cetus and melvin issues
  Point at corrected salinity restoring files for oEC60to30v3 and oRRS18to6v3
  Make sure clean clm cleans up obj dir
  ...
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@singhbalwinder singhbalwinder singhbalwinder approved these changes

Assignees

@singhbalwinder singhbalwinder

Labels
Atmosphere BFB PR leaves answers BFB
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
2 participants
@ndkeen @singhbalwinder