Difference between pages "Bugfixes" and "ARCHER porting"
From UKCA
(Difference between pages)
(→RADAER update: add new cat) |
|||
| Line 1: | Line 1: | ||
| + | Several of the [[UKCA Release Jobs]] have been ported from HECToR to ARCHER, and their [[UKCA_Release_Jobs#Job_List|jobids]] are detailed on that page. |
||
| − | This page covers bug and bugfixes found in various UKCA branches or jobs. This page will be updated whenever a new bug/bugfix is found. |
||
| + | ==Timings== |
||
| − | If you have any concerns, please contact [[User:Nla27 | Luke Abraham]]. If you think that you have found a bug, please report it to him, or alternatively you can add it to the [[Talk:Bugfixes | bugfix talk page]] (or click on the ''discussion'' link above). |
||
| + | [[File:CCMI_timings_ARCHER.png|300px|thumb|right|Scaling and efficiency profiling for N48L60 UKCa CheS+ on ARCHER]] |
||
| − | =Compiler Bugs and Bugfixes= |
||
| + | The plot shows how UKCA scales when ported to ARCHER. The most efficient number of nodes to use is 3 (8x9), although the model will run fastest on 6 nodes (e.g. 16x9 or 12x12). Moving to 3 nodes from 6 will increase the run-time by 50%. |
||
| − | ===Cray compiler on HECToR XE6=== |
||
| + | ==Code Changes== |
||
| − | Please also see advice from [http://cms.ncas.ac.uk/index.php/component/content/1583?task=view NCAS-CMS]. |
||
| + | ===Cray cce Fortran Compiler=== |
||
| − | To allow the VN7.3_UKCA_CheM branch to compile, please include the following compiler file over-ride: |
||
| + | Corrections required for the Cray cce compiler on archer can be found on the [[Bugfixes#Cray_compiler_on_ARCHER|bugfixes page]]. |
||
| − | /home/ukca/comp_overides/hector_cce_7.3_asad_chem_diags |
||
| − | |||
| − | =Code Bugs and Bugfixes= |
||
| − | |||
| − | ===Corrections to VN7.3_UKCA_CheM branch=== |
||
| − | |||
| − | '''The latest revision number is 9278.''' |
||
| − | |||
| − | There have been four bugfixes to this branch, one correcting output units (revision 5887), one correcting conversion factors in any StratChem jobs (revision 6356), one correcting a bug win wet-deposition (revision 6791), and one fixing the <code>ukca_transform_halogen</code> routine (revision 9278) which affected the Total N tracer in Stratospheric Chemistry jobs. This also requires an update to the initialisation of this Total N tracer (tracer 4, 34004) - see [[Bugfixes#Total_N_tracer_initial_condition | below]]. |
||
| − | |||
| − | ===Scenario script on PUMA=== |
||
| − | |||
| − | This script is located at <code>/home/ukca/bin/scenario</code>. |
||
| − | |||
| − | This was giving incorrect values for CFC-11, CFC-12, and Methyl Bromide (CH3Br), the numbers being too high. This bug has been corrected as of 2011/10/18, but you should double-check your input scripts. |
||
| − | |||
| − | =UMUI Bugs and Bugfixes= |
||
| − | |||
| − | ==UMUI Panel Settings== |
||
| − | |||
| − | ===Total N tracer initial condition=== |
||
| − | |||
| − | The initial condition for the Total N tracer needs to be updated along with the fcm revision of the VN7.3_UKCA_CheM branch listed [[Bugfixes#Corrections_to_VN7.3_UKCA_CheM_branch | above]]. This file is in the same directory as the previous initial condition file, and is called <code>CheS_init_NOy.anc</code>. The UKCA UMUI jobs have been updated accordingly. |
||
| − | |||
| − | ===Lightning Emissions=== |
||
| − | |||
| − | Some jobs may be having issues with lightning NOx emissions. This will not affect all jobs, only those that take the section 0 prognostic cloud base height and cloud top height, rather than the section 5 diagnostic values (this can be checked in <code>ukca_setd1defs.F90</code> or in the UKCA STASH hand-edit). If you need the section 0 prognostics, please make sure following is set: |
||
| − | |||
| − | Model Selection --> |
||
| − | Atmosphere --> |
||
| − | Scientific Parameters and Sections --> |
||
| − | Section by section choices --> |
||
| − | Section 5: Convection --> |
||
| − | Continue (follow on window) --> |
||
| − | "Apply time decay of convective clouds" = 0 |
||
| − | |||
| − | If this is not the case these two fields will be empty and there will be no lightning NOx produced by UKCA. This bug does not affect the VN7.3_UKCA_CheM branch, which uses section 5 diagnostics. |
||
| − | |||
| − | ==UMUI Hand-Edits== |
||
| − | |||
| − | ===RADAER update=== |
||
| − | |||
| − | There is an issue with the following hand-edits on PUMA: |
||
| − | |||
| − | /home/ukca/hand_edits/VN7.3/std/radaer_indir1_on_l38.ed |
||
| − | /home/ukca/hand_edits/VN7.3/std/radaer_indir1_on_l38_HECToR.ed |
||
| − | /home/ukca/hand_edits/VN7.3/std/ukca_mode_cdnc_l38.ed |
||
| − | |||
| − | The line |
||
| − | |||
| − | &USE NAME="UPUKCA", LOCN=6 |
||
| − | |||
| − | should have been set to |
||
| − | |||
| − | &USE NAME="UPUKCA", LOCN=1 |
||
| − | |||
| − | which has now been done. However, if a personal copy of this hand-edit has been taken, it will also need to be corrected. If you have not taken a copy of this hand-edit the correction will be applied the next time your job is processed. |
||
| − | [[Category:Support]] |
||
Revision as of 17:12, 14 January 2014
Several of the UKCA Release Jobs have been ported from HECToR to ARCHER, and their jobids are detailed on that page.
Timings
The plot shows how UKCA scales when ported to ARCHER. The most efficient number of nodes to use is 3 (8x9), although the model will run fastest on 6 nodes (e.g. 16x9 or 12x12). Moving to 3 nodes from 6 will increase the run-time by 50%.
Code Changes
Cray cce Fortran Compiler
Corrections required for the Cray cce compiler on archer can be found on the bugfixes page.
