Create experiment for cpl_gray+swamp+ocn #70
Conversation
Before, input was the old version with no continent. Now input folders should include the necessary files
Each build directory had some remnants of building the executables previously. Removing that now, only leaving genmake_local
|
@qwert2266 Thanks for sharing your set-up, this is great. I will take a look in the coming days and will see if there is something to add or change. |
also add ".gitignore" and update shares_code (removing CVS headers)
Also save atm & ocn multi-procs SIZE.h (suffix: .24procs, .3procs) and add single-procs version (easier for testing).
|
@qwert2266 I did some cleaning so that I can now compile for single-proc component (easier for testing). I still have issues to run, because the |
|
And regarding number of procs to use to run atm & ocn as well as at which level to create the |
|
Hey Jean-Michel,
I updated the SIZE.h and Initiation files, sorry about that. They were from
a run that didn't have ENSO, I will commit the new ones now. I'm not sure
where the tRef settings are, could you point me to that and I'll make the
same change?
Just to clarify, you want me to change the readme so that it's clear the
default version does not have multiple processes, but leave in that there
is an option to include this?
Thanks,
PJ
…On Thu, Mar 7, 2024 at 12:11 PM Jean-Michel Campin ***@***.***> wrote:
And regarding number of procs to use to run atm & ocn as well as at which
level to create the rank_? dir:
For now, I think it's easier to use the same convention as for the other
coupled set-up, so that one can use tools/run_cpl_test to check that
everything works.
In the future we could try to add new arguments to run_cpl_test so enable
to change number of procs (this is already there but as hard coded
parameters) and to have the rank dir one level up (but this requires also
to adjust prepare_run in input_* dirs.
And for now, we would just keep a multi-proc SIZE.h version (I don't like
too much the suffix ".3procs" and ".24procs" that I put there and could use
soemthing else) and document this.
—
Reply to this email directly, view it on GitHub
<#70 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AQXV3RS5XYQIAQDAC6VXPPLYXCNVFAVCNFSM6AAAAABCLHLC5KVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTSOBUGAZDSMBXGA>
.
You are receiving this because you were mentioned.Message ID:
***@***.***>
|
Updated to make consistent the necessary things to model ENSO, which are the GGL90 package and adding vertical levels to the ocean
…266/verification_other into Adding_cpl_gray+swamp+ocn
|
Nevermind, I updated tRef and sRef
…On Thu, Mar 7, 2024 at 12:11 PM Jean-Michel Campin ***@***.***> wrote:
And regarding number of procs to use to run atm & ocn as well as at which
level to create the rank_? dir:
For now, I think it's easier to use the same convention as for the other
coupled set-up, so that one can use tools/run_cpl_test to check that
everything works.
In the future we could try to add new arguments to run_cpl_test so enable
to change number of procs (this is already there but as hard coded
parameters) and to have the rank dir one level up (but this requires also
to adjust prepare_run in input_* dirs.
And for now, we would just keep a multi-proc SIZE.h version (I don't like
too much the suffix ".3procs" and ".24procs" that I put there and could use
soemthing else) and document this.
—
Reply to this email directly, view it on GitHub
<#70 (comment)>,
or unsubscribe
<https://github.com/notifications/unsubscribe-auth/AQXV3RS5XYQIAQDAC6VXPPLYXCNVFAVCNFSM6AAAAABCLHLC5KVHI2DSMVQWIX3LMV43OSLTON2WKQ3PNVWWK3TUHMYTSOBUGAZDSMBXGA>
.
You are receiving this because you were mentioned.Message ID:
***@***.***>
|
|
@qwert2266 Thanks for the update the ocean set-up. I am going to check few other things (and may be clean-up little things) and will let you know how it goes. After that we could review the README once everything is in place. |
- remove CVS Header and Name - import changed from standard version in main code (to reduce diff with) - reduce "numDiags" in DIAGNOSTICS_SIZE.h (but still much larger than what is currently used).
Assuming all rank's dir are directly put in cpl_gray+swamp+ocn/, adjust script "prepare_run" and remove (duplicated) bin files that are available elsewhere ; Also remove unused file "input_ocn/flat_4km.bin" and move "SST_APE_1.bin" from input.ape/ to input/ in exp. atm_gray (a more logical place where it can be generated with only minor changes to gendata.m there). -
to be used for multi-processor ocn & atm
also adjust some technical details
2 participants
Working on creating experiment, readme is here now and will add other code later