FAMOUS

Main.DebuggingHPCx History

Hide minor edits - Show changes to markup

February 21, 2007, at 03:02 PM by Annette -
Changed line 68 from:
to:
Restore
February 21, 2007, at 03:01 PM by Annette -
Changed line 68 from:
to:
Restore
February 21, 2007, at 03:00 PM by Annette -
Changed lines 66-68 from:
Add your file
to:
Add your file, containing the line of compile_vars to be overwritten
Restore
February 21, 2007, at 02:59 PM by Annette -
Changed lines 56-57 from:
$UMDIR/vn4.5/source/compile_vars
to:
Open $UMDIR/vn4.5/source/compile_vars
The compiler options follow FCOM_OPTS: 
@fort    FCOM_OPTS=-qfixed -q64 -qrealsize=8 -qintsize=8
Changed lines 66-68 from:
to:
Add your file
Restore
February 21, 2007, at 02:56 PM by Annette -
Changed lines 21-22 from:
/hpcx/devt/n02/n02-ncas/aosprey/um/mods/debug/timer3a_write.mf77
to:
/hpcx/devt/n02/n02-ncas/aosprey/um/mods/debug/timer3a_write.mf77
Restore
February 21, 2007, at 02:55 PM by Annette -
Changed lines 10-12 from:
Variable name: XLFRTEOPTS
Value: namelist=old:buffering=disable_all
to:
Variable name: XLFRTEOPTS
Value: namelist=old:buffering=disable_all
Changed lines 41-43 from:
Variable name: SKIP_TAR_COMPDIR
Value: true
to:
Variable name: SKIP_TAR_COMPDIR
Value: true
Changed lines 52-53 from:
to:

Changing the compiler options:

  • To look at the defult compiler options:
$UMDIR/vn4.5/source/compile_vars
  • To add a compile override file:
(Sub-Model Ind → Compilaton & Mods → User-defined Compile Option Overrides)
Restore
February 21, 2007, at 02:42 PM by Annette -
Changed lines 1-3 from:

Tips for Debugging the FAMOUS on HPCx

to:

Tips for Debugging FAMOUS on HPCx

Restore
February 21, 2007, at 02:42 PM by Annette -
Changed lines 4-5 from:

To get more information from the .leave file:

to:

To get more information in the .leave file:

Restore
February 21, 2007, at 02:42 PM by Annette -
Changed lines 1-3 from:

Tips for Debugging the UM on HPCx

to:

Tips for Debugging the FAMOUS on HPCx

Changed lines 31-53 from:

@]

to:

@]

To look at the source code:

  • The source files will usually be in $UCOMPDIR or $RECONDIR for the reconfiguration, defined in (Sub Model Ind → File & Directory Naming). Otherwise they have been zipped up and put into a tar archive. To retrieve them follow the instructions from the README file in the job output directory.
  • To set the job so that it does not automatically zip up the source files:
(Sub-Model Ind → Script Inserts and Modifications)
Add the environment variable:
Variable name: SKIP_TAR_COMPDIR
Value: true

Looking at individual processor output:

  • Each processor writes it’s output to an individual pe file in the output directory. Generally only the pe0 is copied to the .leave file, and the others are deleted. To keep all of the pe files:
(Sub-Model Ind → Output Management)
Select:
Output option: “Always keep output from PEs in DATAW”
Restore
February 21, 2007, at 02:24 PM by Annette -
Changed line 15 from:
  1. (Sub-Model Ind → Sub-Model Ind Section Options → Misc Secs 94–97)
to:
(Sub-Model Ind → Sub-Model Ind Section Options → Misc Secs 94–97)
Changed line 19 from:
  1. (Sub-Model Ind → Compilation and Mods → Modifications for the model)
to:
(Sub-Model Ind → Compilation and Mods → Modifications for the model)
Restore
February 21, 2007, at 02:23 PM by Annette -
Changed lines 8-9 from:
  • (Sub-Model Independent → Script Inserts and Modifications)
Add the envrionment variable:
to:
(Sub-Model Independent → Script Inserts and Modifications)
Add the envrionment variable:
Changed line 15 from:
  • (Sub-Model Ind → Sub-Model Ind Section Options → Misc Secs 94–97)
to:
  1. (Sub-Model Ind → Sub-Model Ind Section Options → Misc Secs 94–97)
Changed line 19 from:
  • (Sub-Model Ind → Compilation and Mods → Modifications for the model)
to:
  1. (Sub-Model Ind → Compilation and Mods → Modifications for the model)
Restore
February 21, 2007, at 02:22 PM by Annette -
Changed line 9 from:
Add the envrionment variable:
to:
Add the envrionment variable:
Changed line 23 from:
[@
to:

[@

Restore
February 21, 2007, at 02:22 PM by Annette -
Changed line 8 from:
(Sub-Model Independent → Script Inserts and Modifications)
to:
  • (Sub-Model Independent → Script Inserts and Modifications)
Changed line 15 from:
(Sub-Model Ind → Sub-Model Ind Section Options → Misc Secs 94–97)
to:
  • (Sub-Model Ind → Sub-Model Ind Section Options → Misc Secs 94–97)
Changed line 19 from:
(Sub-Model Ind → Compilation and Mods → Modifications for the model)
to:
  • (Sub-Model Ind → Compilation and Mods → Modifications for the model)
Added line 22:
Restore
February 21, 2007, at 02:21 PM by Annette -
Changed line 15 from:
Sub-model Independent Section Options → Section 97
to:
(Sub-Model Ind → Sub-Model Ind Section Options → Misc Secs 94–97)
Changed lines 18-20 from:
Compilation and Modifications → Modifications for the model
Add:
/hpcx/devt/n02/n02-ncas/aosprey/um/mods/debug/timer3a_write.mf77
to:
(Sub-Model Ind → Compilation and Mods → Modifications for the model)
Add the mod:
/hpcx/devt/n02/n02-ncas/aosprey/um/mods/debug/timer3a_write.mf77 
*IDENT TIMER3AWRITE
*/
*/ Print timed routines
*/
*DECLARE TIMER3A
*D TIMER3A.177
         WRITE(6,*) section_name,' action= ',action
Restore
February 21, 2007, at 02:16 PM by Annette -
Changed lines 6-7 from:
  • Disable buffering so the UM prints out everything before it crashes.
to:
  • Disable buffering so the UM prints out everything as it happens.
Restore
February 21, 2007, at 02:15 PM by Annette -
Changed lines 1-3 from:

Basic Tips for Debugging the UM on HPCx

to:

Tips for Debugging the UM on HPCx

Restore
February 21, 2007, at 02:15 PM by Annette -
Changed lines 1-3 from:

Some Tips for Debugging the UM on HPCx

to:

Basic Tips for Debugging the UM on HPCx

Restore
February 21, 2007, at 02:14 PM by Annette -
Changed lines 8-9 from:
Sub-Model Independent → Script Inserts and Modifications
Add:
to:
(Sub-Model Independent → Script Inserts and Modifications)
Add the envrionment variable:
Changed lines 13-14 from:
  • Get the UM to write out when it enters and leaves a timed routine.
to:
  • Get the UM to print when it enters and leaves a timed routine.
Restore
February 21, 2007, at 02:12 PM by Annette -
Changed lines 1-5 from:

Debugging the UM on HPCx

Things that I find helpful to get more information from the .leave file:

to:

Some Tips for Debugging the UM on HPCx

To get more information from the .leave file:

Restore
February 21, 2007, at 11:54 AM by Annette -
Changed lines 13-14 from:
  • To get an idea of what the UM was doing when it crashed, get it to write a message when it enters a timed routine.
to:
  • Get the UM to write out when it enters and leaves a timed routine.
Restore
February 20, 2007, at 04:33 PM by Annette -
Added line 16:
Select:
Restore
February 20, 2007, at 04:32 PM by Annette -
Changed line 15 from:
In Sub-Model Independent → Sub-model Independent Section Options → Section 97
to:
Sub-model Independent Section Options → Section 97
Changed line 17 from:
In Sub-Model Independent → Compilation and Modifications → Modifications for the model
to:
Compilation and Modifications → Modifications for the model
Restore
February 20, 2007, at 04:32 PM by Annette -
Changed lines 14-18 from:
In Sub-Model Independent → Sub-model Independent Section Options → Section 97
Section 97 (timer): 3A
In Sub-Model Independent → Compilation and Modifications → Modifications for the model
Add:
/hpcx/devt/n02/n02-ncas/aosprey/um/mods/debug/timer3a_write.mf77
to:
In Sub-Model Independent → Sub-model Independent Section Options → Section 97
Section 97 (timer): 3A
In Sub-Model Independent → Compilation and Modifications → Modifications for the model
Add:
/hpcx/devt/n02/n02-ncas/aosprey/um/mods/debug/timer3a_write.mf77
Restore
February 20, 2007, at 04:31 PM by Annette -
Changed lines 7-11 from:
In Sub-Model Independent → Script Inserts and Modifications
Add:
Variable name: XLFRTEOPTS
Value: namelist=old:buffering=disable_all
to:
Sub-Model Independent → Script Inserts and Modifications
Add:
Variable name: XLFRTEOPTS
Value: namelist=old:buffering=disable_all
Restore
February 20, 2007, at 04:31 PM by Annette -
Changed lines 7-11 from:
In Sub-Model Independent → Script Inserts and Modifications
Add:
Variable name: XLFRTEOPTS
Value: namelist=old:buffering=disable_all
to:
In Sub-Model Independent → Script Inserts and Modifications
Add:
Variable name: XLFRTEOPTS
Value: namelist=old:buffering=disable_all
Changed lines 13-17 from:
In Sub-Model Independent → Sub-model Independent Section Options → Section 97
Section 97 (timer): 3A
In Sub-Model Independent → Compilation and Modifications → Modifications for the model
Add:
/hpcx/devt/n02/n02-ncas/aosprey/um/mods/debug/timer3a_write.mf77
to:
In Sub-Model Independent → Sub-model Independent Section Options → Section 97
Section 97 (timer): 3A
In Sub-Model Independent → Compilation and Modifications → Modifications for the model
Add:
/hpcx/devt/n02/n02-ncas/aosprey/um/mods/debug/timer3a_write.mf77
Restore
February 20, 2007, at 04:30 PM by Annette -
Changed lines 1-3 from:

Debugging the UM on [HPCx]

to:

Debugging the UM on HPCx

Restore
February 20, 2007, at 04:30 PM by Annette -
Restore
February 20, 2007, at 04:30 PM by Annette -
Changed lines 1-4 from:

Debugging the UM on HPCx

Things that I find helpful

to:

Debugging the UM on [HPCx]

Things that I find helpful to get more information from the .leave file:

  • Disable buffering so the UM prints out everything before it crashes.
In Sub-Model Independent → Script Inserts and Modifications
Add:
Variable name: XLFRTEOPTS
Value: namelist=old:buffering=disable_all
  • To get an idea of what the UM was doing when it crashed, get it to write a message when it enters a timed routine.
In Sub-Model Independent → Sub-model Independent Section Options → Section 97
Section 97 (timer): 3A
In Sub-Model Independent → Compilation and Modifications → Modifications for the model
Add:
/hpcx/devt/n02/n02-ncas/aosprey/um/mods/debug/timer3a_write.mf77
Restore
February 20, 2007, at 04:19 PM by Annette -
Added lines 1-4:

Debugging the UM on HPCx

Things that I find helpful

Restore
Page last modified on February 21, 2007, at 03:02 PM by Annette