View source for GNU Fortran compiler

On this page, we discuss compiling GEOS-Chem with the GNU Fortran compiler (aka <code>gfortran</code>).

== Overview ==

We have added several modifications to facilitate compiling GEOS-Chem with the GNU Fortran (aka <code>gfortran</code> compiler).    In the process, we have also corrected some coding issues that were flagged by GNU Fortran.  These changes will be added to the v11-01 public release code.

=== GNU Fortran versions tested with GEOS-Chem ===

The [[GEOS-Chem Support Team]] has experimented using different versions of GNU Fortran to build GEOS-Chem.  Here are our results.

{| border=1 cellspacing=0 cellpadding=5 
|- bgcolor="#CCCCCC"
!width="100px"|Version
!width="175px"|Machine
!width="700px"|Status

|-valign="top"
|gfortran 4.4.7
|odyssey.rc.fas.harvard.edu
|
*GEOS-Chem builds into an executable.
*GEOS-Chem simulations run to completion.
*Can be used with the TotalView debugger.

|-valign="top"
|gfortran 4.8.2
|odyssey.rc.fas.harvard.edu
|
*GEOS-Chem builds into an executable.
*GEOS-Chem simulations run to completion.
*Can NOT be used with the TotalView debugger.
**[[#GNU Fortran 4.8 and higher is not compatible with Totalview|gfortran 4.8.2 loads the DWARF-4 debugging library by default]].  
**TotalView can only be used with DWARF-2 or DWARF-3.

|-valign="top"
|gfortran 5.2.0
|odyssey.rc.fas.harvard.edu
|
*GEOS-Chem does not compile
**gfortran 6.1.0 [[#THREADPRIVATE_error_in_GNU_Fortran_5_and_higher|refuses to allow entire common blocks to be placed in an <code>!$OMP THREADPRIVATE statement</code>]]. This occurs in the 3rd-party ISORROPIA header file <code>ISORROPIA/isrpia.inc</code>.

|-valign="top"
|gfortran 6.1.0
|odyssey.rc.fas.harvard.edu
|
*GEOS-Chem does not compile
**gfortran 6.1.0 [[#THREADPRIVATE_error_in_GNU_Fortran_5_and_higher|refuses to allow entire common blocks to be placed in an <code>!$OMP THREADPRIVATE statement</code>]]. This occurs in the 3rd-party ISORROPIA header file <code>ISORROPIA/isrpia.inc</code>.
 
|}

We invite you to test with other versions of GNU Fortran.  Please send your results to the [[GEOS-Chem Support Team]].

--[[User:Bmy|Bob Yantosca]] ([[User talk:Bmy|talk]]) 19:40, 4 October 2016 (UTC)

== Environment settings for GNU Fortran ==

Here is some information about how you can customize your Unix environment to use the GNU Fortran compiler.

=== Using the module command to load GNU Fortran and related libraries ===

On most Unix clusters, the <code>module</code> command is used to load the GNU Fortran compiler library into your Unix enviroment, along with any related libraries (e.g. netCDF, OpenMPI).  For example, we use the following commands on the Harvard cluster (odyssey.rc.fas.harvard.edu):

 # These commands load GNU Fortran 4.8.2
 module load gcc/4.8.2-fasrc01      openmpi/1.10.1-fasrc01
 module load netcdf/4.3.3.1-fasrc02 netcdf-fortran/4.4.2-fasrc01

You can ask your IT staff what the corresponding commands would be on your particular cluster.

The <code>module</code> command should automatically define several environment variables for you:

{| border=1 cellspacing=0 cellpadding=5 
|- bgcolor="#CCCCCC"
!width="150px"|Variable
!width="150px"|Expected setting
!width="600px"|Description

|-valign="top"
|<code>FC</code>
|<code>gfortran</code>
|Name of the GNU Fortran compiler

|-valign="top"
|<code>CC</code>
|<code>gcc</code>
|Name of the GNU C compiler

|-valign="top"
|<code>CXX</code>
|<code>g++</code>
|Name of the GNU C++ compiler

|-valign="top"
|<code>NETCDF_HOME</code>
|System-dependent
|Path to the root netCDF folder

|-valign="top"
|<code>NETCDF_INCLUDE</code>
|System-dependent
|Path to the netCDF include folder (e.g. <code>$NETCDF_HOME/include</code>)

|-valign="top"
|<code>NETCDF_LIB</code>
|System-dependent
|Path to the netCDF library folder (e.g. <code>$NETCDF_HOME/lib</code> or <code>$NETCDF_HOME/lib64</code>)

|-valign="top"
|<code>NETCDF_FORTRAN_HOME</code>
|System-dependent
|Path to the root netCDF Fortran folder

|-valign="top"
|<code>NETCDF_FORTRAN_INCLUDE</code>
|System-dependent
|Path to the netCDF Fortran include folder (e.g. <code>$NETCDF_FORTRAN_HOME/include</code>)

|-valign="top"
|<code>NETCDF_FORTRAN_LIB</code>
|System-dependent
|Path to the netCDF Fortran library folder (e.g. <code>$NETCDF_FORTRAN_HOME/lib</code> or <code>$NETCDF_FORTRAN_HOME/lib64</code>)

|}

If these variables are not automatically set by the module command on your system (or if your system does not use the module command:

*Set the <code>FC</code>, <code>CC</code>, and <code>CXX</code> variables automatically in your startup script (e.g. <code>.bashrc</code> or <code>.cshrc</code>).
*Ask your IT staff where the netCDF library paths are located, and set the <code>NETCDF_HOME</code>, <code>NETCDF_INCLUDE</code>, and <code>NETCDF_LIB</code> environment variables accordingly.

Depending on your system, you may find that the netCDF Fortran libraries are installed into a different folder than the rest of the netCDF library files.  If this is the case, then the <code>module</code> command should automatically define the <code>NETCDF_FORTRAN_HOME</code>, <code>NETCDF_FORTRAN_INCLUDE</code>, and <code>NETCDF_FORTRAN_LIB</code> environment variables.  If not, then ask your IT staff what the proper paths are so that you can set these variables manually.

<span style="color:red">'''NOTE: Setting the <code>OMP_STACKSIZE</code> environment variable will make it easier to switch between different compilers on your system.  The <code>KMP_STACKSIZE</code> environment variable only works with the Intel Fortran Compiler but not with GNU Fortran.'''</span>

--[[User:Bmy|Bob Yantosca]] ([[User talk:Bmy|talk]]) 21:45, 4 October 2016 (UTC)

=== Requesting sufficient stack memory for GEOS-Chem ===

In order to run GEOS-Chem with GNU Fortran, you must request the maximum amount of stack memory in your Unix environment.  (The stack memory is where local automatic variables and temporary <code>!$OMP PRIVATE</code> variables will be created.)  Add the following lines to your system startup file:

{| border=1 cellspacing=0 cellpadding=5 
|- bgcolor="#CCCCCC"
!width="300px"|If you use bash.<br>add this to your .bashrc file
!width="300px"|If you use csh or tcsh,<br>add this to your .cshrc file

|-valign="top"
|
 ulimit -s unlimited
 export OMP_STACKSIZE=500m
|
 limit stacksize unlimited
 setenv OMP_STACKSIZE 500m
|}

The <code>ulimit -s unlimited</code> (for bash) or <code>limit stacksize unlimited</code> commands tell the Unix shell to use the maximum amount of stack memory available.

The environment variable <code>OMP_STACKSIZE</code> must also be set to a very large number.  In this example, we are nominally requesting 500 MB of memory.  But in practice, this will tell the GNU Fortran compiler to use the maximum amount of stack memory available on your system.  The value <code>500m</code> is a good round number that is larger than the amount of stack memory on most computer clusters, but you can change this if you wish.

--[[User:Bmy|Bob Yantosca]] ([[User talk:Bmy|talk]]) 19:26, 4 October 2016 (UTC)

== Compilation options ==

The sections below contain useful information about the GNU Fortran compiler options that are used for GEOS-Chem.

=== Optimization options ===

Please see the [https://gcc.gnu.org/onlinedocs/gcc/Optimize-Options.html#Optimize-Options ''Optimize Options'' section of the GNU Compiler Collection manual] for detailed information about GNU Fortran's optimization settings.

The default GNU Fortran optimization settings for GEOS-Chem are: <code>-O3 -funroll-loops</code>.

--[[User:Bmy|Bob Yantosca]] ([[User talk:Bmy|talk]]) 22:01, 21 September 2016 (UTC)

=== Debugging options ===

Please see the [https://gcc.gnu.org/onlinedocs/gcc/Debugging-Options.html#Debugging-Options ''Debugging Options'' section of the GNU Compiler Collection manual] for detailed information about GNU Fortran's debugging settings.

--[[User:Bmy|Bob Yantosca]] ([[User talk:Bmy|talk]]) 20:16, 22 September 2016 (UTC)

=== List of commonly-used compilation options ===

Here are the GNU Fortran compilation options currently used by GEOS-Chem:

{| border=1 cellspacing=0 cellpadding=5 
|- bgcolor="#CCCCCC"
!width="275px"|Option
!width="600px"|Description
!width="125px"|How invoked in GEOS-Chem?

|- bgcolor="#CCFFFF"
!
!Normal compiler settings
! 

|-valign="top"
|<code>-cpp</code>
|Turns on the C-preprocessor, to evaluate <code>#if</code> and <code>#define</code> statements in the source code.
|Default setting

|-valign="top"
|<code>-fautomatic</code>
|This option places local variables (scalars and arrays of all types), except those declared as SAVE, on the run-time stack.  It is as if the variables were declared with the AUTOMATIC attribute.  It does not affect variables that have the SAVE attribute or ALLOCATABLE attribute, or variables that appear in an EQUIVALENCE statement or in a common block.
|Default setting

|-valign="top"
|<code>-fconvert-big-endian</code>
|Specifies that the format will be big endian for integer data and big endian IEEE floating-point for real and complex data.  This only affects file I/O to/from binary files (such as binary punch files) but not ASCII, netCDF, or other file formats.
|Default setting

|-valign="top"
|<code>-fno-align-commons</code>
|Prevents the compiler from padding bytes anywhere in common blocks and structures.  Padding can affect numerical precision.
|Default setting

|-valign="top"
|<code>-fopenmp</code>
|Enables OpenMP parallelization commands.
|Default setting

|-valign="top"
|<code>-funroll-loops</code>
|Typically improves performance on code using iterative DO loops by unrolling them and is probably generally appropriate for Fortran, though it is not turned on at any optimization level. Note that outer loop unrolling isn't done specifically; decisions about whether to unroll a loop are made on the basis of its instruction count.

Also, no `loop discovery' is done, so only loops written with DO benefit from loop optimizations, including—but not limited to—unrolling. Loops written with IF and GOTO are not currently recognized as such. This option unrolls only iterative DO loops, not DO WHILE loops. 
|Default setting

|-valign="top"
|<code>-march=native</code>
|This selects the CPU to generate code for at compilation time by determining the processor type of the compiling machine. Using <code>-march=native</code> enables all instruction subsets supported by the local machine (hence the result might not run on different machines).  We use this option for compiling GEOS-Chem because it is the most portable.
|Default setting

|-valign="top"
|<code>-O3</code>
|Performs nearly all supported optimizations that do not involve a space-speed tradeoff, plus a few more optimizations for function inlining and vectorization.  For more information, please see the [https://gcc.gnu.org/onlinedocs/gcc/Optimize-Options.html#Optimize-Options ''Optimize Options'' section of the GNU Compiler Collection manual].
|Default setting

|-valign="top"
|<code>-std=legacy</code>
|Tells GNU Fortran not to halt compilation when encountering code that does not adhere to the Fortran 95, 2003, or 2008 standards.  Gfortran is a much stricter compiler, so turning this option on will tell Gfortran to be more lenient.  
|Default setting

|-valign="top"
|<code>-w</code>
|Turns off most informational warnings.
|Default setting

|- bgcolor="#CCFFFF"
!
!Special compiler settings
! 

|-valign="top"
|<code>-fdefault-real-8</code>
|This option tells the compiler to treat variables that are declared as <tt>REAL</tt> as <tt>REAL*8</tt>

''NOTE: This option is not used globally, but is only applied to certain indidvidual files (mostly from third-party codes like ISORROPIA.  Current GEOS-Chem programming practice is to use either <tt>REAL*4</tt> or <tt>REAL*8</tt> instead of <tt>REAL</tt>, which avoids confusion.''
|Used as needed

|-valign="top"
|<code>-freal-4-real-8</code>
|This option tells the compiler to treat variables that are declared as <tt>REAL</tt> as <tt>REAL*8</tt>

''NOTE: This option is not used globally, but is only applied to certain indidvidual files (mostly from third-party codes like ISORROPIA.  Current GEOS-Chem programming practice is to use either <tt>REAL*4</tt> or <tt>REAL*8</tt> instead of <tt>REAL</tt>, which avoids confusion.''
|Used as needed

|-valign="top"
|<code>-mcmodel=medium</code>
|This option is used to tell Gfortran to use more than 2GB of static memory.  This avoids a [[#Relocation truncated to fit error|specific type of memory error]] that can occur if you compile GEOS-Chem for use with an extremely high-resolution grid (e.g. 0.25&deg; x 0.3125&deg; nested grid).
|Default setting

|- bgcolor="#CCFFFF"
!
!Settings only used for debugging
! 

|-valign="top"
|<code>-fbacktrace</code>
|This option tells the compiler to generate extra information in the object file to provide source file traceback information when a severe error occurs at run time.  When the severe error occurs, source file, routine name, and line number correlation information is displayed along with call stack hexadecimal addresses (program counter trace).  This option increases the size of the executable program, but has no impact on run-time execution speeds.  It functions independently of the debug option.
|<code>TRACEBACK=yes</code>

|-valign="top"
|<code>-fcheck-array-temporaries</code>
|Checks to see if any array temporaries are created.  Depending on how you write your subroutine and function calls, the compiler may need to create a temporary array to hold the values in the array before it passes them to the subroutine.  For detailed information, please see our [[Passing array arguments efficiently in GEOS-Chem]] wiki page.
|<code>DEBUG=yes</code>

|-valign="top"
|<tt>-fcheck-bounds</tt>
|Check for [[Common GEOS-Chem error messages#Array-out-of-bounds_error|array-out-of-bounds errors]].  This is invoked when you compile GEOS-Chem with the <code>BOUNDS=yes</code> Makefile option.  ''NOTE: Only use this option <code>-fcheck-bounds</code> for debugging, as this option will cause GEOS-Chem to execute more slowly!''
|<code>BOUNDS=yes</code>

|-valign="top"
|<code>-ffpe-trap=invalid,zero,overflow</code>
|This option will cause GEOS-Chem to halt when floating-point errors are encountered.  This can happen if an equation results in a denormal value, e.g. <code>NaN</code>, or <code>+/-Infinity</code>.  Common causes of floating-point errors are divisions where the denominator becomes zero.
|<code>FPE=yes</code>

|-valign="top"
|<code>-finit-real-snan</code>
|This option will set local automatic variables to a signaling NaN.  This will make it easier for the compiler to detect undefined variables.
|<code>FPE=yes</code>

|-valign="top"
|<code>-g</code>
|Tells the compiler to generate full debugging information in the object file.  This will cause a debugger (like Totalview) to display the actual lines of source code, instead of hexadecimal addresses (which is gibberish to anyone except hardware engineers).
|<code>DEBUG=yes</code>

|-valign="top"
|<code>-gdwarf-2</code>
|Tells the compiler to generate full debugging information using the DWARF-2 library standard.
|<code>DEBUG=yes</code>

|-valign="top"
|<code>-gdwarf-2</code>
|Tells the compiler to strictly adhere to the DWARF-2 debugging library standard.
|<code>DEBUG=yes</code>

|-valign="top"
|<code>-O0</code>
|Turns off all optimization.  Source code instructions (e.g. DO loops, IF blocks) and numerical expressions are evaluated in precisely the order in which they are listed, without being internally rewritten by the optimizer.  This is necessary for using a debugger (like Totalview).
|<code>DEBUG=yes</code>

|-valign="top"
|<code>-Wall</code>
|Enables some common compiler warnings that you probably would not enable out unless you were debugging.
|<tt>DEBUG=yes</tt>

|-valign="top"
|<code>-Warray-temporaries</code>
|Warn about array temporaries generated by the compiler. The information generated by this warning is sometimes useful in optimization, in order to avoid
such temporaries. Used in conjunction with <code>-fcheck-array-temporaries</code>.
|<code>DEBUG=yes</code>

|-valign="top"
|<code>-Wconversion</code>
|Warn about implicit conversions that are likely to change the value of the expression after conversion. Implied by <code>-Wall</code>.
|<code>DEBUG=yes</code>

|-valign="top"
|<code>-Wextra</code>
|<code>-Wextra</code> Enables some warning options for usages of language features which may be problematic. This currently includes <code>-Wcompare-reals</code> and <code>-Wunused-parameter</code>.
|<code>DEBUG=yes</code>

|}

--[[User:Bmy|Bob Yantosca]] ([[User talk:Bmy|talk]]) 21:55, 21 September 2016 (UTC)

=== Typical settings for a GEOS-Chem simulation === 

The normal GEOS-Chem build uses the following Gfortran compiler flags:

 -cpp -w -std=legacy -fautomatic -fno-align-commons -fconvert=big-endian -fno-range-check
 -march=native -fopenmp -mcmodel=medium -O3 -funroll-loops

whereas a debugging run (meant to execute in a debugger such as TotalView) will typically use these flags:

 -cpp -w -std=legacy -fautomatic -fno-align-commons -fconvert=big-endian -fno-range-check -march=native
 -fopenmp -mcmodel=medium -g gdwarf-2 -gstrict-dwarf -O0 -Wall -Wextra -Warray-temporaries -Wconversion 
 -fcheck-array-temporaries -fbounds-check -fbacktrace -ffpe-trap=invalid,zero,overflow, -finit-real=snan

--[[User:Bmy|Bob Yantosca]] ([[User talk:Bmy|talk]]) 20:15, 22 September 2016 (UTC)

== Modifications made to GEOS-Chem for GNU Fortran ==

The following tables list the modifications that had to be made in order to compile GEOS-Chem with GNU Fortran.

=== Module in GeosCore ===

{| border=1 cellspacing=0 cellpadding=5 
|- bgcolor="#CCCCCC"
!width="200px"|Module
!width="450px"|Code removed (in <span style="color:red">RED</span>)
!width="600px"|Code added (in <span style="color:green">GREEN</span>)

|-valign="top"
|
 carbon_mod.F
|
 GLOB_DARO2(I,J,L,<span style="color:red">:</span>,JHC-5) = DELHC(<span style="color:red">:</span>)
 . . .
 GLOB_DARO2(I,J,L,<span style="color:red">:</span>,4) = DELHC(<span style="color:red">:</span>)
|
 GLOB_DARO2(I,J,L,<span style="color:green">1:2</span>,JHC-5) = DELHC(<span style="color:green">1:2</span>)
 . . .
 GLOB_DARO2(I,J,L,<span style="color:green">1:2</span>,4) = DELHC(<span style="color:green">1:2</span>)
 
 <span style="color:green"># Now only copy the 1st 2 elements of DELHC 
 # into GLOB_DARO2, which avoids an array-size mismatch</span>

|-valign="top"
|
 convection_mod.F
|
 IF ( AER <span style="color:red">== .TRUE.</span> ) THEN
|
 IF ( AER ) THEN

|-valign="top"
|
 diag3.F
|
 (Input_Opt%LSKYRAD(2)<span style="color:red">.EQ.</span>TRUE.))
 . . .
 (Input_Opt%LSKYRAD(1)<span style="color:red">.EQ.</span>TRUE.))
|
 (Input_Opt%LSKYRAD(2)<span style="color:green">.EQV.</span>TRUE.))
 . . .
 (Input_Opt%LSKYRAD(1)<span style="color:green">.EQV.</span>TRUE.))

|-valign="top"
|
 diag_mod.F
|
 <span style="color:red">INTEGER, ALLOCATABLE :: AD71_COUNT
 INTEGER, ALLOCATABLE :: AD71_HRCT
 INTEGER, ALLOCATABLE :: AD71_LHR
 INTEGER, ALLOCATABLE :: AD71_LDAY</span>
|
 <span style="color:green">INTEGER              :: AD71_COUNT
 INTEGER              :: AD71_HRCT
 INTEGER              :: AD71_LHR
 INTEGER              :: AD71_LDAY
 
 # Scalar variables cannot be declared ALLOCATABLE.</span>

|-valign="top"
|
 diag48_mod.F
|
 (Input_Opt%LSKYRAD(2)<span style="color:red">.EQ.</span>TRUE.))
 . . .
 (Input_Opt%LSKYRAD(1)<span style="color:red">.EQ.</span>TRUE.))
|
 (Input_Opt%LSKYRAD(2)<span style="color:green">.EQV.</span>TRUE.))
 . . .
 (Input_Opt%LSKYRAD(1)<span style="color:green">.EQV.</span>TRUE.))

|-valign="top"
|
 emissions_mod.F90
|
 CALL EMISSCARBON( ... etc ... )
 IF ( RC /= GC_SUCCESS ) RETURN 
|
 <span style="color:green">IF ( Input_Opt%ITS_A_FULLCHEM_SIM   .or. &
      Input_Opt%ITS_AN_AEROSOL_SIM ) THEN</span>
    CALL EMISSCARBON( ... etc ... )
    IF ( RC /= GC_SUCCESS ) RETURN 
 <span style="color:green">ENDIF 
 
 # Need to prevent EMISSCARBON from being called
 # for simulations without carbon aerosols</span>

|-valign="top"
|
 flexchem_mod.F90
|
  write(*,'(i,a3,a85)') D,' | ',EQN_NAMES(D)
|
  WRITE( 6, '(i<span style="color:green">8,a3,a85)' ) D,' | ',EQN_NAMES(D)

|-valign="top"
|
 geosfp_read_mod.F90
|
 <span style="color:red">INTEGER</span>, SAVE      :: first = .TRUE.
|
 <span style="color:green">LOGICAL</span>, SAVE      :: first = .TRUE.

|-valign="top"
|
 hcoi_gc_diagn_mod.F90
|
 IF ( YesOrNo <span style="color:red">==</span> .FALSE. ) THEN
|
 IF ( YesOrNo <span style="color:green">.eqv.</span> FALSE ) THEN 

|-valign="top"
|
 hcoi_gc_main_mod.F90
|
 IF ( Input_Opt%LUCX <span style="color:red">/=</span> LTMP ) THEN	   
 . . .
 IF ( Input_Opt%LSCHEM <span style="color:red">/=</span> LTMP ) THEN
 . . .
 IF ( Input_Opt%LCHEM <span style="color:red">/=</span>. LTMP .and. 
|

 IF ( Input_Opt%LUCX <span style="color:green">.neqv.</span> LTMP ) THEN
 . . .
 IF ( Input_Opt%LSCHEM <span style="color:green">.neqv.</span> LTMP ) THEN
 . . .
 IF ( Input_Opt%LCHEM <span style="color:green">.neqv./</span> LTMP .and.              

|-valign="top"
|
 input_mod.F
|
 READ( SUBSTRS(1:N), '(i)') CFCYEAR
 . . .
 write(MSG,'(I,a,L)') '<>', Input_Opt%TS_DYN, ...
 . . .
 Input_Opt%ND63_TRACERS(1:ND63) = TRACERS(1:N_ND63)
 . . .
 IF ( LCH4BUD <span style="color:red">.EQ. .TRUE.</span> ) THEN
|
 READ( SUBSTRS(1:N), '(i<span style="color:green">4</span>)') CFCYEAR
 . . .
 write(MSG,'(I<span style="color:green">8</span>,a,L)') '<>', Input_Opt%TS_DYN, ...
 . . .
 Input_Opt%ND63_TRACERS(1:<span style="color:green">N_</span>ND63) = TRACERS(1:N_ND63)
 . . .
 IF ( LCH4BUD ) THEN

|-valign="top"
|
 isoropiaII_mod.F
|
 <span style="color:red">#IF</span> !defined ( USE_REAL_8 )
       HNO3_sav(I,J,L) = SNGL(HNO3_UGM3) 
 <span style="color:red">#ELIF</span>
       HNO3_sav(I,J,L) = HNO3_UGM3
 <span style="color:red">#ENDIF</span>
|
 <span style="color:green">#if</span> !defined( USE_REAL_8 )
       HNO3_sav(I,J,L) = SNGL(HNO3_UGM3)
 <span style="color:green">#else</span>
       HNO3_sav(I,J,L) = HNO3_UGM3
 <span style="color:green">#endif</span>   

|-valign="top"
|
 ocean_mercury_mod.F
|
 IF ( <span style="color:red">ND03</span> .and. NN == ID_Hg_tot )
|
 IF ( <span style="color:green">( ND03 > 0 )</span> .and. <span style="color:green">(</span> NN == ID_Hg_tot <span style="color:green">)</span> )
 
 <span style="color:green"># Use parentheses to cast to LOGICAL type

|-valign="top"
|
 merra2_read_mod.F90
|
 <span style="color:red">INTEGER</span>, SAVE      :: first = .TRUE.
|
 <span style="color:green">LOGICAL</span>, SAVE      :: first = .TRUE.

|-valign="top"
|
 ndxx_setup.F
|
 <span style="color:red">INTEGER</span> :: IT_IS_A_CH3I_SIM
 <span style="color:red">INTEGER</span> :: IT_IS_A_FULLCHEM_SIM
 <span style="color:red">INTEGER</span> :: IT_IS_A_MERCURY_SIM
 <span style="color:red">INTEGER</span> :: IT_IS_A_TAGO3_SIM
 <span style="color:red">INTEGER</span> :: IT_IS_A_H2HD_SIM
|
 <span style="color:green">LOGICAL</span> :: IT_IS_A_CH3I_SIM
 <span style="color:green">LOGICAL</span> :: IT_IS_A_FULLCHEM_SIM
 <span style="color:green">LOGICAL</span> :: IT_IS_A_MERCURY_SIM
 <span style="color:green">LOGICAL</span> :: IT_IS_A_TAGO3_SIM
 <span style="color:green">LOGICAL</span> :: IT_IS_A_H2HD_SIM

|-valign="top"
|
 ndxx_setup.F
|
 <span style="color:red">IF ( ALLOCATED( AD71_COUNT  ) ) DEALLOCATE( AD71_COUNT  )
 IF ( ALLOCATED( AD71_HRCT   ) ) DEALLOCATE( AD71_HRCT   )
 IF ( ALLOCATED( AD71_LDAY   ) ) DEALLOCATE( AD71_LDAY   )
 IF ( ALLOCATED( AD71_LHR    ) ) DEALLOCATE( AD71_LHR    )</span>
|
 
|-valign="top"
|
 seasalt_mod.F
|
 <span style="color:red"> IF (DMID(ID) .ge. R0*2e+0_fp .and. ... 
 &          THEN</span>
|
 <span style="color:green"> IF ( DMID(ID) .ge. R0*2e+0_fp   .and.
 &     DMID(ID) .le. R1*2e+0_fp ) THEN
 
 # Break IF statement into 2 lines</span>

|-valign="top"
|
 seasalt_mod.F
|
 IF ( LMPOA <span style="color:red">> 0</span> ) THEN
|
 IF ( LMPOA ) THEN

|-valign="top"
|
 strat_chem_mod.F90
|
 <span style="color:red">'A3O2',    'ACET', etc.</span>
|
 <span style="color:green">'A3O2    ', 'ACET    ', etc.
 
 # All strings in array constructors must
 # have the same number of spaces</span>

|-valign="top"
|
 strat_chem_mod.F90
|
 <span style="color:red">TYPE(BrPointers)      :: BrPtrDay(6) 
 TYPE(BrPointers)      :: BrPtrNight(6)</span> 
|
 <span style="color:green">TYPE(BrPointers), POINTER :: BrPtrDay(:) 
 TYPE(BrPointers), POINTER :: BrPtrNight(:) </span>
 . . .
 <span style="color:green"><nowiki>!</nowiki> BrPtrDay and BrPtrNight have to be allocated dynamically
 <nowiki>!</nowiki> because they are pointers (bmy, 10/3/16)
 ALLOCATE( BrPtrDay  ( 6 ), STAT=errCode )
 ALLOCATE( BrPtrNight( 6 ), STAT=errCode )</span>
 . . .
 <span style="color:green">IF ( ASSOCIATED( BrPtrDay   ) ) DEALLOCATE( BrPtrDay   )
 IF ( ASSOCIATED( BrPtrNight ) ) DEALLOCATE( BrPtrNight )
 
 # Dynamically allocate the BrPtrDay and BrPtrNight arrays
 # so as to not have to declare them SAVEd.</span>

|-valign="top"
|
 ucx_mod.F90
|
 <span style="color:red">JDIF_OUT = SIND(JMAX_OUT)-SIND(JMIN_OUT)</span>
 . . .
 <span style="color:red">JDIF_TMP = SIND(JMAX_TMP)-SIND(JMIN_TMP)</span>
|
 <span style="color:green">USE PhysConstants,      ONLY : PI_180</span>
 . . .
  <span style="color:green">JDIF_OUT = SIN( JMAX_OUT * PI_180 )
 &         - SIN( JMIN_OUT * PI_180 )</span>
 . . .
  <span style="color:green">   JDIF_TMP = SIN( JMAX_TMP * PI_180 )
 &            - SIN( JMIN_TMP * PI_180 )
 
 # Remove unsupported SIND function</span>

|}

--[[User:Bmy|Bob Yantosca]] ([[User talk:Bmy|talk]]) 19:45, 28 September 2016 (UTC)

=== Modules in GeosUtil ===

{| border=1 cellspacing=0 cellpadding=5 
|- bgcolor="#CCCCCC"
!width="200px"|Module
!width="450px"|Code removed (in <span style="color:red">RED</span>)
!width="600px"|Code added (in <span style="color:green">GREEN</span>)

|-valign="top"
|
 error_mod.F
|
 #if defined( LINUX_IFORT ) 
|
 #if defined( LINUX_IFORT ) <span style="color:green">|| defined( LINUX_GFORTRAN )</span>

|-valign="top"
|
 geos_timers_mod.F
|
 IF ( (SavedTimers(TimerLoc)%ENABLED) <span style="color:red">.eq. .true.</span>)
|
 IF ( SavedTimers(TimerLoc)%ENABLED )

|-valign="top"
|
 henry_coeffs_mod.F
|
 'H2O2',   ! Jacob et al. 2000 
 'CH3I',   ! Moore et al. 1995
 <span style="color:red">'DMS'</span>,    ! De Bruyn et al. 1995
 'ACET' )/ ! Benkelberg et al 1995
|
 'H2O2',   ! Jacob et al. 2000 
 'CH3I',   ! Moore et al. 1995
 <span style="color:green">'DMS ',</span>   ! De Bruyn et al. 1995
 'ACET' )/ ! Benkelberg et al 1995
 
 <span style="color:green"># All strings in an array constructor
 # must have the same number of spaces</span>

|}

--[[User:Bmy|Bob Yantosca]] ([[User talk:Bmy|talk]]) 19:13, 28 September 2016 (UTC)

=== Modules in Headers ===

{| border=1 cellspacing=0 cellpadding=5 
|- bgcolor="#CCCCCC"
!width="200px"|Module
!width="450px"|Code removed (in <span style="color:red">RED</span>)
!width="600px"|Code added (in <span style="color:green">GREEN</span>)

|-valign="top"
|
 input_opt_mod.F90
|
 <span style="color:red">INTEGER</span>                    :: LND51_HDF
 . . .
 <span style="color:red">INTEGER</span>                    :: LND51b_HDF
 . . .
 <span style="color:red">INTEGER</span>                    :: LWINDO_SE
 . . .
 <span style="color:red">INTEGER</span>                    :: LWINDO_CU
|
 <span style="color:green">LOGICAL</span>                     :: LND51_HDF
 . . .
 <span style="color:green">LOGICAL</span>                     :: LND51b_HDF
 . . .
 <span style="color:green">LOGICAL</span>                     :: LWINDO_SE
 . . .
 <span style="color:green">LOGICAL</span>                     :: LWINDO_CU

|}

--[[User:Bmy|Bob Yantosca]] ([[User talk:Bmy|talk]]) 19:49, 28 September 2016 (UTC)

=== Modules in HEMCO/Core ===

{| border=1 cellspacing=0 cellpadding=5 
|- bgcolor="#CCCCCC"
!width="200px"|Module
!width="450px"|Code removed (in <span style="color:red">RED</span>)
!width="600px"|Code added (in <span style="color:green">GREEN</span>)

|-valign="top"
|
 hco_calc_mod.F90
|
 IF ( UseConc <span style="color:red">/=</span> Dct%Dta%IsConc ) THEN
|
 IF ( UseConc <span style="color:green">.neqv.</span> Dct%Dta%IsConc ) THEN
 
 <span style="color:green"># Use .eqv. and .neqv. instead of == or /=
 # when comparing LOGICAL variables directly.</span>

|-valign="top"
|
 hco_datacont_mod.F90
|
  TYPE(ListCont), POINTER :: TmpLct <span style="color:red">=> NULL()</span>
|
  TYPE(ListCont), POINTER :: TmpLct
  . . .
  <span style="color:green">TmpLct => NULL()</span>

|-valign="top"
|
 hco_diagn_mod.F90
|
 INTERFACE Diagn_Update
    <span style="color:red">MODULE PROCEDURE Diagn_UpdateSP
    MODULE PROCEDURE Diagn_UpdateDP</span>
 END INTERFACE
 
 <span style="color:red"># Removed DiagnUpdateSP and DiagnUpdateDP
 # which used OPTIONAL arguments</span>
|
 <span style="color:green">PRIVATE :: Diagn_UpdateSp0d
 PRIVATE :: Diagn_UpdateSp0d
 PRIVATE :: Diagn_UpdateSp2d
 PRIVATE :: Diagn_UpdateSp3d
 PRIVATE :: Diagn_UpdateDp0d
 PRIVATE :: Diagn_UpdateDp2d
 PRIVATE :: Diagn_UpdateDp3d</span>
 ...
 INTERFACE Diagn_Update
    <span style="color:green">MODULE PROCEDURE Diagn_UpdateSp0d
    MODULE PROCEDURE Diagn_UpdateSp2d
    MODULE PROCEDURE Diagn_UpdateSp3d
    MODULE PROCEDURE Diagn_UpdateDp0d
    MODULE PROCEDURE Diagn_UpdateDp2d
    MODULE PROCEDURE Diagn_UpdateDp3d</span>
 END INTERFACE
 
 <span style="color:green"># Added new routines DiagnUpdateSp* and DiagUpdateDp*
 # to avoid using OPTIONAL arguments in a MODULE INTERFACE</span>

|-valign="top"
|
 hco_interp_mod.F90
|
 <span style="color:red">WHERE ( REGFRACS > MAXFRACS ) 
    MAXFRACS = REGR_4D
    INDECES  = IVAL
 END WHERE</span>
|
 <span style="color:green">DO T  = 1, NTIME
 DO L  = 1 ,NLEV
 DO J  = 1, HcoState%NY
 DO I2 = 1, HcoState%NX
    IF ( REGFRACS(I2,J,L,T) > MAXFRACS(I2,J,L,T) ) THEN
       MAXFRACS(I2,J,L,T) = REGR_4D(I2,J,L,T)
       INDECES (I2,J,L,T) = IVAL
    ENDIF
 ENDDO
 ENDDO
 ENDDO
 ENDDO</span>
   
|-valign="top"
|
 hco_unit_mod.F90
|
 <span style="color:red">'1',             &
 'count',         &
 'unitless',      &</span>
 etc.
|
 <span style="color:green">'1          ',   &
 'count      ',   &
 'unitless   ',   &</span>
 etc.
  
 <span style="color:green"># All strings in array constructors must have the
 # same # of spaces or else Gfortran chokes</span>
 
|}

--[[User:Bmy|Bob Yantosca]] ([[User talk:Bmy|talk]]) 18:21, 26 September 2016 (UTC)

=== Modules in HEMCO/Extensions ===

{| border=1 cellspacing=0 cellpadding=5 
|- bgcolor="#CCCCCC"
!width="200px"|Module
!width="450px"|Code removed (in <span style="color:red">RED</span>)
!width="600px"|Code added (in <span style="color:green">GREEN</span>)

|-valign="top"
|
 hcox_gfed_mod.F90
|
 REAL(sp), POINTER   :: TmpPtr(:,:)  <span style="color:red">=> NULL()</span>
 |
 REAL(sp), POINTER   :: <span style="color:green">TmpPtr(:,:)</span>
 ...
 <span style="color:green">TmpPtr => NULL()
 
 # Setting a pointer to NULL where it is declared
 # turns the pointer into a SAVEd variable.</span>

|-valign="top"
|
 hcox_gfed_mod.F90
|
 <nowiki>!</nowiki> Get pointers to GFED3 data
 IF ( IsGFED3 ) THEN
    .... etc ...
 
 <nowiki>!</nowiki> Get pointers to GFED4 data
 ELSEIF ( IsGFED4 ) THEN
     ... etc ...
  
 ENDIF
 
  <span style="color:red"><nowiki>!</nowiki> Make sure HUMTROP does not exceed one
  WHERE ( HUMTROP > 1.0_sp ) 
     HUMTROP = 1.0_sp
  END WHERE</span>
|
 <nowiki>!</nowiki> Get pointers to GFED3 data
 IF ( IsGFED3 ) THEN
    .... etc ...
 	 
    <span style="color:green"><nowiki>!</nowiki> Make sure HUMTROP does not exceed one
    WHERE ( HUMTROP > 1.0_sp ) 
       HUMTROP = 1.0_sp
    END WHERE</span>
 
 <nowiki>!</nowiki> Get pointers to GFED4 data
 ELSEIF ( IsGFED4 ) THEN
    ... etc ...
 ENDIF
 
 <span style="color:green"># HUMTROP is only defined for GFED3, so the
 # WHERE statement should go in the IfGFED3 block.</span>

|-valign="top"
|
 hcox_paranox_mod.F90
|
 VARS(4) = <span style="color:red">ASIND( SC5(I,J) )</span>
 VARS(5) = <span style="color:red">ASIND( ExtState%SUNCOS%Arr%Val(I,J) )</span>
|
 VARS(4) = <span style="color:green">ASIN( SC5(I,J)                     ) / HcoState%Phys%PI_180</span>
 VARS(5) = <span style="color:green">ASIN( ExtState%SUNCOS%Arr%Val(I,J) ) / HcoState%Phys%PI_180
 
 # ASIND is not supported in Gfortran.  Use ASIN instead to compute
 # the arcsin and convert from degrees to radians manually.

|-valign="top"
|
 hcox_seasalt_mod.F90
|
 ALLOCATE ( NR  ( NSALT ), STAT=AS )
 IF ( AS/=0 ) THEN
   CALL HCO_ERROR( 'Cannot allocate NR', RC )
   RETURN
 ENDIF
 <span style="color:red">SS_DEN = 2200.d0</span>
 
 ALLOCATE ( SS_DEN  ( NSALT ), STAT=AS ) 
|
 ALLOCATE ( NR  ( NSALT ), STAT=AS )
 IF ( AS/=0 ) THEN
   CALL HCO_ERROR( 'Cannot allocate NR', RC )
   RETURN
 ENDIF
 <span style="color:green">NR = 0</span>
 
 ALLOCATE ( SS_DEN  ( NSALT ), STAT=AS ) 
 
 <span style="color:green"># Don't refer to SS_DEN before it is allocated</span>

|}

--[[User:Bmy|Bob Yantosca]] ([[User talk:Bmy|talk]]) 21:07, 26 September 2016 (UTC)

=== Modules in NcdfUtil ===

{| border=1 cellspacing=0 cellpadding=5 
|- bgcolor="#CCCCCC"
!width="200px"|Module
!width="450px"|Code removed (in <span style="color:red">RED</span>)
!width="600px"|Code added (in <span style="color:green">GREEN</span>)

|-valign="top"
|
 ncdf_mod.F90
|
  <span style="color:red">PRIVATE :: NC_VAR_WRITE_INT
  PRIVATE :: NC_VAR_WRITE_R4
  PRIVATE :: NC_VAR_WRITE_R8
|
  <span style="color:green">PRIVATE :: NC_VAR_WRITE_INT_1D
  PRIVATE :: NC_VAR_WRITE_INT_2D
  PRIVATE :: NC_VAR_WRITE_INT_3D
  PRIVATE :: NC_VAR_WRITE_INT_4D
  PRIVATE :: NC_VAR_WRITE_R4_1D
  PRIVATE :: NC_VAR_WRITE_R4_2D
  PRIVATE :: NC_VAR_WRITE_R4_3D
  PRIVATE :: NC_VAR_WRITE_R4_4D
  PRIVATE :: NC_VAR_WRITE_R8_1D
  PRIVATE :: NC_VAR_WRITE_R8_2D
  PRIVATE :: NC_VAR_WRITE_R8_3D
  PRIVATE :: NC_VAR_WRITE_R8_4D
  
  <span style="color:green">#Added separate, overloaded routine 
  # for each combination of array size and INT, REAL*4, REAL*8.  
  # This avoids using OPTIONAL arguments in overloaded routines, 
  #which Gfortran hates.</span>

|-valign="top"
|
 ncdf_mod.F90
|
  </span>INTERFACE NC_VAR_WRITE
     <span style="color:red">MODULE PROCEDURE NC_VAR_WRITE_INT
     MODULE PROCEDURE NC_VAR_WRITE_R4
     MODULE PROCEDURE NC_VAR_WRITE_R8</span>
  END INTERFACE NC_VAR_WRITE
|
  </span>INTERFACE NC_VAR_WRITE
     <span style="color:green">MODULE PROCEDURE NC_VAR_WRITE_INT_1D
     MODULE PROCEDURE NC_VAR_WRITE_INT_2D
     MODULE PROCEDURE NC_VAR_WRITE_INT_3D
     MODULE PROCEDURE NC_VAR_WRITE_INT_4D
     MODULE PROCEDURE NC_VAR_WRITE_R4_1D
     MODULE PROCEDURE NC_VAR_WRITE_R4_2D
     MODULE PROCEDURE NC_VAR_WRITE_R4_3D
     MODULE PROCEDURE NC_VAR_WRITE_R4_4D
     MODULE PROCEDURE NC_VAR_WRITE_R8_1D
     MODULE PROCEDURE NC_VAR_WRITE_R8_2D
     MODULE PROCEDURE NC_VAR_WRITE_R8_3D
     MODULE PROCEDURE NC_VAR_WRITE_R8_4D</span>
  END INTERFACE NC_VAR_WRITE
 
|-valign="top"
|
 ncdf_mod.F90
|
 READ( tUnit(L1:L2), <span style="color:red">'(i)'</span>, IOSTAT=STAT ) tYr 
 READ( tUnit(L1:L2), <span style="color:red">'(i)'</span>, IOSTAT=STAT ) tMt
 READ( tUnit(L1:L2), <span style="color:red">'(i)</span>', IOSTAT=STAT ) tDy
 READ( tUnit(L1:L2), <span style="color:red">'(i)</span>', IOSTAT=STAT ) tHr
 READ( tUnit(L1:L2), <span style="color:red">'(i)</span>', IOSTAT=STAT ) tMn
 READ( tUnit(L1:L2), <span style="color:red">'(i)</span>', IOSTAT=STAT ) tSc
|
 READ( tUnit(L1:L2), <span style="color:green">'(i4)'</span>, IOSTAT=STAT ) tYr 
 READ( tUnit(L1:L2), <span style="color:green">'(i2)</span>', IOSTAT=STAT ) tMt
 READ( tUnit(L1:L2), <span style="color:green">'(i2)</span>', IOSTAT=STAT ) tDy
 READ( tUnit(L1:L2), <span style="color:green">'(i2)</span>', IOSTAT=STAT ) tHr
 READ( tUnit(L1:L2), <span style="color:green">'(i2)</span>', IOSTAT=STAT ) tMn 
 READ( tUnit(L1:L2), <span style="color:green">'(i2)</span>', IOSTAT=STAT ) tSc
 
 <span style="color:green"># Gfortran cannot have a generic integer 
 #width in a FORMAT statement.</span>

|-valign="top"
|
 ncdf_mod.F90
|
 READ( TIMEUNIT(L1:L2), <span style="color:red">'(i)'</span>, IOSTAT=STAT ) YYYY
 READ( TIMEUNIT(L1:L2), <span style="color:red">'(i)'</span>, IOSTAT=STAT ) MM
 READ( TIMEUNIT(L1:L2), <span style="color:red">'(i)'</span>, IOSTAT=STAT ) DD
 READ( TIMEUNIT(L1:L2), <span style="color:red">'(i)'</span>, IOSTAT=STAT ) HH
|
 READ( TIMEUNIT(L1:L2), <span style="color:green">'(i4)'</span>, IOSTAT=STAT ) YYYY
 READ( TIMEUNIT(L1:L2), <span style="color:green">'(i2)'</span>, IOSTAT=STAT ) MM
 READ( TIMEUNIT(L1:L2), <span style="color:green">'(i2)'</span>, IOSTAT=STAT ) DD
 READ( TIMEUNIT(L1:L2), <span style="color:green">'(i2)'</span>, IOSTAT=STAT ) HH
 
 <span style="color:green"># Gfortran cannot have a generic integer 
 #width in a FORMAT statement.</span>

|}

--[[User:Bmy|Bob Yantosca]] ([[User talk:Bmy|talk]]) 15:14, 29 September 2016 (UTC)

=== Fixes for technical issues in the specialty simulation modules ===

In addition to the modifications listed in the tables above, we fixed a few technical issues were in the various GEOS-Chem specialty simulation modules.  These issues were discovered by running complete set of [[GEOS-Chem Unit Tester|GEOS-Chem unit tests]] using the GNU Fortran compiler.

*[[Mercury#Avoid_floating-point_exception_in_OCEAN_MERCURY_READ|Avoid floating-point exception in OCEAN_MERCURY_READ]]
*[[POPs_simulation#Avoid_floating-point_exception_in_routine_CHEM_POPGP|Avoid floating-point error in routine CHEM_POPGP]]
*[[Aerosol-only_simulation#Fixed_undefined_molecular_weight_of_HNO3_in_routine_SEASALT_CHEM|Fixed undefined molecular weight of HNO3 in routine SEASALT_CHEM]]

--[[User:Bmy|Bob Yantosca]] ([[User talk:Bmy|talk]]) 16:54, 11 October 2016 (UTC)

== Validation ==

To be added soon.

== Known issues ==

=== Now us ASIN instead of obsolete function ASIND ===

In HEMCO module <code>HEMCO/Extensions/hcox_paranox_mod.F90</code>, we had to replace the non-standard function ASIND (which returns the arcsine of an argument in degrees) with the ASIN function (which returns the arcsin in radians).  The proper unit conversion was then applied.

The code in <span style="color:red">RED</span> was deleted:

 VARS(4) = <span style="color:red">ASIND( SC5(I,J) )</span>
 VARS(5) = <span style="color:red">ASIND( ExtState%SUNCOS%Arr%Val(I,J) )</span>

And the code in <span style="color:green">GREEN</span> was added:

 VARS(4) = <span style="color:green">ASIN( SC5(I,J)                     ) / HcoState%Phys%PI_180</span>
 VARS(5) = <span style="color:green">ASIN( ExtState%SUNCOS%Arr%Val(I,J) ) / HcoState%Phys%PI_180

This modification does not cause any numerical differences in the output.  All difference tests returned identical results.

--[[User:Bmy|Bob Yantosca]] ([[User talk:Bmy|talk]]) 21:00, 30 September 2016 (UTC)

=== Now use SIN instead of obsolete function SIND ===

In the UCX chemistry module <code>GeosCore/ucx_mod.F</code>, we had to replace the non-standard function SIND (which returns the sine of an argument in degrees) with the SIN function (which returns the sine in radians).  The argument was converted from degrees to radians before being passed to SIN.

The code in <span style="color:red">RED</span> was deleted:

 <span style="color:red">JDIF_OUT = SIND(JMAX_OUT)-SIND(JMIN_OUT)</span>
 . . .
 <span style="color:red">JDIF_TMP = SIND(JMAX_TMP)-SIND(JMIN_TMP)</span>

And the code in <span style="color:green">GREEN</span> was added:

 <span style="color:green">USE PhysConstants,      ONLY : PI_180</span>
 . . .
  <span style="color:green">JDIF_OUT = SIN( JMAX_OUT * PI_180 )
 &         - SIN( JMIN_OUT * PI_180 )</span>
 . . .
  <span style="color:green">   JDIF_TMP = SIN( JMAX_TMP * PI_180 )
 &            - SIN( JMIN_TMP * PI_180 )
 
This change causes very small numerical differences, at the level of numerical noise, only in the "Standard" and "UCX" chemical mechanisms.  We attribute these differences to conversion between degrees and radians, probably with a different value of PI in each case.

--[[User:Bmy|Bob Yantosca]] ([[User talk:Bmy|talk]]) 21:14, 30 September 2016 (UTC)

=== THREADPRIVATE error in GNU Fortran 5 and higher ===

GNU Fortran v5 and higher seems to not like how the <code>!$OMP THREADPRIVATE</code> statements in the ISORROPIA code are written.  It does not let you place a <code>COMMON</code> block name in an <code>!$OMP THREADPRIVATE</code> statement.  We are still looking into a workaround for this.

'''''[[User:Bmy|Bob Yantosca]] wrote:'''''

<blockquote>I tried loading these modules on the holy2a81302 node of Odyssey:</blockquote>

      Currently Loaded Modules:
       1) git/2.1.0-fasrc01     5) gmp/6.1.1-fasrc02    9) openmpi/1.10.3-fasrc01
       2) perl/5.10.1-fasrc04   6) mpfr/3.1.4-fasrc02  10) zlib/1.2.8-fasrc07
       3) IDL/8.5.1-fasrc01     7) mpc/1.0.3-fasrc04   11) hdf5/1.8.12-fasrc12
       4) ncl/6.1.2             8) gcc/6.1.0-fasrc01   12) netcdf/4.1.3-fasrc09

<blockquote>And then I tried to compile GEOS-Chem with Gfortran 6.1.0, but I got a bunch of these errors:</blockquote>

      !$OMP THREADPRIVATE( /CASE/ ) Error: !$OMP THREADPRIVATE statement 
      is not allowed inside of BLOCK DATA at (1) isrpia.inc:232:72:

<blockquote>This is occurring in a section of 3rd-party, legacy code that is not easily rewritten.  I looked on the internet and found [http://openmp.org/forum/viewtopic.php?f=3&t=1866 this forum on Openmp.org].  So it appears that it is not an OpenMP problem, but Gfortran is being petty and pedantic here.  Gfortran seems to be very unfriendly to legacy codes.</blockquote>

--[[User:Bmy|Bob Yantosca]] ([[User talk:Bmy|talk]]) 20:37, 28 September 2016 (UTC)

=== GNU Fortran 4.8 and higher is not compatible with Totalview ===

GNU Fortran 4.8 and higher versions save debugging information to disk using the DWARF library, version 4, by default.  This makes code compiled with GNU Fortran incompatible with the Totalview debugger, version 8.8.0.1, which can only accept DWARF v2 or DWARF v3 input.  The GEOS-Chem Support Team noticed this on our local computer cluster.  We are still looking into a workaround for this.

--[[User:Bmy|Bob Yantosca]] ([[User talk:Bmy|talk]]) 21:39, 30 September 2016 (UTC)