Difference between revisions of "Intel Fortran Compiler"
(→Timing results: IFORT 11 vs. IFORT 10) |
(→Performance) |
||
Line 127: | Line 127: | ||
== Performance == | == Performance == | ||
+ | |||
+ | === 7-day time test results === | ||
+ | |||
+ | Please see [[GEOS-Chem_performance#7-day_time_tests|this list of timing results]] done with [[GEOS-Chem v10-01]] and [[GEOS-Chem v11-01]]. | ||
=== Timing results: IFORT 11 vs. IFORT 10 === | === Timing results: IFORT 11 vs. IFORT 10 === | ||
Line 208: | Line 212: | ||
=== Comparison between IFORT 9.1 and IFORT 10.1 === | === Comparison between IFORT 9.1 and IFORT 10.1 === | ||
+ | |||
+ | [[Image:Obsolete.jpg]] | ||
+ | |||
+ | <span style="color:red">'''''The GEOS-Chem versions listed below are obsolete. Please see [[GEOS-Chem_performance#7-day_time_tests|this list of timing results]] done with more recent versions of GEOS-Chem.'''''</span> | ||
The table shows the wallclock time and mean OH for several GEOS-Chem simulations that were done in order to compare Intel Fortran Compiler (IFORT) v9.1 vs. v10.1.013. The simulations had all these things in common: | The table shows the wallclock time and mean OH for several GEOS-Chem simulations that were done in order to compare Intel Fortran Compiler (IFORT) v9.1 vs. v10.1.013. The simulations had all these things in common: |
Revision as of 20:22, 3 January 2017
This page contains information about the Intel Fortran Compiler (aka "IFORT" compiler). The IFORT compiler is the preferred compiler for use with GEOS-Chem.
Contents
- 1 Documentation
- 2 Environment settings for Intel Fortran
- 3 Performance
- 4 Optimization
- 5 Known issues
- 5.1 Resetting stacksize for Linux
- 5.2 Out of memory asking for NNNNN
- 5.3 Bug fix for GEOS-Chem compiled with Intel Fortran Compiler 12
- 5.4 Error compiling with IFORT 12 and Mac OS X
- 5.5 Compilation error with IFORT 12
- 5.6 Relocation truncated to fit error
- 5.7 Problems with IFORT 11.0.xxx
- 5.8 Problem with IFORT 11 and GEOS-Chem adjoint
- 5.9 Incompatibility between IFORT 11 and OS version
- 5.10 Error in partition.f when compiling with IFORT 10
- 5.11 KPP not compatibile with IFORT 9.1
- 5.12 Speedup With Hyperthreading on Nehalem chips
- 5.13 Performance bottleneck caused by inefficient subroutine calls
- 5.14 Bugs in the IFORT compiler cause HEMCO to segfault
Documentation
NOTE: Recent Intel compiler versions (e.g. IFORT 12, IFORT 13) are now packaged and sold under the name Intel Fortran Composer XE (or something similar). Many GEOS-Chem users still use older versions (i.e. IFORT 11, IFORT 10).
You can find more information about the Intel Fortran Compiler here:
- Version 11.1
- Version 12.1
- Version 15.0
- Determining the cause of SIGSEGV or SIGBUS errors
- Dr. Fortran's Blog
Also, normally when you installs the Intel Fortran compilers, you also will install the C and C++ compilers. These compilers are not needed for GEOS-Chem, but they will be needed if you install libraries (e.g. netCDF or HDF5) on your system.
--Bob Yantosca (talk) 18:19, 3 November 2016 (UTC)
Environment settings for Intel Fortran
Here is some information about how you can customize your Unix environment to use the Intel Fortran compiler.
On most Unix clusters, the module command is used to load the Intel Fortran compiler library into your Unix enviroment, along with any related libraries (e.g. netCDF, OpenMPI). For example, we use these commands on the Harvard cluster (odyssey.rc.fas.harvard.edu):
# These commands load Intel Fortran 15 module load intel/15.0.0-fasrc01 openmpi/1.8.3-fasrc02 netcdf/4.1.3-fasrc04 # These commands load Intel Fortran 13 module load intel/13.0.079-fasrc01 openmpi/1.8.1-fasrc05 netcdf/4.1.3-fasrc06 # These commands load Intel Fortran 11 module load intel/11.1 GEOS-Chem-Libraries
You can ask your IT staff what the corresponding commands would be on your particular cluster.
The module command should automatically define several environment variables for you:
Variable | Expected setting | Description |
---|---|---|
FC | ifort | Name of the Intel Fortran compiler |
CC | icc | Name of the Intel C compiler |
CXX | icpc | Name of the Intel C++ compiler |
NETCDF_HOME | System-dependent | Path to the root netCDF folder |
NETCDF_INCLUDE | System-dependent | Path to the netCDF include folder (e.g. $NETCDF_HOME/include) |
NETCDF_LIB | System-dependent | Path to the netCDF library folder (e.g. $NETCDF_HOME/lib or $NETCDF_HOME/lib64) |
NETCDF_FORTRAN_HOME | System-dependent | Path to the root netCDF Fortran folder |
NETCDF_FORTRAN_INCLUDE | System-dependent | Path to the netCDF Fortran include folder (e.g. $NETCDF_FORTRAN_HOME/include) |
NETCDF_FORTRAN_LIB | System-dependent | Path to the netCDF Fortran library folder (e.g. $NETCDF_FORTRAN_HOME/lib or $NETCDF_FORTRAN_HOME/lib64) |
If these variables are not automatically set by the module command on your system (or if your system does not use the module command), then:
- Set the FC, CC, and CXX variables manually in your startup script (e.g. .bashrc or .cshrc).
- Ask your IT staff where the netCDF library paths are located, and set the NETCDF_HOME, NETCDF_INCLUDE, and NETCDF_LIB environment variables accordingly.
Depending on your system, you may find that the netCDF Fortran libraries are installed into a different folder than the rest of the netCDF library files. If this is the case, then the module command should automatically define the NETCDF_FORTRAN_HOME, NETCDF_FORTRAN_INCLUDE, and NETCDF_FORTRAN_LIB environment variables. If not, then ask your IT staff what the proper paths are so that you can set these variables manually.
--Bob Yantosca (talk) 19:34, 4 October 2016 (UTC)
Requesting sufficient stack memory for GEOS-Chem
In order to run GEOS-Chem with Intel Fortran, you must request the maximum amount of stack memory in your Unix environment. (The stack memory is where local automatic variables and temporary !$OMP PRIVATE
variables will be created.) Add the following lines to your system startup file:
If you use bash. add this to your .bashrc file |
If you use csh or tcsh, add this to your .cshrc file |
---|---|
ulimit -s unlimited export OMP_STACKSIZE=500m |
limit stacksize unlimited setenv OMP_STACKSIZE 500m |
The ulimit -s unlimited (for bash) or limit stacksize unlimited commands tell the Unix shell to use the maximum amount of stack memory available.
The environment variable OMP_STACKSIZE 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 500m 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.
NOTE: Setting the OMP_STACKSIZE environment variable will make it easier to switch between different compilers on your system. The KMP_STACKSIZE environment variable only works with the Intel Fortran Compiler but not with GNU Fortran.
--Bob Yantosca (talk) 21:35, 4 October 2016 (UTC)
Performance
7-day time test results
Please see this list of timing results done with GEOS-Chem v10-01 and GEOS-Chem v11-01.
Timing results: IFORT 11 vs. IFORT 10
The GEOS-Chem versions listed below are obsolete. Please see this list of timing results done with more recent versions of GEOS-Chem.
Description
The table below shows the wallclock time and mean OH for several GEOS-Chem simulations that were done in order to compare IFORT 10.1.013 vs. IFORT 11.1.069. The simulations had all these things in common:
- GEOS-Chem v8-02-04
- 4x5 GEOS-5 met fields for month of 2008/07
- 1-month of simulation (0 GMT 2005/07/01 to 0 GMT 2005/08/01)
- Base compiler options: -cpp -w -O2 -auto -noalign -convert big_endian
- KPP compiler turned on
- Linoz stratospheric chemistry turned on
- All simulations ran on virtual machines (kvm guests) with the following characteristics:
Machine Type x86_64 Operating System Linux Operating System Release 2.6.18-128.1.6.el5_lustre.1.8.0 CPU Count 8 CPUs CPU Speed 2659 MHz Memory Total 12008076.000 KB Swap Space Total 18415600.000 KB
Results
Run | IFORT version |
# CPUs | Wall clock (mm:ss) |
Parallel % | Mean OH (1e5 molec/cm3) |
---|---|---|---|---|---|
1 | 10.1.013 | 4 | 02:10:51 | 384.7% | 12.5205894678448 |
2 | 11.1.069 | 4 | 02:09:14 | 382.7% | 12.5217430768752 |
3 | 10.1.013 | 8 | 01:17:13 | 757.1% | 12.5205894678448 |
4 | 11.1.069 | 8 | 01:18:44 | 753.4% | 12.5213489686705 |
Here are some plots of surface ozone from the benchmarks simulations:
NOTES:
- The wall times and parallel % are more or less identical when moving from IFORT 10 to IFORT 11.
- The ideal parallelization percentages are 400% (on 4p) and 800% (on 8p).
- The differences in the surface ozone is approximately between IFORT 10 and IFORT 11 is due to numerical differences in the libraries and optimization. The absolute magnitude of the differences is approximately 1 ppt of ozone.
--Bob Y. 16:08, 25 March 2010 (EDT)
Comparison between IFORT 9.1 and IFORT 10.1
The GEOS-Chem versions listed below are obsolete. Please see this list of timing results done with more recent versions of GEOS-Chem.
The table shows the wallclock time and mean OH for several GEOS-Chem simulations that were done in order to compare Intel Fortran Compiler (IFORT) v9.1 vs. v10.1.013. The simulations had all these things in common:
- GEOS-Chem v8-01-01
- 4x5 GEOS-5 met fields
- 1-week of simulation (0 GMT 2008/01/01 to 0 GMT 2008/01/08)
- Base compiler options: -cpp -w -auto -noalign -convert big_endian
- Runs were done on the Harvard "Ceres" cluster (OS type "linux-rhel5-x86_64")
Run | IFORT version |
# CPUs | Optimization options | Wall clock (mm:ss) |
Speedup from IFORT 9.1 to IFORT 10.1 |
Speedup from 4 to 8 CPUs w/ the same compiler |
Mean OH (1e5 molec/cm3) |
---|---|---|---|---|---|---|---|
1 | 9.1 | 4 | -O2 | 36:16 | 11.2913755849576 | ||
2 | 10.1 | 4 | -O2 | 33:55 | 6.48% | 11.2913755842197 | |
3 | 9.1 | 4 | -O3 | 37:26 | 11.2913755849576 | ||
4 | 10.1 | 4 | -O3 | 33:36 | 10.24% | 11.2913755838124 | |
5 | 9.1 | 8 | -O2 | 24:15 | 33.13% | 11.2913755849576 | |
6 | 10.1 | 8 | -O2 | 22:46 | 6.12% | 32.88% | 11.2913755842197 |
7 | 9.1 | 8 | -O3 | 23:36 | 36.95% | 11.2913755849576 | |
8 | 10.1 | 8 | -O3 | 22:31 | 4.59% | 32.99% | 11.2913755838124 |
9 | 9.1 | 8 | -O3 -ipo -no-prec-div -static | 23:03 | 11.2913764967223 | ||
10 | 10.1 | 8 | -O3 -ipo -no-prec-div -static | 21:56 | 4.84% | 11.0809209646817 |
NOTES about the table:
- The column Speedup from IFORT 9.1 to IFORT 10.1 compares the wall clock time of equivalent runs done with IFORT 9.1 and IFORT 10.1. For example, the 6.48% speedup listed for Run #2 is comparing Run #2 to Run #1. Similarly Run #4 is compared against Run #3, etc.
- The column Speedup from 4 to 8 CPUs w/ the same compiler compares the wall clock time between runs with 4 CPUs and 8 CPUs for the same compiler (i.e. 4 CPUs on IFORT 9 vs 8 CPUs on IFORT 9, and ditto for IFORT 10). For example, the 33.13% speedup listed for Run #5 is comparing Run #5 to Run #1. Similarly, Run #6 is compared against Run #2, etc.
- The compiler options -O3 -ipo -non-prec-div -static correspond to IFORT's -fast optimization option. Using this option results in a mean OH concentration that is different than with the simpler optimization options of -O2 and -O3. This is because the -fast option sacrifices numerical accuracy for speed.
- With IFORT 9.1, switching from -O2 to -O3 does not change the mean OH concentration. Thus the bpch files of the runs were binary identical to each other.
- With IFORT 10.1, switching from -O2 to -O3 changes the mean OH concentration slightly. This implies that there are slight differences in the chemistry. However all runs done with -O2 have the same mean OH, as do all runs done with -O3.
PLOTS:
- Run #2 vs Run #1 (i.e. IFORT 9.1 vs IFORT 10.1 w/ -O2 on 4 CPUs)
- Run #3 vs Run #4 (i.e. IFORT 9.1 vs IFORT 10.1 w/ -O3 on 4 CPUs)
- Run #5 vs Run #9 (i.e. -O2 vs -fast with IFORT 9.1)
- Run #6 vs Run #10 (i.e. -O2 vs -fast with IFORT 10.1)
TAKE-HOME MESSAGE:
- IFORT 10.1 is always faster than the equivalent run with IFORT 9.1.
- IFORT 10.1 does indeed seem to optimize code better on machines with multi-core chipsets.
- For example: Run #6 (w/ IFORT 10) is 89 seconds faster per week than Run #5 (w/ IFORT 9) on 8 CPUs. This implies that a 52-week simulation with IFORT 10 on 8 CPUs would finish ~1hr 15m earlier than the equivalent IFORT 9 run.
- Switching from 4 to 8 CPU's results in a ~33% speedup for both IFORT 9.1 and IFORT 10.1.
- In general, switching from -O2 to -O3 (while using the same # of CPU's) does not result in a significant speedup. This is true for both IFORT 9.1 and IFORT 10.1.
OUR RECOMMENDATIONS:
- If possible, use IFORT 10.1 instead of IFORT 9.1
- Use the following compiler options (see Makefile.ifort):
- FFLAGS = -cpp -w -O2 -auto -noalign -convert big_endian
--Bob Y. 16:46, 16 April 2008 (EDT)
Upgrading to IFORT 10.1 does not seem to fix the stacksize problem listed below. You still need to manually reset the stacksize limit to a large positive number for both Linux and Altix platforms.
--Bob Y. 12:41, 25 April 2008 (EDT)
Optimization
In this section we present information about the various optimization options available in the Intel Fortran Compiler.
Optimization options
Here is a quick reference table of IFORT's optimization options (taken from the online Intel Fortran Compiler User and Reference Guides.
Option | Description | How invoked in GEOS-Chem? |
---|---|---|
-O0 | Turns off all optimizations. Math expressions will be evaluated in the same order in which they are written, which is necessary for debugging. If you are using a debugger (such as Totalview), compile with -g -O0. | DEBUG=yes or OPT=-O0 |
-O1 | Enables optimizations for speed and disables some optimizations that increase code size and affect speed. The -O1 option may improve performance for applications with very large code size, many branches, and execution time not dominated by code within loops.
Setting -O1 automatically sets the following options:
|
OPT=-O1 |
-O2 (aka -O) | Enables optimizations for speed. This is the generally recommended optimization level.
This option also enables:
On Linux and Mac OS X systems, if -g is specified, -O2 is turned off and -O0 is the default unless -O2 (or -O1 or -O3) is explicitly specified in the command line together with -g. |
Default setting |
-O3 | Enables -O2 optimizations plus more aggressive optimizations, such as prefetching, scalar replacement, and loop and memory access transformations.
Enables optimizations for maximum speed, such as:
On Linux and Mac OS X systems, the -O3 option sets option -fomitframe-pointer. The -O3 optimizations may not cause higher performance unless loop and memory access transformations take place. The optimizations may slow down code in some cases compared to -O2 optimizations. The -O3 option is recommended for applications that have loops that heavily use floating-point calculations and process large data sets. |
OPT=-O3 |
--Bob Y. 11:14, 3 October 2013 (EDT)
Recommended compilation and optimization options for GEOS-Chem
In this section, we present information about the compilation and optimization options that are invoked when you compile a GEOS-Chem simulation.
List of commonly-used compilation options
Here are the IFORT compilation options currently used by GEOS-Chem:
Option | Description | How invoked in GEOS-Chem? |
---|---|---|
Normal compiler settings | ||
-cpp | Turns on the C-preprocessor, to evaluate #if and #define statements in the source code.
|
Default setting |
-w | Suppresses all compiler warnings. This is mainly a convenience to prevent excessive output to the screen or log file.
NOTE: Most compiler warnings are harmless. Execution does not stop when a warning is displayed, unlike an error message, which causes program execution to halt at the point where the error occurred. |
Default setting |
-O2 | Optimizes the source code for speed, without taking too many liberties with numerical precision. For more information, please see the optimization options section above. | Default setting |
-auto | 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 |
-noalign | Prevents the compiler from padding bytes anywhere in common blocks and structures. Padding can affect numerical precision. | Default setting |
-convert big_endian | 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 |
-vec-report0 | Tells the compiler to suppress printing "LOOP HAS BEEN VECTORIZED" messages. This reduces the amount of output that is sent to the screen and/or GEOS-Chem log file. | Default setting |
-fp-model source | Rounds intermediate results to source-defined precision and enables value-safe optimizations. Basically, this tells the compiler not to take too many liberties with how numerical expressions are evaluated. For more information about this option, please see our precision-safe optimization section below. This option can be disabled by compiling GEOS-Chem with the PRECISE=no Makefile option. | Default setting |
Special compiler settings | ||
-r8 | This option tells the compiler to treat variables that are declared as REAL as REAL*8 (as opposed to REAL*4 .
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 |
Used as needed |
-mcmodel=medium | This option is used to tell IFORT to use more than 2GB of static memory. This avoids a 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° x 0.3125° nested grid). | Default setting |
-shared-intel (formerly -i-dynamic) |
This option needs to be used in conjunction with -mcmodel=medium. It causes Intel-provided libraries to be linked in dynamically instead of statically (which is the default). | Default setting |
-ipo | This option enables interprocedural optimization between files. This is also called multifile interprocedural optimization (multifile IPO) or Whole Program Optimization (WPO). When you specify this option, the compiler performs inline function expansion for calls to functions defined in separate files.
NOTE: Yuxuan Wang found that this option was useful for certain nested-grid simulations. See the this wiki post below for more information. |
IPO=yes |
-static | This option prevents linking with shared libraries. It causes the executable to link all libraries statically.
NOTE: Yuxuan Wang found that this option was useful for certain nested-grid simulations. See the this wiki post below for more information. |
IPO=yes |
Settings only used for debugging | ||
-debug all | Tells the compiler turn on all debug error output. | DEBUG=yes |
-g | 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). | DEBUG=yes |
-O0 | Turns off all optmization. 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). | DEBUG=yes |
-check bounds (aka -CB) | Check for array-out-of-bounds errors. This is invoked when you compile GEOS-Chem with the BOUNDS=yes Makefile option. NOTE: Only use -CB for debugging, as this option will cause GEOS-Chem to execute more slowly! | DEBUG=yes |
-check arg_temp_created | 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. | DEBUG=yes |
-traceback | 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. | TRACEBACK=yes |
-fpe0 | This option will cause GEOS-Chem to halt if any type of floating-point error is encountered. This can happen if an equation results in a denormal value, e.g. NaN, or +/-Infinity. Common causes of floating-point errors are divisions where the denominator becomes zero. NOTE: The default compiler setting is -fpe3, which will convert many of these denormal values to zeros and then continue execution. |
FPE=yes |
-ftrapuv | This option will assign a large numeric value to all local automatic variables. This makes it easier to identify numerical errors caused by improper initialization. | FPE=yes |
--Bob Y. 11:21, 3 October 2013 (EDT)
Typical settings for a GEOS-Chem simulation
The normal GEOS-Chem build uses the following IFORT compiler flags:
-cpp -w -O2 -auto -noalign -convert big_endian -vec-report0 -fp-model source -openmp
whereas a debugging run (meant to execute in a debugger such as TotalView) will typically use these flags:
-cpp -w -O0 -auto -noalign -convert big_endian -g -DDEBUG -check arg_temp_created -debug all -fp-model source -fpe0 -ftrapuv -check bounds
NOTE: In order to avoid running out of memory if you compiling GEOS-Chem at extremely high resolution (e.g. the 0.25° x 0.3125° nested grids), we recommend adding the following flags:
-mcmodel=medium -shared-intel
These are automatically set when you compile with the NETCDF=yes or HDF=yes compiler options (in GEOS-Chem v9-01-03 and higher).
--Bob Y. 17:34, 29 February 2012 (EST)
Precision-safe optimization
You can use the following Intel Fortran Compiler options to select how aggressively you would like to optimize floating-point operations.
Default behavior
-fp-model fast
Example source code:
REAL T0, T1, T2; ... T0 = 4.0E + 0.1E + T1 + T2;
When this option is specified, the compiler applies the following semantics:
- Additions may be performed in any order
- Intermediate expressions may use single, double, or extended precision
- The constant addition may be pre-computed, assuming the default rounding mode
Using these semantics, the following shows some possible ways the compiler may interpret the original code:
REAL T0, T1, T2; ... T0 = (T1 + T2) + 4.1E;
or
REAL T0, T1, T2; ... T0 = (T1 + 4.1E) + T2;
Preferred alternative
-fp-model source (aka -fp-model precise)
Example source code:
REAL T0, T1, T2; ... T0 = 4.0E + 0.1E + T1 + T2;
When this option is specified, the compiler applies the following semantics:
- Additions are performed in program order, taking into account any parentheses
- Intermediate expressions use the precision specified in the source code
- The constant addition may be pre-computed, assuming the default rounding mode
Using these semantics, the following shows a possible way the compiler may interpret the original code:
REAL T0, T1, T2; ... T0 = ((4.1E + T1) + T2);
Summary
If you do not select any -fp-model option, the Intel Fortran Compiler will default to -fp-model fast. As you can see from the examples above, this may not optimize the code in the same way each time. This can lead to minor numerical noise in the output, as was seen in ISORROPIA II.
To avoid this situation, we recommend compiling all source code files with -fp-model source. This will be the new default in GEOS-Chem v9-01-02.
Reference: Intel® Fortran Floating-point Operations; Document Number: 315892-003US
--Bob Y. 17:01, 25 August 2011 (EDT)
Optimization options for faster runs
Yuxuan Wang told us about the optimization options: -ipo and -static and said these options would speed up the simulations. I've tested these options on our system at Harvard. The run with the new options show very tiny differences (much less than 1% over 1 month) compared to a run with the old options only. For a full-chemistry run (43 tracers) on 4x5 resolution and 4 processors, the run time is about 10% shorter than previously.
These options are especially efficient to handle the transport. So in simulations with a faster chemistry (like tagged tracers simulations), we expect to see a higher gain in time. For example, the time for a methane run is shorten by about 30 %.
To use these options, compile GEOS-Chem with the IPO=yes Makefile option, e.g.
make -j4 IPO=yes
--Ccarouge 15:54, 8 September 2009 (EDT)
--Bob Y. 17:50, 29 February 2012 (EST)
Optimization level for debugging
If you would like to run your code in a debugger, such as Totalview, you must use the following compiler switches:
-g -O0
Using -O0 will ensure that the source code gets executed in the same order in which it is written (i.e. this disables all compiler optimizations). The -g switch will tell the debugger to display lines of source code instead of hexadecimal memory addresses (which are more or less gibberish unless you are a hardware engineer).
GEOS-Chem will add these switches automatically for you if you compile with the DEBUG=yes option.
--Bob Y. 15:28, 22 February 2012 (EST)
Known issues
Resetting stacksize for Linux
Overall machine memory limits are set with the Unix limit command. If you use csh or tcsh, you can set the following commands in your ~/.cshrc file:
# Max out machine limits limit cputime unlimited # NOTE: "Unlimited" is not truly unlimited. It limit filesize unlimited # will set the given limit to the maximum value limit datasize unlimited # determined by your hardware configuration limit stacksize unlimited limit coredumpsize unlimited limit memoryuse unlimited limit vmemoryuse unlimited limit descriptors unlimited limit memorylocked unlimited limit maxproc unlimited
Or if you use bash, you can add these to commands your ~/.bashrc file:
# Max out machine limits ulimit -t unlimited # cputime ulimit -f unlimited # filesize ulimit -d unlimited # datasize ulimit -s unlimited # stacksize ulimit -c unlimited # coredumpsize ulimit -m unlimited # memoryuse ulimit -v unlimited # vmemoryuse ulimit -n unlimited # descriptors ulimit -l unlimited # memorylocked ulimit -u unlimited # maxproc
NOTE: depending on your particular OS build (Linux, CentOS, Fedora, Ubuntu), not all of these limits will be used.
It is important to set the stacksize memory to the maximum value, because this will determine the amount of memory available for temporary variables, which are:
- variables that are not declared with the SAVE attribute
- variables that are not located at the top of a module
- variables that are not located in a common block.
However, one quirk is that the stacksize memory for child processes (i.e. processes spawned by CPUs within !$OMP PARALLEL DO
loops) are not set by the stacksize limit, but instead by the OMP_STACKSIZE environment variable. If OMP_STACKSIZE is not set with a high enough value, then your GEOS-Chem simulation may think it doesn't have enough memory to proceed, and may die with a segmentation fault error.
The fix for this situation is to make sure that you set OMP_STACKSIZE to a high value. It's OK if the value you give to OMP_STACKSIZE is larger than the largest amount of memory on your system. As long as it's set to a high positive number it will work.
If you use csh or tcsh, you can add this command to your ~/.cshrc file.
# Reset the child stack size to a large positive number # (It's OK if this is larger than the max value, it's just a kludge) setenv OMP_STACKSIZE 500m
Or if you use bash, add this command to your ~/.bashrc file:
# Reset the child stack size to a large positive number # (It's OK if this is larger than the max value, it's just a kludge) export OMP_STACKSIZE=500m
Resetting the OMP_STACKSIZE environment variable in this manner usually will correct the following errors:
- Segmentation fault encountered after TPCORE initialization
- Stack overflow
- forrtl: error (76): IOT trap signal
NOTE: We now recommend that you use the OMP_STACKSIZE variable instead of the KMP_STACKSIZE variable. This will make it easier to switch between compilers on your system. OMP_STACKSIZE works with all compilers, but KMP_STACKSIZE only works with Intel Fortran.
--Bob Y. 14:05, 5 November 2014 (EST)
Out of memory asking for NNNNN
Debra Weisenstein wrote:
- I'm trying to compile GEOS-CHEM (the stratospheric beta version that Seb Eastham at MIT is working on) on hpc at Harvard, and though it compiled for me early before, I've now been getting an "out of memory" error every time it tries to compile GeosCore/diag49_mod.F. I changed the compile option for mcmodel to large. Here is the compile line and error.
ifort -cpp -w -O2 -auto -noalign -convert big_endian -vec-report0 -mcmodel=large -i-dynamic -fp-model source -openmp -Dmultitask -I../Headers -module ../mod -I/home/dkweis/include -c diag49_mod.F Fatal compilation error: Out of memory asking for 36864.
- I've gone back and tried compiling older versions that I've compiled and run before and get a similar error:
ifort -cpp -w -O2 -auto -noalign -convert big_endian -vec-report0 -fp-model source -openmp -Dmultitask -DAPM -I../Headers -module ../mod -c dao_mod.F Fatal compilation error: Out of memory asking for 20480.
- Any idea what is going on? The "top" command shows 301780k of free memory (out of 8GB) and 906164k of free swap.
Bob Yantosca wrote:
- I googled the error message (“Out of memory asking for ….”) and I found this post online. The person who replied to this post gave 3 suggestions:
- Use -O2.
- Split up the source into smaller files if possible (may be difficult if it's a module.)
- Raise your datasize limit
- I'd start with the last. If you can't resolve it, please submit a test case to Intel Premier Support.
- So it may be that your datasize limit is not maxed out in your computational environment. You can check the setting of the system limits by typing:
limit
- at the Unix prompt. You’ll get a screen like this:
[54 bmy]% limit datasize unlimited stacksize unlimited ... etc ...
- Depending on your platform (i.e. combination of hardware + operating system) you may not have all of these limits available. The two most important limits for GEOS-Chem are datasize and stacksize. These should be set to unlimited. For good measure, you can also set the other available limits for your platform to unlimited as well. Follow the instructions in this wiki post. One caveat: if you try to set a limit that does not exist on your platform to unlimited, you will get a warning message from the Unix shell. You can then delete the offending limit setting from your .cshrc or .bashrc file.
--Bob Y. 14:58, 25 July 2013 (EDT)
Bug fix for GEOS-Chem compiled with Intel Fortran Compiler 12
Prasad Kasibhatla wrote:
- I am experiencing the IFORT 12 compile failure of GEOS-Chem v9-01-03 with nested grid options turned on in define.h. The compile seems to be failing in strat_chem_mod.F90 (see end of this msg). And to add - the compile succeeds for v9-01-03l with IFORT 12. I noticed that strat_chem_mod.F90 appeared post-v9-01-03l.
ifort -cpp -w -O2 -auto -noalign -convert big_endian -vec-report0 -mcmodel=medium -i-dynamic -fp-model source -openmp -Dmultitask -I../Headers -module ../mod -I/opt/geos-netcdf-4/include -c -free strat_chem_mod.F90 : catastrophic error: **Internal compiler error: segmentation violation signal raised** Please report this error along with the circumstances in which it occurred in a Software Problem Report. Note: File and line given may not be explicit cause of this error. compilation aborted for strat_chem_mod.F90 (code 1) make[3]: *** [strat_chem_mod.o] Error 1 make[3]: Leaving directory `/nfs/fire/psk9/Code.v9-01-03/GeosCore' make[2]: *** [lib] Error 2 make[2]: Leaving directory `/nfs/fire/psk9/Code.v9-01-03/GeosCore' make[1]: *** [all] Error 2 make[1]: Leaving directory `/nfs/fire/psk9/Code.v9-01-03/GeosCore' make: *** [all] Error 2
Bob Yantosca wrote:
- I have this version of IFORT installed locally:
[67 bmy Code.v9-02]% ifort -V Intel(R) Fortran Intel(R) 64 Compiler XE for applications running on Intel(R) 64, Version 12.1.3.293 Build 20120212 Copyright (C) 1985-2012 Intel Corporation. All rights reserved. FOR NON-COMMERCIAL USE ONLY
- And I got the exact same error as you did. I looked on the internet and it seemed that the issue is related to how IFORT 12 deals w/ OpenMP DO loops (click HERE and HERE).
- A little brute-force debugging revealed the offending parallel DO loop in strat_chem_mod.F90, in routine CALC_STE:
! Determine mean tropopause level for the period !$OMP PARALLEL DO & !$OMP DEFAULT( SHARED ) & !$OMP PRIVATE( I, J ) DO J = 1,JJPAR DO I = 1,IIPAR LTP(I,J) = NINT( TPauseL(I,J) / TPauseL_Cnt ) ENDDO ENDDO !$OMP END PARALLEL DO
- I also noticed that this DO loop is only executed for global simulations. Immediately preceding it there is an #if statement that exits out of the routine if you are doing an nested grid run:
#if defined( NESTED_NA ) || defined( NESTED_CH ) || defined( NESTED_EU ) ! This method only works for a global domain. ! It could be modified for nested domains if the total mass flux across the ! boundaries during the period is taken into account. RETURN #endif
- So given that this code doesn’t get used for nested runs, I added an #else block so that the code isn’t compiled in when you compile for nested grids:
#if defined( NESTED_NA ) || defined( NESTED_CH ) || defined( NESTED_EU ) ! This method only works for a global domain. ! It could be modified for nested domains if the total mass flux across the ! boundaries during the period is taken into account. RETURN !------------------------------------------------------------------------------ ! Prior to 10/5/12: ! Since the rest of this code isn't needed for the nested grid, wrap it ! in an #else statement. This might help the code to compile for IFORT 12. !#endif !------------------------------------------------------------------------------ #else ! Determine mean tropopause level for the period !$OMP PARALLEL DO & !$OMP DEFAULT( SHARED ) & !$OMP PRIVATE( I, J ) DO J = 1,JJPAR DO I = 1,IIPAR LTP(I,J) = NINT( TPauseL(I,J) / TPauseL_Cnt ) ENDDO ENDDO !$OMP END PARALLEL DO ... etc ... #endif END SUBROUTINE Calc_STE
- I’ve compiled this with the NA nested grid, the CH nested grid, and for 2 x 2.5 and 4 x 5 global GEOS-5 simulations, and the code does not get that compilation error. I’ve pushed this to the GEOS-Chem repository as a last-minute bug fix. I’m not sure why the error happens in the first place w/ the IFORT 12 compiler, but in any case, this fixes it.
--Bob Y. 13:52, 5 October 2012 (EDT)
Error compiling with IFORT 12 and Mac OS X
David Lary wrote:
- I am using Intel(R) 64, Version 12.1.3.289 Build 20120130 on Mac OS X 10.7.4. When I typed make, I get the following error:
ld: library not found for -lcrt1.10.6.o make[2]: *** [exe] Error 1 make[1]: *** [all] Error 2 make: *** [all] Error 2
Bob Yantosca replied:
- I think this is an issue particular to IFORT running on Mac OS X. It is probably not finding the proper library. I found a couple of posts (post #1 and post #2) on the internet that describe this error.
- Post #2 recommends adding the following path to your LD_LIBRARY_PATH environment variable:
/Developer/SDKs/MacOSX10.4u.sdk/usr/lib. (i.e. $(SDKROOT)/usr/lib)
--Bob Y. 10:43, 26 June 2012 (EDT)
Compilation error with IFORT 12
Geert Vinken wrote:
- I was trying to compile the GEOS-Chem v9-01-03k version, but I was getting a strange error when trying to do this with ifort 12.1.2:
biofuel_mod.F(363): error #5082: Syntax error, found IDENTIFIER 'X1' when expecting one of: ( % [ : . = => GENERIC8_1x1 = GENERIC_1x1 ----------------------------^ biofuel_mod.F(363): error #6404: This name does not have a type, and must have an explicit type. [GENERIC8_1] GENERIC8_1x1 = GENERIC_1x1 ------------------^ biofuel_mod.F(363): error #6460: This is not a field name that is defined in the encompassing structure. [X1] GENERIC8_1x1 = GENERIC_1x1 ----------------------------^ biofuel_mod.F(363): error #6366: The shapes of the array expressions do not conform. [X1] GENERIC8_1x1 = GENERIC_1x1 ----------------------------^ compilation aborted for biofuel_mod.F (code 1)
- Compiling with ifort 11.1 shows no problems at this module. Any of you ever heard of this problem or know a way around it? If I uncomment this line the model compiles, so it's just this line (and apparently the combination of 8_1x1 =) that's causing the problem.
- Also, renaming the GENERIC8_1x1 array to GENERIC8 seems to solve the problem.
Bob Yantosca replied:
- I think the default behavior of IFORT 12.1 has changed to adopt one of the more modern Fortran standards (F2003 or F2008). I found this entry in the Intel Fortran Compiler 12.1 manual which describes how one specifies floating-point constants.
- As shown in the entry above, the IFORT 12 compiler now interprets an underscore immediately preceded and followed by numbers as an alternate way to specify numbers in scientific notation. For example, the traditional way of representing a REAL*4 and REAL*8 constants:
REAL*4, PARAMETER :: PI = 3.14159e0 REAL*8, PARAMETER :: PI = 3.141592658979323d0
- can now also be represented as:
REAL*4, PARAMETER :: PI = 3.14159e0_4 REAL*8, PARAMETER :: PI = 3.141592658979323e0_8
- Therefore, the string:
GENERIC8_1x1
- is probably now parsed by the compiler as an exponential. This more than likely separates the string into substrings GENERIC8_1x1 and x1, which is what generates the error.
- I have looked for a compiler switch in the IFORT 12.1 manual that would disable this feature, but have not been successful. The short-term solution is to rename the variables so that it does not have an underscore surrounded by two numbers (e.g. from GENERIC8_1x1 to GENERIC_8_1x1).
--Bob Y. 18:13, 23 May 2012 (EDT)
Relocation truncated to fit error
If your code uses many large arrays, or if you are compiling an ultra-fine resolution version of GEOS-Chem (e.g. a 0.25° x 0.3125° GEOS-FP nested grid), then you may see this type of error:
Relocation truncated to fit: R_X86_64_32S against `.bss' Error"
The wording you get may differ slightly than the example shown above.
Long story short: IFORT is telling you that your program is trying to use more than 2GB of statically-allocated data (i.e. data space that is not declared with an ALLOCATABLE statement) at compile time. The default setting in IFORT is to expect to use less than 2GB of memory, so you are hitting the upper limit.
The solution is simple: recompile your code with the following compiler flags:
-mcmodel=medium -shared-intel
The -mcmodel=medium flag will tell IFORT that you expect to use more than 2GB of statically-allocated memory in your program. However, this also requires that you use link using dynamic libraries instead of the normal shared libraries. Using the -shared-intel flag will turn on the dynamic library linking. (Starting with GEOS-Chem v9-01-03, these compiler flags will be applied to the build sequence automatically.)
IMPORTANT NOTE! If your code links to any libraries such as HDF or netCDF, then you MUST rebuild each library, making sure that the Fortran and C compilers use the -mcmodel=medium option. Please see our Installing libraries for GEOS-Chem page for examples.
GEOS-Chem v9-01-03 and higher will automatically set these flags for you.
For more information, please see the following links:
--Bob Y. (talk) 18:24, 25 September 2015 (UTC)
Problems with IFORT 11.0.xxx
You should use GEOS-Chem with IFORT 11.1.058 or higher versions. Please see the discussion below about problems in the earlier versions of IFORT 11.0.xxx:
Tzung-May Fu wrote:
- I tested the Intel Fortran v11.0.074 compiler, but found that it is incompatible with the GC code. This is related to the partition.f bug that I reported earlier. (Actually, I'm not sure there is a bug in partition.f any more, unless you have also run into it with IFORT v10).
- I ran a 1-day simulation, using Bob's v8-01-03 standard run release, with no change at all. Using Intel Fortran v10.1.015, I was able to replicate Bob's standard run. However, when I switched to Intel Fortran v11.0.074, I ran into the error in partition.f, due to the CONCNOX-SUM1 < 0d0 check. Here's the error message in log:
=============================== GEOS-CHEM ERROR: STOP 30000 STOP at partition.f ===============================
- I then tried Bob's fix to partition.f. This time the run finishes, warning the user about the CONCNOX-SUM1 < 0d0 issue. But the output result is completely wacky!!! Below you can compare the surface Ox concentrations, using
- The (B) spatial pattern is completely off. NOx is also affected and shows the similar weird pattern.
- I'm pretty sure the problem is in the chemistry part. I've tried turning off the optimization but the problem persists. Perhaps there is some problem with the way IFORTv11 treats floating points? Also, I am not sure if IFORTv11 caused the weird model result, or if IFORTv11 caused some issues in chemistry, and the partition.f 'fix' subsequently lead to the weird result.
- Long story short, it seems like IFORTv11 is not a good choice for now, and that the 'fix' to partition.f should not be implemented.
Philippe Le Sager wrote:
- Thanks for testing Ifort11. We did run into the partition bug with Ifort10 after fixing tpcore. So I doubt that the weird result is related to that partition fix, and it is probably just a problem with IFORT 11.
Bob Yantosca wrote:
- You might have to go thru the IFORT 11 manuals to see if any default behavior has changed (i.e. optimization, compiler options, etc). It may not just be the concnox thing but something else in the numerics that is particular to IFORT 11.
- There is usually a "What's new" document w/ every Intel compiler release. Maybe that has some more information, you could look at it.
Bob Yantosca wrote:
- I've also heard from some folks @ NASA that IFORT 11.0 was problematic. They claim that IFORT 11.1 is much better. You may want to look into this in the meantime.
--Bob Y. 16:50, 7 October 2009 (EDT)
Eric Sofen wrote:
- Both Becky Alexander and I have run into problems with IFORT 11.1. When either of us run offline aerosol simulations compiled on IFORT 11.1, the simulation compiles and runs without errors, but the sulfur budgets are way off. The problems seem to be occurring in the deposition code, as Becky's simulations end up with very little deposition, but at the same time, the S burdens are too low. In my case, the deposition ends up being an order of magnitude too high. Changing back to IFORT 10 fixed both of these problems.
--Eric Sofen 13:32, 22 October 2009
Yuxuan Wang wrote:
- From our interaction with the Intel people, ifort 11.1.056 should work for GEOS-Chem. The GC version we tested at Tsinghua is v8-02-01 (nested-grid China with GEOS-5 meteorology). The platform we tested is Nehalem from Intel, with the following compilation options:
-cpp -w -static -fno-alias -O2 -safe_cray_ptr -no-prec-sqrt -no-prec-div -auto -noalign -convert big_endian
- Not sure whether these options will work for Mac OSX. From the testing, we found that codes compiled with ifort 11.1.056 ran at 2% faster than ifort 10.1.008.
--Bob Y. 14:59, 4 November 2009 (EST)
Problem with IFORT 11 and GEOS-Chem adjoint
Nicolas Bousserez wrote:
- We have been struggling for some time with the following problem when running GC adjoint (v8-02-01):
"OMP abort: Initializing libguide.so,
- but found libguide.so already initialized".
- After some investigations it seems like it is a linker error generated when different parts of the program try to link both static and dynamic verions of the OpenMP runtime. There is an option in ifort 11 to have openmp linked statically, which theoretically should fix this problem.
- But using ifort 11 for GC seems to cause other problems and this compilation option doesn't exist with ifort 10. The fact is that Daven Henze, who is using ifort 10 and a linux platform similar to ours never got the above problem. Has anyone got this error before? My platform configuration is the following:
Linux node9 2.6.9-89.0.23.ELsmp #1 SMP Wed Mar 17 06:49:21 EDT 2010 x86_64 x86_64 x86_64 GNU/Linux
- If anyone running GC adjoint has a similar configuration please let me know what is your Makefile (using netcdf libraries) configuration and which version of ifort you're using so that I can do some testing.
--Bob Y. 09:43, 8 April 2011 (EDT)
Incompatibility between IFORT 11 and OS version
If you are using a the Intel Fortran Compiler version 11, you may encounter some incompatibilities with your operating system, which might require an OS upgrade.
Nicolas Bousserez wrote:
- We have been struggling for some time with the following problem when running GC adjoint (v8-02-01). We get this error:
"OMP abort: Initializing libguide.so, but we but found libguide.so already initialized".
- After some investigations it seems like it is a linker error generated when different parts of the program try to link both static and dynamic verions of the OpenMP runtime. There is an option in ifort 11 to have openmp linked statically, which theoretically should fix this problem. But using ifort 11 for GC seems to cause other problems and this compilation option doesn't exist with ifort 10. The fact is that Daven Henze, who is using ifort 10 and a linux platform similar to ours never got the above problem. Has anyone got this error before? My platform configuration is the following:
Linux node9 2.6.9-89.0.23.ELsmp #1 SMP Wed Mar 17 06:49:21 EDT 2010 x86_64 x86_64 x86_64 GNU/Linux
Nicolas Bousserez wrote:
- For what it's worth, this is the oldest OS we're using:
Linux terra-01.vpn.as.harvard.edu 2.6.18-194.3.1.el5_lustre.1.8.4 #1 SMP Fri Jul 9 21:55:24 MDT 2010 x86_64 x86_64 x86_64 GNU/Linux
- and this is the newest:
Linux kvm-12.s.as.harvard.edu 2.6.18-194.32.1.el5 #1 SMP Wed Jan 5 17:52:25 EST 2011 x86_64 x86_64 x86_64 GNU/Linux
- What is relevant is the 2.6.9 and the 2.6.18, not the compilation dates. It means you're running something equivalent to a RHEL4/CentOS-4 kernel instead of RHEL5/CentOS-5, which has implications for your libraries, compatibility, bugs, security, etc. I would guess that you've been updating RHEL4 (first released 2005) or equivalent for several years, and ifort11 was released during the era of RHEL5 (first released 2007), so it wouldn't be too surprising if there were a library incompatibility. I don't know whether that is the cause of your symptom, but it might be. (RHEL6 was released late last year and v12 of the Intel compilers have also been released. CentOS-6 will be out soon.)
- You can continue to use the older compiler with the older OS, but I'd recommend upgrading the OS, which is worth doing anyway.
--Bob Y. 10:25, 13 April 2011 (EDT)
Error in partition.f when compiling with IFORT 10
Prasad Kasibhatla reported an error in routine partition.f which caused a GEOS-Chem v9-01-01 execution to halt. Upon further investigation, we found that this error only occurs when compiling GEOS-Chem with the Intel Fortran Compiler version 10 (aka IFORT 10) when selecting -O2 optimization and -openmp parallelization.
We recommend that all GEOS-Chem users upgrade to Intel Fortran Compiler version 11 (aka IFORT 11). If you must use IFORT 10, then we recommend that you compile the entire code with the -O1 optimization option. For more information, please see this wiki post.
--Bob Y. 09:17, 29 June 2011 (EDT)
KPP not compatibile with IFORT 9.1
Please see this wiki post about how problems compiling the KPP solver with IFORT 9.1.
Speedup With Hyperthreading on Nehalem chips
Hyperthreading is when a job uses more threads than there are actual CPU cores. I've noticed that using 16 threads ($OMP_NUM_THREADS = 16) on an 8-core system (2 x quad core Intel Nehalem X5570's) leads to a 15% speedup over using 8 threads. These tests were with GEOS-Chem v8-02-03, full chemistry, 2x2.5, ifort 10.1.021, and
FFLAGS = -cpp -w -O3 -auto -noalign -convert big_endian -g -traceback -CB -vec-report0.
This does not have a positive impact when using earlier generations of Intel chips (Harpertown or Clovertown).
--Daven Henze 1:42, 16 December 2009 (MDT)
Performance bottleneck caused by inefficient subroutine calls
Special care has to be taken when passing pointer arrays or sub-fields of dervied type objects to subroutines. If this is done incorrectly, it can cause a huge performance slowdown. Please see the discussion on our Passing array arguments efficiently in GEOS-Chem wiki page for full details.
--Bob Y. 10:49, 10 June 2013 (EDT)
Bugs in the IFORT compiler cause HEMCO to segfault
The GEOS-Chem Support Team has recently determined that bugs in the Intel Fortran Compiler versions 14 and 15 have caused the Harvard-NASA Emissions Component (aka HEMCO) to halt with segmentation faults. For more information, please see these wiki posts: