Configuring incomplete, errors occurred

[TianYangMY]

Name and Institution (Required)

Name: lish Institution: SUN YAT-SEN UNIVERSITY

Description of your issue or question

When Configuring with cmake, the following error occurs.

CMake Error at ESMA_cmake/compiler/flags/Intel_Fortran.cmake:59 (message): Unknown processor. Contact Matt Thompson Call Stack (most recent call first): ESMA_cmake/compiler/esma_compiler.cmake:9 (include) ESMA_cmake/esma.cmake:48 (include) CMakeLists.txt:55 (include)

-- Configuring incomplete, errors occurred!

Please provide as much detail as possible. Always include the GCHP version number and any relevant configuration and log files. GCHP v14.2.3 module load compiler/intel/2021.3.0 module load mpi/intelmpi/2021.3.0 module load mathlib/hdf5/1.12.0-intel-2021 module load mathlib/netcdf/4.6.1-intel-2021 module load compiler/cmake/3.20.4 export CC=icc export CXX=icpc export FC=ifort esmf 8.4.1

[lizziel]

Hi @TianYangMY, I am pinging Matt Thompson on this issue since the error message indicates contacting him. He is a developer of the ESMA_cmake software used within GCHP.

@mathomp4, this error is occurring using ESMA_cmake 3.8.0. We have not had any reports of this before so it must be specific to the configuration of the system @TianYangMY is using. Do you have any ideas what could be wrong? The error is occuring here.

cmake_host_system_information(RESULT proc_decription QUERY PROCESSOR_DESCRIPTION)
if (${proc_decription} MATCHES "EPYC")
   set (COREAVX2_FLAG "-march=core-avx2")
elseif (${proc_decription} MATCHES "Intel")
   set (COREAVX2_FLAG "-march=core-avx2")
   # Previous versions of GEOS used this flag, which was not portable
   # for AMD. Keeping here for a few versions for historical purposes.
   #set (COREAVX2_FLAG "-xCORE-AVX2")
else ()
   message(FATAL_ERROR "Unknown processor. Contact Matt Thompson")
endif ()

[mathomp4]

@TianYangMY What type of machine are you building on? You must be on something new!

My first thought would be to edit ESMA_cmake/compiler/flags/Intel_Fortran.cmake and add a line after the cmake_host_system_information call to display what CMake thinks your system is:

cmake_host_system_information(RESULT proc_decription QUERY PROCESSOR_DESCRIPTION)
message(STATUS "proc_description: ${proc_description}")

[TianYangMY]

@mathomp4 I used GCHP on the High-performance computing sytem, the processor is X86.

I add a line after the cmake_host_system_information:

cmake_host_system_information(RESULT proc_decription QUERY PROCESSOR_DESCRIPTION)
message(STATUS "proc_description: ${proc_description}")

After running CMake again, still the same error, and CMake doesn't seem to be matching to my system.New information is：

proc_description: 
CMake Error at ESMA_cmake/compiler/flags/Intel_Fortran.cmake:60 (message):
  Unknown processor.  Contact Matt Thompson
Call Stack (most recent call first):
  ESMA_cmake/compiler/esma_compiler.cmake:9 (include)
  ESMA_cmake/esma.cmake:48 (include)
  CMakeLists.txt:55 (include)

[mathomp4]

Well that is baffling. I have never seen CMake not return something there. Can you try adding this code:

## Print out the processor description
cmake_host_system_information(RESULT proc_description QUERY PROCESSOR_DESCRIPTION)
message(STATUS "Processor description: ${proc_description}")
## Print out the processor name
cmake_host_system_information(RESULT proc_name QUERY PROCESSOR_NAME)
message(STATUS "Processor name: ${proc_name}")
## Print out the processor serial number if HAS_SERIAL_NUMBER is true else print
## out that the processor has no serial number
cmake_host_system_information(RESULT has_serial QUERY HAS_SERIAL_NUMBER)
if (has_serial)
  cmake_host_system_information(RESULT proc_serial QUERY PROCESSOR_SERIAL_NUMBER)
  message(STATUS "Processor serial number: ${proc_serial}")
else ()
  message(STATUS "Processor has no serial number")
endif ()
## Print out CMAKE_HOST_SYSTEM_PROCESSOR
message(STATUS "CMAKE_HOST_SYSTEM_PROCESSOR: ${CMAKE_HOST_SYSTEM_PROCESSOR}")

and see what CMake spits out?

Can you also send the output of:

awk '/^$/ { exit } { print }' /proc/cpuinfo

which should leverage what the system knows.

Beyond that, I suppose a workaround for you would be to put:

   set (COREAVX2_FLAG "-march=core-avx2")

instead of the message(FATAL_ERROR) in:

else ()
   message(FATAL_ERROR "Unknown processor. Contact Matt Thompson")
endif ()

Most modern AMD and Intel chips support core-avx2. (Though it's hard for me to assure that without knowing what the processor is.)

[TianYangMY]

Thank you! @mathomp4
I add the code:

## Print out the processor description
cmake_host_system_information(RESULT proc_description QUERY PROCESSOR_DESCRIPTION)
message(STATUS "Processor description: ${proc_description}")
## Print out the processor name
cmake_host_system_information(RESULT proc_name QUERY PROCESSOR_NAME)
message(STATUS "Processor name: ${proc_name}")
## Print out the processor serial number if HAS_SERIAL_NUMBER is true else print
## out that the processor has no serial number
cmake_host_system_information(RESULT has_serial QUERY HAS_SERIAL_NUMBER)
if (has_serial)
  cmake_host_system_information(RESULT proc_serial QUERY PROCESSOR_SERIAL_NUMBER)
  message(STATUS "Processor serial number: ${proc_serial}")
else ()
  message(STATUS "Processor has no serial number")
endif ()
## Print out CMAKE_HOST_SYSTEM_PROCESSOR
message(STATUS "CMAKE_HOST_SYSTEM_PROCESSOR: ${CMAKE_HOST_SYSTEM_PROCESSOR}")

The output is :

-- Processor description: 64 core Hygon C86 7285 32-core Processor
-- Processor name: Unknown Hygon family
-- Processor has no serial number
-- CMAKE_HOST_SYSTEM_PROCESSOR: x86_64
-- proc_description: 64 core Hygon C86 7285 32-core Processor
CMake Error at ESMA_cmake/compiler/flags/Intel_Fortran.cmake:77 (message):
  Unknown processor.  Contact Matt Thompson
Call Stack (most recent call first):
  ESMA_cmake/compiler/esma_compiler.cmake:9 (include)
  ESMA_cmake/esma.cmake:48 (include)
  CMakeLists.txt:55 (include)

The output of this code awk '/^$/ { exit } { print }' /proc/cpuinfo is :

processor       : 0
vendor_id       : HygonGenuine
cpu family      : 24
model           : 1
model name      : Hygon C86 7285 32-core Processor
stepping        : 1
microcode       : 0x80901047
cpu MHz         : 2000.000
cache size      : 512 KB
physical id     : 0
siblings        : 32
core id         : 0
cpu cores       : 32
apicid          : 0
initial apicid  : 0
fpu             : yes
fpu_exception   : yes
cpuid level     : 13
wp              : yes
flags           : fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm constant_tsc art rep_good nopl nonstop_tsc extd_apicid amd_dcm aperfmperf eagerfpu pni pclmulqdq monitor ssse3 fma cx16 sse4_1 sse4_2 movbe popcnt aes xsave avx f16c rdrand lahf_lm cmp_legacy svm extapic cr8_legacy abm sse4a misalignsse 3dnowprefetch osvw skinit wdt tce topoext perfctr_core perfctr_nb bpext perfctr_l2 hw_pstate sme retpoline_amd ssbd ibpb vmmcall fsgsbase bmi1 avx2 smep bmi2 rdseed adx smap clflushopt sha_ni xsaveopt xsavec xgetbv1 clzero irperf xsaveerptr arat npt lbrv svm_lock nrip_save tsc_scale vmcb_clean flushbyasid decodeassists pausefilter pfthreshold avic v_vmsave_vmload vgif overflow_recov succor smca
bogomips        : 3999.44
TLB size        : 2560 4K pages
clflush size    : 64
cache_alignment : 64
address sizes   : 43 bits physical, 48 bits virtual
power management: ts ttp tm hwpstate eff_freq_ro [13] [14]

I changed the code:

 cmake_host_system_information(RESULT proc_decription QUERY PROCESSOR_DESCRIPTION)
message(STATUS "proc_description: ${proc_description}")
if (${proc_decription} MATCHES "EPYC")
   set (COREAVX2_FLAG "-march=core-avx2")
elseif (${proc_decription} MATCHES "Intel")
   set (COREAVX2_FLAG "-march=core-avx2")
   # Previous versions of GEOS used this flag, which was not portable
   # for AMD. Keeping here for a few versions for historical purposes.
   #set (COREAVX2_FLAG "-xCORE-AVX2")
else ()
    set (COREAVX2_FLAG "-march=core-avx2")
#   message(FATAL_ERROR "Unknown processor. Contact Matt Thompson")
endif ()

The configuration was successful !

But We I compiled GCHP, there are a lot of Warnings such as:

 /work/home/lish325/GCHP_14/GCHP/src/FMS/mpp/mpp_efp.F90(764): remark #8291: Recommended relationship between field width 'W' and the number of fractional digits 'D' in this edit descriptor is 'W>=D+7'.
      write (mesg,'("mpp_efp_list_sum_across_PEs error at ",i6," val was ",ES12.6, ", prec_error = ",ES12.6)') &
-----------------------------------------------------------------------------^
/work/home/lish325/GCHP_14/GCHP/src/FMS/mpp/mpp_efp.F90(764): remark #8291: Recommended relationship between field width 'W' and the number of fractional digits 'D' in this edit descriptor is 'W>=D+7'.
      write (mesg,'("mpp_efp_list_sum_across_PEs error at ",i6," val was ",ES12.6, ", prec_error = ",ES12.6)') &

ifort: warning #10182: disabling optimization; runtime debug checks enabled
ifort: warning #10182: disabling optimization; runtime debug checks enabled
ifort: warning #10182: disabling optimization; runtime debug checks enabled
ifort: warning #10182: disabling optimization; runtime debug checks enabled
ifort: warning #10182: disabling optimization; runtime debug checks enabled

[ 64%] Building Fortran object src/fms_r8/CMakeFiles/fms_r8.dir/tracer_manager/tracer_manager.F90.o
[ 64%] Building Fortran object src/fms_r8/CMakeFiles/fms_r8.dir/field_manager/fm_util.F90.o
[ 64%] Building Fortran object src/fms_r8/CMakeFiles/fms_r8.dir/oda_tools/oda_core.F90.o
[ 64%] Building Fortran object src/fms_r8/CMakeFiles/fms_r8.dir/diag_manager/diag_output.F90.o
/work/home/lish325/GCHP_14/GCHP/src/FMS/tracer_manager/tracer_manager.F90(912): remark #8291: Recommended relationship between field width 'W' and the number of fractional digits 'D' in this edit descriptor is 'W>=D+7'.
901 FORMAT(E12.6,1x,E12.6)
------------^
/work/home/lish325/GCHP_14/GCHP/src/FMS/tracer_manager/tracer_manager.F90(912): remark #8291: Recommended relationship between field width 'W' and the number of fractional digits 'D' in this edit descriptor is 'W>=D+7'.
901 FORMAT(E12.6,1x,E12.6)
---------------------^
/work/home/lish325/GCHP_14/GCHP/src/FMS/tracer_manager/tracer_manager.F90(911): remark #8291: Recommended relationship between field width 'W' and the number of fractional digits 'D' in this edit descriptor is 'W>=D+7'.
900 FORMAT(A,2(1x,E12.6))
-------------------^
/work/home/lish325/GCHP_14/GCHP/src/FMS/tracer_manager/tracer_manager.F90(1278): remark #8291: Recommended relationship between field width 'W' and the number of fractional digits 'D' in this edit descriptor is 'W>=D+7'.
  700 FORMAT (3A,E12.6,A,F10.6)

@lizziel I don't know if these warnings affect the proper functioning of GCHP, but in the end the GCHP installed successfully!

[100%] Built target GCHP
[100%] Built target GCHP_GridComp
[100%] Built target gchp
Install the project...
-- Install configuration: "RelWithDebInfo"

[mathomp4]

Ahhh. A Hygon processor! I know of Hygon but I've never encountered one before!

From your cpuinfo output it should be avx2 compatible, so the flag you used should work. I'll add it to ESMA_cmake as a supported processor for Intel.

ETA: Well, I'll add support once you can confirm things run, @TianYangMY. I'm fairly confident it will, but better to be sure. 😄

[lizziel]

Hurray! @TianYangMY, many warnings during build is normal for GCHP. There is actually a separate issue open about trying to suppress them: https://github.com/geoschem/GCHP/issues/322. For now you can just ignore them.

[TianYangMY]

Thank you ！@mathomp4 @lizziel I tried to run GCHP, but encountered a strange problem! I used the following script to submit GCHP to run:

#!/bin/bash
#
#SBATCH -n 96
#SBATCH -N 2
#SBATCH -t 2:00:00
#SBATCH -p xahcnormal
#SBATCH --mem=80G
#SBATCH -o out.%j
#SBATCH -e err.%j

source gchp.env

#################################################################
#
# ADDITIONAL PRE-RUN CONFIGURATION
#
# If a subsequent command fails, treat it as fatal (don't continue)
set -e

# For remainder of script, echo commands to the job's log file
set -x

# Unlimit resources to prevent OS killing GCHP due to resource usage/
# Alternatively you can put this in your environment file.
ulimit -c 0                  # coredumpsize
ulimit -l unlimited          # memorylocked
ulimit -u 50000              # maxproc
ulimit -v unlimited          # vmemoryuse
ulimit -s unlimited          # stacksize
module list

# Define log name to include simulation start date
start_str=$(sed 's/ /_/g' cap_restart)
log=gchp.${start_str:0:13}z.log

# Update config files, set restart symlink, and do sanity checks
source setCommonRunSettings.sh
source setRestartLink.sh
source checkRunSettings.sh

# srun -n 96 -N 2 -m plane=24 --mpi=pmix ./gchp > ${log}
mpirun -np 96 ./gchp > ${log}

#################################################################
#
# POST-RUN COMMANDS
#

# Rename mid-run checkpoint files, if any. Discard file if time corresponds
# to run start time since duplicate with initial restart file.
chkpnts=$(ls Restarts)
for chkpnt in ${chkpnts}
do
    if [[ "$chkpnt" == *"gcchem_internal_checkpoint."* ]]; then
       chkpnt_time=${chkpnt:27:13}
       if [[ "${chkpnt_time}" = "${start_str:0:13}" ]]; then
          rm ./Restarts/${chkpnt}
       else
          new_chkpnt=./Restarts/GEOSChem.Restart.${chkpnt_time}z.c${N}.nc4
          mv ./Restarts/${chkpnt} ${new_chkpnt}
       fi
    fi
done
# If new start time in cap_restart is okay, rename restart file
# and update restart symlink
new_start_str=$(sed 's/ /_/g' cap_restart)
if [[ "${new_start_str}" = "${start_str}" || "${new_start_str}" = "" ]]; then
    echo "ERROR: GCHP failed to run to completion. Check the log file for more information."
    rm -f Restarts/gcchem_internal_checkpoint
    exit 1
else
    N=$(grep "CS_RES=" setCommonRunSettings.sh | cut -c 8- | xargs )
    mv Restarts/gcchem_internal_checkpoint Restarts/GEOSChem.Restart.${new_start_str:0:13}z.c${N}.nc4
    source setRestartLink.sh
fi

The gchp.log shows that it just started running normally, but I noticed that the gchp.log output:

     object: 331,name: STATE_PSC
            type: Field
     object: 332,name: T_DAVG
            type: Field
     object: 333,name: T_PREVDAY
            type: Field
     object: 334,name: TropLev
            type: Field
     object: 335,name: WetDepNitrogen
            type: Field

After this gchp.log no longer outputs, but no errors are reported and the tasks I submitted keep running. This seems to indicate that GCHP has stopped running but is still hogging the CPU, what could this be?

allPEs.txt gchp.20190101_0000z.txt logfile.000000.out.txt

[lizziel]

This looks like an issue with the restart file. What simulation are you running (e.g. standard versus benchmark) and what version of the model are you using?

[lizziel]

Also, please make a new issue for this since it is a new subject. I will close out this issue after we move there.