f951: sorry, unimplemented: Graphite loop.. cannot be used

Discussion on computers, ROMS installation and compiling

Moderators: arango, robertson

Post Reply
Message
Author
barack99
Posts: 17
Joined: Thu Aug 31, 2017 4:33 pm

f951: sorry, unimplemented: Graphite loop.. cannot be used

#1 Post by barack99 » Sun Sep 03, 2017 11:22 am

Dear All,

I am a new ROMS user, I got the error message below while trying to run test case: Sed_floc_toy_check. Could anyone offer some help?

Many thanks

Barack

f951: sorry, unimplemented: Graphite loop optimizations cannot be used (-fgraphite, -fgraphite-identity, -floop-block, -floop-interchange, -floop-strip-mine, -floop-parallelize-all, and -ftree-loop-linear)
make: *** [/data/COAWST/Projects/Sed_floc_toy_check/Build/mod_kinds.o] Error 1
Barack

User avatar
kate
Posts: 3718
Joined: Wed Jul 02, 2003 5:29 pm
Location: IMS/UAF, USA

Re: f951: sorry, unimplemented: Graphite loop.. cannot be us

#2 Post by kate » Sun Sep 03, 2017 10:13 pm

Which compiler are you using? Those look like gnu fortran options.

barack99
Posts: 17
Joined: Thu Aug 31, 2017 4:33 pm

Re: f951: sorry, unimplemented: Graphite loop.. cannot be us

#3 Post by barack99 » Mon Sep 04, 2017 1:33 am

Hi Kate,

That error occured with gfortran. When I tried another test (Sedbed_toy) with Ifort/openMPI, I got another error message below.

Could you help?

Many Thanks & Regards
Barack

cd /data/projects/COAWST/Projects/Sedbed_toy_check/Build; /usr/local/software/OpenMPI/2.0.2-intel-2016.u3/bin/mpif90 -c -heap-arrays -fp-model precise -ip -O2 u2dbc_im.f90
u2dbc_im.f90(171): error #6404: This name does not have a type, and must have an explicit type. [FIRST_2D_STEP]
IF (FIRST_2D_STEP) THEN
----------^
u2dbc_im.f90(171): error #6341: A logical data type is required in this context. [FIRST_2D_STEP]
IF (FIRST_2D_STEP) THEN
----------^
compilation aborted for u2dbc_im.f90 (code 1)
make: *** [/data/projects/COAWST/Projects/Sedbed_toy_check/Build/u2dbc_im.o] Error 1
Barack

User avatar
kate
Posts: 3718
Joined: Wed Jul 02, 2003 5:29 pm
Location: IMS/UAF, USA

Re: f951: sorry, unimplemented: Graphite loop.. cannot be us

#4 Post by kate » Mon Sep 04, 2017 2:33 am

I don't know where you are getting those extra compile flags from. My roms wants to compile with gfortran like this:

Code: Select all

cd Build; /usr/bin/gfortran -c -frepack-arrays -O3 -ffast-math strings.f90
Did you do a "make clean" before trying with a different compiler? FIRST_2D_STEP is coming from globaldefs.h and should not be seen by the compiler. Is there some other error above that from the cpp phase?

barack99
Posts: 17
Joined: Thu Aug 31, 2017 4:33 pm

Re: f951: sorry, unimplemented: Graphite loop.. cannot be us

#5 Post by barack99 » Mon Sep 04, 2017 3:28 pm

I did "make clean", but I dont know where the error came from?

Do you have any further suggestion?

Thanks
Barack

User avatar
kate
Posts: 3718
Joined: Wed Jul 02, 2003 5:29 pm
Location: IMS/UAF, USA

Re: f951: sorry, unimplemented: Graphite loop.. cannot be us

#6 Post by kate » Tue Sep 05, 2017 4:22 pm

I would go back to gfortran and try to figure out where those extra options came from. Do you have an FFLAGS environment variable set?

barack99
Posts: 17
Joined: Thu Aug 31, 2017 4:33 pm

Re: f951: sorry, unimplemented: Graphite loop.. cannot be us

#7 Post by barack99 » Thu Sep 07, 2017 4:41 am

Dear Kate,

I have: FFLAGS := -frepack-arrays

I am able to compile MCT/SCRIPT_COAWST, and SWAN alone, but when compiling: Projects\Inlet_test\Coupled... I got another error below.

Where could I find m_coupling.mod? or m_coupling.f90/ftn?

Thanks

swan_iounits.f90:14:10:

USE M_COUPLING
1
Fatal Error: Can't open module file ‘m_coupling.mod’ for reading at (1): No such file or directory
compilation terminated.
Barack

User avatar
kate
Posts: 3718
Joined: Wed Jul 02, 2003 5:29 pm
Location: IMS/UAF, USA

Re: f951: sorry, unimplemented: Graphite loop.. cannot be us

#8 Post by kate » Thu Sep 07, 2017 4:44 pm

To find things, I like to use grep. In this case you can use the -ir option for recursive, case insensitive:

Code: Select all

grep -ir m_coupling .
In addition to finding the "USE" lines, you get:

Code: Select all

./SWAN/Src/swmod2.ftn:      MODULE M_COUPLING
./SWAN/Src/swmod2.ftn:      END MODULE M_COUPLING
It also matches instances of ocn2atm_coupling, oh well.

How to get things to compile in the correct order is another issue entirely, something I don't know how SWAN is attempting.

barack99
Posts: 17
Joined: Thu Aug 31, 2017 4:33 pm

Re: f951: sorry, unimplemented: Graphite loop.. cannot be us

#9 Post by barack99 » Fri Sep 08, 2017 12:02 am

All right! :D

After changing gfortran/MPI versions, and re-building netCDF libraries, I have successfully compiled my COAWST! I am able to run all testcases in 'Projects' without any error!

My current versions are: GCC 5.4.0, OpenMPI 2.0.2, Zlib1.2.11/HDF51.10.1, netcdf-C 4.4.1.1 and netCDF-Fortran 4.4.0.

Thanks & Regards
Barrack
Barack

okami_zxh

Re: f951: sorry, unimplemented: Graphite loop.. cannot be us

#10 Post by okami_zxh » Tue Sep 12, 2017 3:23 pm

barack99 wrote:All right! :D

After changing gfortran/MPI versions, and re-building netCDF libraries, I have successfully compiled my COAWST! I am able to run all testcases in 'Projects' without any error!

My current versions are: GCC 5.4.0, OpenMPI 2.0.2, Zlib1.2.11/HDF51.10.1, netcdf-C 4.4.1.1 and netCDF-Fortran 4.4.0.

Thanks & Regards
Barrack
Hi Barrack,
Which version of gfortran you are using for COAWST?
Xiaohui

barack99
Posts: 17
Joined: Thu Aug 31, 2017 4:33 pm

Re: f951: sorry, unimplemented: Graphite loop.. cannot be us

#11 Post by barack99 » Fri Sep 22, 2017 10:23 pm

As I mentioned GCC version is 5.4, so that is gfortran.
Barack

SivaHeramb
Posts: 2
Joined: Mon Aug 05, 2019 1:39 pm
Location: Indian Institute of Technology Delhi

Re: f951: sorry, unimplemented: Graphite loop.. cannot be us

#12 Post by SivaHeramb » Thu Sep 12, 2019 3:32 pm

Barrack[/quote]
barack99 wrote:
cd /data/projects/COAWST/Projects/Sedbed_toy_check/Build; /usr/local/software/OpenMPI/2.0.2-intel-2016.u3/bin/mpif90 -c -heap-arrays -fp-model precise -ip -O2 u2dbc_im.f90
u2dbc_im.f90(171): error #6404: This name does not have a type, and must have an explicit type. [FIRST_2D_STEP]
IF (FIRST_2D_STEP) THEN
----------^
u2dbc_im.f90(171): error #6341: A logical data type is required in this context. [FIRST_2D_STEP]
IF (FIRST_2D_STEP) THEN
----------^
compilation aborted for u2dbc_im.f90 (code 1)
make: *** [/data/projects/COAWST/Projects/Sedbed_toy_check/Build/u2dbc_im.o] Error 1
I got the same error (with the same file and also because of FIRST_2D_STEP lines) but when using gfortran for compiling the basic upwelling case.
Is it because of any compilers incompatibility?
My current versions are almost same as you mentioned.
Can you help me out with this error?

Thanks and Regards,
Siva Heramb

Post Reply