Hello Neks, I ran an svn update this morning (rev 510). Since then I can't compile prex/prenek anymore. I tried with and without the big memory support enabled. Here is what I get in both cases : [17:46:58 michael@dsktp-lmfa:~/Codes/nek5_svn/trunk/tools ] $ ./maketools prenek ---------------------- Make prenek... ---------------------- [: 23: prenek: unexpected operator [: 23: prenek: unexpected operator make[1]: entrant dans le répertoire « /home/michael/Codes/nek5_svn/trunk/tools/prenek » gfortran -mcmodel=medium -o /home/michael/bin/prex prenek.o curve.o edit.o build.o build1.o build2.o bound.o plot.o xinterface.o glomod.o legend.o vprops.o iolib.o subs.o zipper2.o postnek6.o screen.o revert.o crs.o mxm.o xdriver.o -L/usr/X11R6/lib -lX11 -lm prenek.o: In function `fprep_': prenek.f:(.text+0x111): relocation truncated to fit: R_X86_64_PC32 against symbol `devs_' defined in COMMON section in prenek.o prenek.f:(.text+0x32b): relocation truncated to fit: R_X86_64_PC32 against symbol `devs_' defined in COMMON section in prenek.o prenek.o: In function `read_params_': prenek.f:(.text+0x625): relocation truncated to fit: R_X86_64_PC32 against symbol `devs_' defined in COMMON section in prenek.o prenek.f:(.text+0x641): relocation truncated to fit: R_X86_64_PC32 against symbol `devs_' defined in COMMON section in prenek.o prenek.f:(.text+0x71b): relocation truncated to fit: R_X86_64_32S against symbol `param_' defined in COMMON section in prenek.o prenek.f:(.text+0x7d6): relocation truncated to fit: R_X86_64_32S against symbol `param_' defined in COMMON section in prenek.o prenek.f:(.text+0x859): relocation truncated to fit: R_X86_64_32S against symbol `param_' defined in COMMON section in prenek.o prenek.f:(.text+0x8dc): relocation truncated to fit: R_X86_64_32S against symbol `param_' defined in COMMON section in prenek.o prenek.f:(.text+0x966): relocation truncated to fit: R_X86_64_32S against symbol `param_' defined in COMMON section in prenek.o prenek.f:(.text+0x96e): relocation truncated to fit: R_X86_64_PC32 against symbol `param_' defined in COMMON section in prenek.o prenek.f:(.text+0x97b): additional relocation overflows omitted from the output collect2: ld returned 1 exit status make[1]: *** [prex] Erreur 1 make[1]: quittant le répertoire « /home/michael/Codes/nek5_svn/trunk/tools/prenek » make: *** [all] Erreur 1 [17:47:51 michael@dsktp-lmfa:~/Codes/nek5_svn/trunk/tools ] $ Do you have any idea on what is happening there? Is it compulsory to switch to the Intel ifort compiler for the new version of prenek? Thanks in advance for any help. Best, Michael B ________________________________________ De : [email protected] [[email protected]] de la part de [email protected] [[email protected]] Date d'envoi : lundi 3 mai 2010 20:14 À : [email protected] Objet : Re: [Nek5000-users] prex & pretex not working Hi Stefan, The command does not do anything. Here is the output of the command. It does not display anything. [kanch1@honest3 tools]$ ./maketools prenek [kanch1@honest3 tools]$ Thanks, Harish. On Mon, 3 May 2010 [email protected] wrote:
Please send the output of './maketools prenek'
Stefan
On May 3, 2010, at 7:10 PM, [email protected] wrote:
Hi Stefan,
Sorry for missing out on that problem report.
But I tried changing the F77 compiler as below and it does not work. "maketools all" does nothing.
F77="ifort -shared-intel"
Also, this is on a different linux box not on my cluster which has GNU compiler.
Thanks,
Harish.
On Mon, 3 May 2010 [email protected] wrote:
I know. There was a problem report on the mailing list already.
This was my answer:
Ok I think I know a way to fix the Intel compiler problem.
To compile prenek (prex) you need to enable the BIGMEM support so you have to use at least the medium memory model. However the Intel compilers wants you to link all Intel-Libs dynamically if you use another memory model than small.
Just set F77= "ifort -shared-intel" and it should work!
hth, Stefan
On May 3, 2010, at 5:45 PM, [email protected] wrote:
Hi Stefan,
I tried compiling prenek on NCSA-Abe machine with ifort & icc. I get the following error, its same error w/ and w/o "BIGMEM" flag.
Thanks,
Harish.
************************************************************************* icc -c -DUNDERSCORE -Dr8 xdriver.c xdriver.c(875): warning #266: function "fprep_" declared implicitly fprep_ (); /* run Fortran Main code */ ^
xdriver.c(123): (col. 5) remark: LOOP WAS VECTORIZED. xdriver.c(125): (col. 5) remark: LOOP WAS VECTORIZED. xdriver.c(869): (col. 5) remark: LOOP WAS VECTORIZED. xdriver.c(869): (col. 5) remark: LOOP WAS VECTORIZED. xdriver.c(747): (col. 4) remark: LOOP WAS VECTORIZED. xdriver.c(823): (col. 4) remark: LOOP WAS VECTORIZED. xdriver.c(700): (col. 5) remark: LOOP WAS VECTORIZED. xdriver.c(333): (col. 5) remark: LOOP WAS VECTORIZED. xdriver.c(340): (col. 5) remark: LOOP WAS VECTORIZED. ifort -o /u/ac/kanch1/nek5_svn/trunk/tools/prex prenek.o curve.o edit.o build.o build1.o build2.o bound.o plot.o xinterface.o glomod.o legend.o vprops.o iolib.o subs.o zipper2.o postnek6.o screen.o revert.o crs.o mxm.o xdriver.o -L/usr/X11R6/lib64 -lX11 -lm prenek.o(.text+0x47d): In function `drivfc_': : relocation truncated to fit: R_X86_64_PC32 chaar_ prenek.o(.text+0x4f9): In function `drivfc_': : relocation truncated to fit: R_X86_64_PC32 chaar_ prenek.o(.text+0x50d): In function `drivfc_': : relocation truncated to fit: R_X86_64_PC32 basics_ prenek.o(.text+0x584): In function `drivfc_': : relocation truncated to fit: R_X86_64_PC32 devs_ prenek.o(.text+0x5b1): In function `drivfc_': : relocation truncated to fit: R_X86_64_PC32 chaar_ prenek.o(.text+0x62d): In function `drivfc_': : relocation truncated to fit: R_X86_64_PC32 chaar_ prenek.o(.text+0x667): In function `drivfc_': : relocation truncated to fit: R_X86_64_32S chaar_ prenek.o(.text+0x740): In function `drivfc_': : relocation truncated to fit: R_X86_64_PC32 modes_ prenek.o(.text+0x789): In function `drivfc_': : relocation truncated to fit: R_X86_64_PC32 chaar_ prenek.o(.text+0x7fc): In function `drivfc_': : relocation truncated to fit: R_X86_64_PC32 chaar_ prenek.o(.text+0x80c): In function `drivfc_': : additional relocation overflows omitted from the output make[1]: *** [prex] Error 1 make[1]: Leaving directory `/u/ac/kanch1/nek5_svn/trunk/tools/prenek' make: *** [all] Error 1 **************************************************************************
On Mon, May 3, 2010 9:38 am, [email protected] wrote:
Can you try on another Linux box? You can get the Intel compilers (non commercial usage) for free. Sorry I don't want to login into other machines unless it's really needed.
Stefan
On May 3, 2010, at 4:34 PM, [email protected] wrote:
Hi Stefan,
I do not have any other compiler on the cluster. I can give you my login to check the problem.
Thanks,
Harish.
On Mon, 3 May 2010 [email protected] wrote:
Hi Harish,
unfortunately I cannot reproduce the problem over there (gcc 4.2.1/ gfortran 4.5). Can you try again using another compiler?
Stefan
On May 3, 2010, at 3:26 PM, [email protected] wrote:
Hello,
I am compiling prenek with GNU compilers, gfortran-4.3.2 and gcc-4.3.2. When I compile it without "BIGMEM" flag it gives an error, the output of it is attached (error.out). With "BIGMEM" flag turned on, there are no compiler errors (compile.out). But when I try to open pretex or prex, I get the following error.
[harish@cmtl hole-in-center]$ prex Killed [harish@cmtl hole-in-center]$ pretex Killed
All the other tools are working fine. Any help in solving this issue is appreciated.
Thanks,
Harish. <error.out><compile.out>_______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
_______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
_______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
_______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
_______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
_______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
_______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
_______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users
_______________________________________________ Nek5000-users mailing list [email protected] https://lists.mcs.anl.gov/mailman/listinfo/nek5000-users