PC GAMESS/Firefly-related discussion club

Alex Granovsky
gran@classic.chem.msu.su

your mistake is that you did not assign working directories
for different instances of the parallel PC GAMESS/Firefly process.

The simplest way to do this is to use -t command line option.

As it is not documented very well in the manuals, I post here the details on how it works.


Syntax:

pcgamess [other options] -t < the tmp dirs prefix >

Examples:

Case 1 (formally correct but deadly):

 -t /tmp 

Each instance will try to create /tmp.process_rank directory for
its scratch files, i.e., /tmp.0, /tmp.1, etc... If you are not
running as root, they will fail resulting in different runtime
errors depending on the error code returned by OS and the
filesystem type.

Case 2 (correct use, note the difference with Case 1!):

 -t /tmp/ 

Each instance will try to create /tmp/pcg.process_rank
directory for its scratch files, i.e., /tmp/pcg.0. /tmp/pcg.1, etc...


Case 3 (correct use):

 -t /tmp/test 

Each instance will try to create /tmp/test.process_rank
directory for its scratch files.

Case 4 (correct use):

 -t /tmp/test/ 

Each instance will try to create /tmp/test/pcg.process_rank
directory for its scratch files.

etc..., etc...


In addition, I would recommend you to specify -r and -p command line
switches for most of the runs. Finally, the proper -ex option is
important for better performance. And remember that all options which
take path/filename as their argument (including -i and -o options)
expect fully qualified filename or path. This is especially important
because the PC GAMESS/Firefly' working directory is usually not the
directory it was launched from!

-Alex Granovsky









On Sun Nov 16 '08 9:09pm, Yongle Li wrote
-----------------------------------------
>Hi, I am a new user of PC-GAMESS/Firefly, when I tested the TDDFT options
> recently I received errors, so I need some help for fixing this problem.

>The computer I used is:
>CPU: Intel-Xeon,
>System: Red Hat  Enterprise Linux AS release 4
>Intel MPI 3.0, Intel MKL 9.0
>The input file is:

> $CONTRL SCFTYP=RHF RUNTYP=ENERGY DFTTYP=B3LYP CITYP=TDDFT $END
> $TDDFT MULT=1 NSTATE=4 $END
> $BASIS  GBASIS=n31 ngauss=6 NDFUNC=1 $END
> $DATA
>H2O
>CNV 2

> O ?????8.0 ?0.0000000000 ?0.0000000000 ?0.7205815395
> H ?????1.0 ?0.0000000000 ?0.7565140024 ?0.1397092302
> $END

>and when I subbed the job with:
>mpirun < /dev/null -n 4 ./pcgamess -i test3.inp -o test3.out /localhome/yongle/PC-GAMESS

>I received the error information:
>========================================
>TID 9329 caught signal 11, exiting.

>Dump of registers follows
>
>
>TID 9328 caught signal 11, exiting.

>Dump of registers follows

>eax :: 0x01de9d60, edx :: 0xf511d3c8

>TID 9330 caught signal 11, exiting.

>Dump of registers follows

>ecx :: 0x0000003e, ebx :: 0xf5103f78
>eax :: 0x1ff67478, esi :: 0xffffa89c, edi :: 0xf514bc40
>edx :: 0xf511d3c8
>ebp :: 0xffffa4b0, esp :: 0xffffa2b0
>eip :: 0x10571717, ecx :: 0x00000002, eflags :: 0x00010286

>ebx :: 0xf5103f78
>cs  :: 0x0023
>ds  :: 0x002b
>es  :: 0x002b
>esi :: 0xffffa89c, ss  :: 0x002b
>fs  :: 0x01cf
>edi :: 0xf514bc40
>gs  :: 0x005b

>Stack backtrace

>ebp :: 0xffffa4b0, esp :: 0xffffa4b0, ebp :: 0xffffa798, eip :: 0xf50f07e0
>esp :: 0xffffa2b0
>esp :: 0xffffa798, ebp :: 0x0b3a2b69, eip :: 0xc01e5453
>eip :: 0x10571717, eflags :: 0x00010286

>cs  :: 0x0023
>ds  :: 0x002b
>es  :: 0x002b
>eax :: 0x1ff67478, ss  :: 0x002b
>fs  :: 0x01cf
>gs  :: 0x005b

>Stack backtrace

>esp :: 0xffffa4b0, ebp :: 0xffffa798, eip :: 0xf50f07e0
>edx :: 0xf511d3c8
>esp :: 0xffffa798, ebp :: 0x0b3a2b69, eip :: 0xc01e5453
>ecx :: 0x00000002, ebx :: 0xf5103f78
>esi :: 0xffffa89c, edi :: 0xf514bc40
>ebp :: 0xffffa4b0, esp :: 0xffffa2b0
>eip :: 0x10571717, eflags :: 0x00010286

>cs  :: 0x0023
>ds  :: 0x002b
>es  :: 0x002b
>ss  :: 0x002b
>fs  :: 0x01cf
>gs  :: 0x005b

>Stack backtrace

>esp :: 0xffffa4b0, ebp :: 0xffffa798, eip :: 0xf50f07e0
>esp :: 0xffffa798, ebp :: 0x0b3a2b69, eip :: 0xc01e5453

>[yongle@cu02 PC-GAMESS]$ rank 2 in job 1  localhost_49714   caused collective abort of all ranks
>  exit status of rank 2: killed by signal 9
>=====================================================

>And in output file:

>          --------------------------
>          R-B3LYP    SCF CALCULATION
>          --------------------------

>     NUCLEAR ENERGY =         9.2267391005
>     MAXIT =   30     NPUNCH=    2
>     EXTRAP=T  DAMP=T  SHIFT=T  RSTRCT=F  DIIS=T  DEM=F  SOSCF=F
>     DENSITY CONV=  1.00E-05
>     MEMORY REQUIRED FOR SCF STEP=     85597 WORDS.

> XC FUNCTIONAL: SLATER + BECKE 88 + HF EXCHANGE, LYP 88 + VWN 1 RPA CORRELATION
>
>
> ITER EX DEM  TOTAL ENERGY      E CHANGE  DENSITY CHANGE    DIIS ERROR
>          * * *   INITIATING DIIS PROCEDURE   * * *
>   1  0  0   -97.011922222   -97.011922222  57.438108747  16.002497901

>

[ This message was edited on Sun Nov 16 '08 at 10:39pm by the author ]