[CP2K:7675] Re: CP2K compiling on CRAYXC40 - cuda version using gnu compiler

[DEC014]

Aman,

I have received very similar errors before on Cray machines (most recently 
a few Cray XC40  and XC30 machines).  Are you getting a Cholesky Error 
along the lines of "Cholesky decompose failed: the matrix is not positive 
definite or ill-conditioned" in your output files (not the error files)? 
 My problems occurred when I tried statically linking Intel MKL libraries 
for the BLAS, LAPACK, SCALAPACK, and BLACS libraries.  Even when I did not 
use the Cray cc wrapper to exclude any conflicts from Libsci_gnu_49 I was 
still getting the error.  

I see you are not using MKL libraries, but you might want to try a 
different version of LibSci Module.  Also, are you compiling in the Batch 
Nodes or Compute nodes?  I've had programs that misbehave when they aren't 
compiled inside the batch nodes.  To compile in a batch node, submit an 
interactive PBS job (qsub -I -A <Allocation> -q debug -l 
select=1:ncpus=24:mpiprocs=24 -l walltime=0:30:00 -N cp2k_build -j oe -o 
cp2k.oe) of course you'll have to change ncpus and mpiprocs to your Cray's 
processors.  When building inside Batch Nodes, please do not use more than 
4 cpus for make (make -j 4).  These Batch nodes are really the PBS MOM 
nodes that dole out the PBS tasks to the compute nodes and Admins can get 
angry when you overload them.  I have attached the Architecture file that 
worked for me for my Cray XC40 build, of course this is for POPT and you'd 
have to make the necessary changes to build a Hybrid Executable.

Regards,
David 

On Thursday, April 14, 2016 at 5:38:10 AM UTC-4, Aman Jindal wrote:
>
> Yeah I did that. Now, once more I am trying to compile plumed again and 
> then cp2k (without this flag).
> Hope it will work. 
> Thank you
>
> On Thu, Apr 14, 2016 at 2:53 PM, Alfio Lazzaro <alfi... at gmail.com> 
> wrote:
>
>> Interesting, there must something before the error you are mentioning. 
>> Have you make "clean" of CP2K before recompiling it with the flag?
>>
>> Alfio
>>
>> On Thursday, April 14, 2016 at 9:10:28 AM UTC+2, Aman Jindal wrote:
>>>
>>> Hello Alfio,
>>> Thanks your valuable response. I tried compiling cp2k after removing the 
>>> flag "-mavx" , but I am getting the following error:
>>> collect2: error: ld returned 1 exit status
>>>
>>> Plumed was compiled using the following commands:
>>> 1) ./configure CXX=CC CC=cc F77=ftn F90=ftn  --prefix=/path/for/exe 
>>> --enable-mpi
>>> 2) make
>>> 3) make install
>>>
>>> Aman
>>>
>>>
>>> On Wed, Apr 13, 2016 at 4:52 PM, Alfio Lazzaro <alfi... at gmail.com> 
>>> wrote:
>>>
>>>> Dear Aman,
>>>> in your arch file I see that you are using "-mavx" flag. Could you 
>>>> remove it and try to recompile? Your error message says that there is 
>>>> something wrong with the flags used for the compilation (I assume the 
>>>> vector instructions), and from your last email it is seems that there is 
>>>> something wrong with PLUMED. Do you know how it was compiled?
>>>>
>>>> Alfio
>>>>
>>>>
>>>> Il giorno mercoledì 13 aprile 2016 11:55:02 UTC+2, Aman Jindal ha 
>>>> scritto:
>>>>>
>>>>> Hello Samuel Andermatt,
>>>>> I found that this error is coming only if I am compiling CP2K(version 
>>>>> 3.0)  with plumed (version 2.2.1). Because I compiled cp2k with and without 
>>>>> plumed today using CRAY-XC30-gfortran.psmp (this time without cuda) arch 
>>>>> file. Now without plumed, it is working fine, but with plumed it is giving 
>>>>> the same error (mentioned earlier).
>>>>>
>>>>> Do you have any idea how can I resolve this issue..
>>>>>
>>>>> Thank you,
>>>>> Aman
>>>>>
>>>>> On Tue, Apr 12, 2016 at 6:36 PM, Aman Jindal <amanj... at gmail.com> 
>>>>> wrote:
>>>>>
>>>>>> Hi, I used CRAY-XC30-gfortran-cuda.psmp arch file and I 
>>>>>> have gcc-5.2.0 version. Attached is the modified arch file.
>>>>>> Thank you
>>>>>> On Tuesday, April 12, 2016 at 6:21:25 PM UTC+5:30, Samuel Andermatt 
>>>>>> wrote:
>>>>>>>
>>>>>>> Which arch file did you use and which compiler version do you have?
>>>>>>>
>>>>>> -- 
>>>>>> You received this message because you are subscribed to the Google 
>>>>>> Groups "cp2k" group.
>>>>>> To unsubscribe from this group and stop receiving emails from it, 
>>>>>> send an email to cp2k+... at googlegroups.com.
>>>>>> To post to this group, send email to cp... at googlegroups.com.
>>>>>> Visit this group at https://groups.google.com/group/cp2k.
>>>>>> For more options, visit https://groups.google.com/d/optout.
>>>>>>
>>>>>
>>>>> -- 
>>>> You received this message because you are subscribed to the Google 
>>>> Groups "cp2k" group.
>>>> To unsubscribe from this group and stop receiving emails from it, send 
>>>> an email to cp2k+... at googlegroups.com.
>>>> To post to this group, send email to cp... at googlegroups.com.
>>>> Visit this group at https://groups.google.com/group/cp2k.
>>>> For more options, visit https://groups.google.com/d/optout.
>>>>
>>>
>>> -- 
>> You received this message because you are subscribed to the Google Groups 
>> "cp2k" group.
>> To unsubscribe from this group and stop receiving emails from it, send an 
>> email to cp2k+... at googlegroups.com.
>> To post to this group, send email to cp... at googlegroups.com.
>> Visit this group at https://groups.google.com/group/cp2k.
>> For more options, visit https://groups.google.com/d/optout.
>>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20160414/13efb3f5/attachment.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: Cray-XC40-gfortran-mkl-smm.popt
Type: application/octet-stream
Size: 1334 bytes
Desc: not available
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20160414/13efb3f5/attachment.obj>