[CP2K-user] [CP2K:20306] Re: Errors in Compiling CP2K 9.1

[Lenard Carroll]

I assumed it would find the path immediately, since it is in my bashrc. In 
any case, I am now recompiling but letting cp2k install cosma

On Tuesday, June 11, 2024 at 4:44:18 PM UTC+2 Frederick Stein wrote:

> Make cannot find the COSMA library. You have told the toolchain to pick it 
> up from your system. You have to add the respective paths either to your 
> PATH variable or to your arch file or, if you do not want to use COSMA, 
> remove the "-D__COSMA" flag from your arch files and recompile.
> Lenard Carroll schrieb am Dienstag, 11. Juni 2024 um 16:37:53 UTC+2:
>
>> I have now avoided editing the bash file, and went with the following:
>> ./install_cp2k_toolchain.sh -j 20 --mpi-mode=openmpi --math-mode=openblas 
>> --with-gcc=system --with-intel=no --with-cmake=system --with-openmpi=system 
>> --with-mpich=no --with-intelmpi=no --with-openblas=system --with-elpa=no 
>> --with-spfft=system --with-spla=system --with-cosma=system --with-mkl=no
>>
>> make -j 20 ARCH=local VERSION="psmp"
>>
>> I have also gone with version 2024.1
>>
>> I installed it in home. The compilation does not succeed, and I get the 
>> error messages attached
>>
>> On Tuesday, June 11, 2024 at 2:31:07 PM UTC+2 Lenard Carroll wrote:
>>
>>> I am sticking to version 9.1 for consistency purposes with other work I 
>>> have done, and because I have managed to compile it before. The newest 
>>> version gave me problems with arguments in the bash file, not 9.1.
>>>
>>> For what it prints, it complained about any argument I would use, such 
>>> as --gpu-ver=V100 or --enable-cuda=yes
>>>
>>> On Tuesday, June 11, 2024 at 2:19:03 PM UTC+2 Frederick Stein wrote:
>>>
>>>> What does it print?
>>>> Why do you actually stick to version 9.1? There is already a version 
>>>> 2024.1.
>>>>
>>>> Lenard Carroll schrieb am Dienstag, 11. Juni 2024 um 14:16:57 UTC+2:
>>>>
>>>>> Hi Frederick
>>>>>
>>>>> I have done this before, but the .sh script wouldn't accept my 
>>>>> arguments. I will try this with version 9.1 again, and post the update
>>>>>
>>>>> On Tuesday, June 11, 2024 at 2:13:42 PM UTC+2 Frederick Stein wrote:
>>>>>
>>>>>> Dear Lenard,
>>>>>> The Toolchain script has several options which you may check by 
>>>>>> running "./install_cp2k_toolchain.sh -h". It should tell you that you can 
>>>>>> turn off features such as MKL or ELPA with flags like "--with-mkl=no" or 
>>>>>> "--with-elpa=no". I kindly ask you to consider the other available options 
>>>>>> as well.
>>>>>> If you have rerun the original toolchain script setting the relevant 
>>>>>> options, post the respective commands+outputs here, if it still fails.
>>>>>> Best,
>>>>>> Frederick
>>>>>>
>>>>>> Lenard Carroll schrieb am Dienstag, 11. Juni 2024 um 13:54:30 UTC+2:
>>>>>>
>>>>>>> Good day, I am trying to compile CP2K 9.1 for use with an AIMD 
>>>>>>> Ryzen-3960X and 2 RTX 3070 Ti GPUs. Ubuntu 20.04
>>>>>>>
>>>>>>> I started off by editing the install_cp2k_toolchain.sh to not use 
>>>>>>> MKL or ELPA. To use OpenMPI, GPU version V100, to enable CUDA and to 
>>>>>>> install a few libraries/packages. Many of them I also compiled manually 
>>>>>>> because I was getting many errors doing it through CP2K's toolchain. I am 
>>>>>>> including the toolchain file just in case.
>>>>>>>
>>>>>>> Unfortunately I have obtained several errors.
>>>>>>>
>>>>>>> The toolchain process goes smoothly, it is the *make *process that 
>>>>>>> is not going well:
>>>>>>>
>>>>>>> I have included the output of the make process
>>>>>>>
>>>>>>>

-- 
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+unsubscribe at googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/cf553df8-9f1e-4651-8f86-119377e7acd5n%40googlegroups.com.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20240611/9f52141b/attachment.htm>