[CP2K-user] [CP2K:11157] Re: van der Waals regtests fail on Intel KNL, and build glitches

Ronald Cohen reco... at gmail.com
Wed Jan 9 13:50:23 CET 2019


OK! Good news. The cp2k.ssmp finished building and the argon06 test gives exactly the correct number with OMP=64
and takes only 4.871 seconds. I will see how it works with my real job. I will also try building the mpi versions. I did notice a problem with the make dependencies.
I had to do a make clean before cp2k.ssmp would build. Otherwise it complains about missing modules and doesn’t try to make them.

I included the flags -mavx512f -mavx512er -mavx512cd -mavx512pf  so that the KNL features would be used.

Thanks for all of your help.

Sincerely,

Ron

---
Ron Cohen
reco... at gmail.com
skypename: ronaldcohen
twitter: @recohen3




> On Jan 9, 2019, at 12:36 PM, Alfio Lazzaro <alfio.... at gmail.com> wrote:
> 
> Well, in any case I'm glad that you found the solution ;) I hope it will solve the other problems.
> 
> 
> 
> Il giorno mercoledì 9 gennaio 2019 12:31:05 UTC+1, Ronald Cohen ha scritto:
> No No! I did not edit the arch file:
> 
> grep CPP arch/local.ssmp
> CPP         =
> 
> 
> ---
> Ron Cohen
> rec... at gmail.com
> skypename: ronaldcohen
> twitter: @recohen3
> 
> 
> 
> 
>> On Jan 9, 2019, at 12:21 PM, Alfio Lazzaro <alfi... at gmail.com> wrote:
>> 
>> To be more precise, probably you have in your arch/local.ssmp
>> 
>> CPP = gcc
>> CPPFLAGS   = -E
>> 
>> Although this is strange (I've never had these flags set in my toolchain), I get a similar problem if I do set in my arch file.
>> This is a bug in the CP2K makefile:
>> 
>> https://github.com/cp2k/cp2k/blob/700dfe71c0b99fbb8510f414e08a90aacc047890/Makefile#L444-L447
>> 
>> I will open a PR for that. Thanks for reporting.
>> 
>> 
>> 
>> 
>> 
>> Il giorno mercoledì 9 gennaio 2019 12:08:44 UTC+1, Alfio Lazzaro ha scritto:
>> OK, for some reason you set
>> 
>> CPP
>> 
>> in the arch file. Could you check under arch directory what is the value of CPP?
>> An easy solution would be to remove it.
>> 
>> 
>> 
>> Il giorno mercoledì 9 gennaio 2019 09:39:41 UTC+1, Ronald Cohen ha scritto:
>> This shows clearly that the file cannot be called fypped:
>> 
>> cp qs_basis_gradient.f90 t.fypped
>> rcohen at tomcat3:~/CP2K/cp2k/src$ gcc -E -cpp t.fypped
>> gcc: warning: t.fypped: linker input file unused because linking not done
>> rcohen at tomcat3:~/CP2K/cp2k/src$ cp t.fypped t.f
>> rcohen at tomcat3:~/CP2K/cp2k/src$ gcc -E -cpp t.f | head
>> # 1 "t.f"
>> # 1 "<built-in>"
>> # 1 "<command-line>"
>> # 1 "t.f"
>> !--------------------------------------------------------------------------------------------------!
>> !   CP2K: A general program to perform molecular dynamics simulations                              !
>> !   Copyright (C) 2000 - 2018  CP2K developers group                                               !
>> !--------------------------------------------------------------------------------------------------!
>> 
>> ! *****************************************************************************
>> 
>> ---
>> Ron Cohen
>> rec... at gmail.com
>> skypename: ronaldcohen
>> twitter: @recohen3
>> 
>> 
>> 
>> 
>>> On Jan 9, 2019, at 9:29 AM, Ronald Cohen <rec... at gmail.com> wrote:
>>> 
>>> It seems that gcc -E will not work on a file called x.fypped . If I call this file x.f90 and do gcc -E -cpp on it preprocesses. So it seems the toolchain doesn’t not work properly.
>>> 
>>> Ron
>>> 
>>> ---
>>> Ron Cohen
>>> rec... at gmail.com
>>> skypename: ronaldcohen
>>> twitter: @recohen3
>>> 
>>> 
>>> 
>>> 
>>>> On Jan 9, 2019, at 9:17 AM, Ronald Cohen <rec... at gmail.com> wrote:
>>>> 
>>>> It seems that the gcc -E doesn’t work:
>>>> 
>>>> /home/rcohen/CP2K/cp2k/tools/toolchain/install/gcc-8.2.0/bin/gcc -E -D__COMPILE_ARCH="\"local\"" -D__COMPILE_DATE="\"Wed Jan  9 08:49:48 CET 2019\"" -D__COMPILE_HOST="\"tomcat3\"" -D__COMPILE_REVISION="\"git:89b110c40\"" -D__DATA_DIR="\"/home/rcohen/CP2K/cp2k/data\"" -I/home/rcohen/CP2K/cp2k/src -D__SHORT_FILE__="\"qs_basis_gradient.F\"" -I'/home/rcohen/CP2K/cp2k/src/' qs_basis_gradient.fypped > qs_basis_gradient.f90
>>>> gcc: warning: qs_basis_gradient.fypped: linker input file unused because linking not done
>>>> 
>>>> 
>>>> It outputs a zero length file:
>>>> 
>>>> -rw-r--r--  1 rcohen tomcat   15467 Jan  9 08:49 qs_basis_gradient.fypped
>>>> -rw-r--r--  1 rcohen tomcat       0 Jan  9 08:55 qs_basis_gradient.f90
>>>> 
>>>> It must work sometimes because some things get built!
>>>> 
>>>> If I use /usr/bin/gcc (gcc version 6.3.0 20170516 (Debian 6.3.0-18+deb9u1)) the same thing happens.
>>>> 
>>>> I do not see anything wrong with qs_basis_gradient.fypped . Why does gcc -E output nothing?
>>>> 
>>>> Ron
>>>> 
>>>> 
>>>> --
>>>> You received this message because you are subscribed to a topic in the Google Groups "cp2k" group.
>>>> To unsubscribe from this topic, visit https://groups.google.com/d/topic/cp2k/gzmRqKNt62U/unsubscribe.
>>>> To unsubscribe from this group and all its topics, 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 a topic in the Google Groups "cp2k" group.
>> To unsubscribe from this topic, visit https://groups.google.com/d/topic/cp2k/gzmRqKNt62U/unsubscribe.
>> To unsubscribe from this group and all its topics, 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 a topic in the Google Groups "cp2k" group.
> To unsubscribe from this topic, visit https://groups.google.com/d/topic/cp2k/gzmRqKNt62U/unsubscribe.
> To unsubscribe from this group and all its topics, send an email to cp2k+uns... 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 --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 849 bytes
Desc: Message signed with OpenPGP
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20190109/104e58ce/attachment.sig>


More information about the CP2K-user mailing list