Community
cancel
Showing results for 
Search instead for 
Did you mean: 
Highlighted
Beginner
118 Views

internal compiler error 101004_111

Hi,

 

I am trying to compile ParaView 5.2.0 RC3 (http://www.paraview.org) with GUI support. The GUI is written in QT and the compiler gives error like following,

[ 88%] Building CXX object Qt/Components/CMakeFiles/pqComponents.dir/pqPresetToPixmap.cxx.o"

: internal error: 101004_111

compilation aborted for /okyanus/users/uturuncoglu/progs/paraview-5.2.0rc3/src/Qt/Components/pqPresetToPixmap.cxx (code 4)

make[2]: *** [Qt/Components/CMakeFiles/pqComponents.dir/pqPresetToPixmap.cxx.o] Error 4

make[1]: *** [Qt/Components/CMakeFiles/pqComponents.dir/all] Error 2make: *** [all] Error 2

I am using icpc version 17.0.0 (gcc version 4.8.5 compatibility) - 2017.0.098 under CentOS Linux release 7.2.1511 (Core) - 3.10.0-327.36.1.el7.x86_64 #1 SMP Sun Sep 18 13:04:29 UTC 2016 x86_64 x86_64 x86_64 GNU/Linux. The same code compiles without any problem with icpc version 16.0.1 (gcc version 4.8.5 compatibility). I just wonder about the meaning of the internal error code 101004_111. Is it possible to solve the issue or do i need to wait until it is fixed by Intel probably in the future releases?

Regards,

--ufuk

 

0 Kudos
15 Replies
Highlighted
Black Belt
118 Views

Intel(R) C++ Intel(R) 64 Compiler for applications running on Intel(R) 64, Versi
on 17.0.1.143 Build 20161005
Copyright (C) 1985-2016 Intel Corporation.  All rights reserved.

pqPresetToPixmap.cxx
pqPresetToPixmap.cxx(32): catastrophic error: cannot open source file "pqPresetT
oPixmap.h"
  #include "pqPresetToPixmap.h"
                               ^

compilation aborted for pqPresetToPixmap.cxx (code 4)

Having current version doesn't compensate for missing include file.

0 Kudos
Highlighted
Beginner
118 Views

Thanks for your help. I think that i mislead to you. Actually, i just put the single source file that causes the internal error in the compiler just for reference. You might want to see the code that Intel C++ compiler have problem with it. In fact that the file is not the whole package (ParaView application) and does not contain all the source codes and header files. So, it is normal that the compiler complains about the header file. I just wonder about the meaning of the internal compiler error with that specific number. If you want to reproduce error, you need to get whole source code and compile it.

http://www.paraview.org/download/

http://www.paraview.org/paraview-downloads/download.php?submit=Download&version=v5.2&type=source&os=...

0 Kudos
Highlighted
Employee
118 Views

Hi, Ufuk

This internal compiler error is a crash in compiler itself, not the program. The number also refers to an internal crash number.

To fix the issue we need to reproduce the issue and create a reproducer if possible and escalate for resolution.

Since I need to download and build the whole source code to reproduce, it will take me sometime for investigation. I will let you know for an update on this issue.

Thanks.

 

0 Kudos
Highlighted
Beginner
118 Views

Okay. Thanks for your help. In the mean time i think it is good to downgrade the compiler version to 2016 which was working. BTW, let me know if you need any help about it. I could also provide the details about the installation. The problem appears when the code try to compile GUI (written in QT4). My test environment was Centos 7.2 and the installed qt packages

qt-settings-19-23.5.el7.centos.noarch
qt-4.8.5-11.el7.x86_64
qt-x11-4.8.5-11.el7.x86_64
qt-devel-4.8.5-11.el7.x86_64 

Regards,

--ufuk

0 Kudos
Highlighted
118 Views

I've also got the similar error while compiling ParaView 5.0.1 with ICC 17.0.1 20161005. Does anybody find the fix to this issue ??

The error log is the following:

[ 96%] Building CXX object Qt/Components/CMakeFiles/pqComponents.dir/pqPropertyWidgetDecorator.cxx.o
[ 96%] Building CXX object Qt/Components/CMakeFiles/pqComponents.dir/pqPropertyWidgetInterface.cxx.o
": internal error: 101004_111

compilation aborted for /home/sergey/intel/ThirdParty-dev/ParaView-5.0.1/Qt/Components/pqPresetToPixmap.cxx (code 4)
make[2]: *** [Qt/Components/CMakeFiles/pqComponents.dir/pqPresetToPixmap.cxx.o] Error 4
make[2]: *** Waiting for unfinished jobs....
make[1]: *** [Qt/Components/CMakeFiles/pqComponents.dir/all] Error 2
make: *** [all] Error 2
Command exited with non-zero status 2
7027.32user 1798.27system 18:54.00elapsed 778%CPU (0avgtext+0avgdata 1336456maxresident)k
7664inputs+8998312outputs (1major+180665480minor)pagefaults 0swaps

 

0 Kudos
Highlighted
Beginner
118 Views

Currently, I have no success with 2017. It is compiled with Intel compiler 2016u1 without any problem. So, downgrading to earlier version of the Intel compiler will solve the problem.

0 Kudos
Highlighted
Employee
118 Views

Hi, Ufuk

Sorry for my late response.

I have setup the environment to build Paraview(the package provided in your link), but cannot reproduce the internal error with icc 17.0 Update1.

What's the compiler flags you used? I used below cmake configure command. Please let me know if anything is missed:

 ccmake ParaView-v5.2.0-RC3 -DCMAKE_CXX_COMPILER=icpc -DCMAKE_CC_COMPILER=icc

 

Thanks.

0 Kudos
Highlighted
Beginner
118 Views

Hi Yolanda,

 

You could find the options that I used to compile ParaView in the attached file. Please let me know, if you need more information.

Regards,

--ufuk

0 Kudos
Highlighted
Employee
118 Views

Hi, Ufuk and Serge

Thank you for the information.

I have reproduced the internal error you reported. Previously I built for debug type and cannot reproduce. After change to release type, it reproduced.

I am still root-causing the problem. I will get back to you shortly when I have a more thorough understanding of the problem.

Thank you.
 

0 Kudos
Highlighted
Beginner
118 Views

Okay. That is great. Anyway, let me know, if you need help about it. Thanks for your help and support.

Best Regards,

--ufuk

0 Kudos
Highlighted
Employee
118 Views

Hi, Ufuk and Serge

This crash relates to below loop from file Qt/Components/pqPresetToPixmap.cxx, line 258 in function "renderIndexedColorTransferFunction":

    for (int i = Nh / 2; i < Nh; ++i)
    {
      double ar = Nv * wmp / static_cast<double>(hmp * Nh);
      double q = (ar >= 1.0) ? ar : 1. / ar;
      if (q < bestQ)
      {
        bestQ = q;
        best = i;
      }
    }

This loop looks unnecessary and caused compiler crash when try to vectorize it. Not sure if this loop is intended, after remove the for loop, the program compiles:

    //for (int i = Nh / 2; i < Nh; ++i)
    //{
      double ar = Nv * wmp / static_cast<double>(hmp * Nh);
      double q = (ar >= 1.0) ? ar : 1. / ar;
      if (q < bestQ)
      {
        bestQ = q;
        best = Nh-1;//i;
      }
    //}

Anyway the internal crash is still a compiler bug and need a fix. I have entered it in our problem tracking system. We will try to resolve this issue as soon as we can. However, please be advised that this issue may have to be targeted to for the next update or major release. I will let you know when I have an update on this issue.

Let me know if this work-around helps.

Thanks.

0 Kudos
Highlighted
118 Views

Yes, your work-around works.

Is it any news about compiler fix for this issue ?

0 Kudos
Highlighted
118 Views

One additional remark. Yes, it was compiled without any error, however the ParaView crashes at the beginning of execution. That's why this workaround cannot be applied to fix the issue.

0 Kudos
Highlighted
Employee
118 Views

Hi, Sergey

Engineer is working on a fix of this issue in high priority.

But for the loop itself, it still looks to me a possibly bug in Paraview. I think the intention of this loop is to find a certain i between Nh/2 to Nh that gets bestQ. However each q is calculated regardless i, which does not looks correct.

I'd suggest to consult Paraview to confirm this code piece.

Thanks. 

 

0 Kudos
Highlighted
118 Views

Hi, Chen.

About my latest post about a crash of ParaView after applying your workaround. I'm sorry it was just a tiny issue of my particular system. I've solved it and now Paraview works. Therefore, your workaround works fine and really solves an issue.

With best regards, Sergey

0 Kudos