[Paraview] Error cross-compiling Catalyst on BG/Q

[Biddiscombe, John A.]

NB cake = cmake + aggressive autocorrect

From: <Biddiscombe>, cscs <biddisco at cscs.ch<mailto:biddisco at cscs.ch>>
Date: Tuesday 17 February 2015 21:20
To: "cook47 at llnl.gov<mailto:cook47 at llnl.gov>" <cook47 at llnl.gov<mailto:cook47 at llnl.gov>>
Cc: "paraview at paraview.org<mailto:paraview at paraview.org>" <paraview at paraview.org<mailto:paraview at paraview.org>>
Subject: Re: [Paraview] Error cross-compiling Catalyst on BG/Q

Yes indeed. What I really meant was you can just compile with cake and use the wrappers, there’s no need to setup a special toolchain and get all the cross-compilation stuff going because the wrappers do all that for you.

On BGQ I am using clang instead of xlc o I’ve got a different setup.

JB

From: <Cook>, "cook47 at llnl.gov<mailto:cook47 at llnl.gov>" <cook47 at llnl.gov<mailto:cook47 at llnl.gov>>
Date: Tuesday 17 February 2015 18:30
To: cscs <biddisco at cscs.ch<mailto:biddisco at cscs.ch>>
Cc: David DeMarle <dave.demarle at kitware.com<mailto:dave.demarle at kitware.com>>, "paraview at paraview.org<mailto:paraview at paraview.org>" <paraview at paraview.org<mailto:paraview at paraview.org>>
Subject: Re: [Paraview] Error cross-compiling Catalyst on BG/Q

Thanks for that information.  Could be a life-saver!  However, please note that I checked with a local expert and he says,  "if you want any optimization on the compute nodes you should use the cross compilers.  Likewise, if you use optimization on your frontend compile, you won't be able to run on the compute nodes.  They're not fully binary compatible and they do not use the same OS.  So your mileage will vary with this approach.”

— Rich

On Feb 12, 2015, at 10:55 PM, Biddiscombe, John A. <biddisco at cscs.ch<mailto:biddisco at cscs.ch>> wrote:

Rich

On the BGQ, the compute nodes and login nodes are binary compatible, so you shouldn’t need to bother with most of the cross compiling stuff. the only thing to watch out for is the MPI and since vtkParseOGLExt and vtkWrapPython etc don’t touch that, you ought to be ok compiling
_as if_ you were going to run on the host node. If you use the MPI wrappers (I don’t), then all will be taken care of.

I don’t think that I have compiled pv+catalyst to run on BGQ, so my comment is only based on theory, but I can set a build going in a minute if it helps (except I’d need to setup mesa which may take a while, so maybe I won’t....)

JB


From: ParaView [mailto:paraview-bounces at paraview.org] On Behalf Of David E DeMarle
Sent: 13 February 2015 01:19
To: Cook, Rich
Cc: ParaView list
Subject: Re: [Paraview] Error cross-compiling Catalyst on BG/Q


On Thu, Feb 12, 2015 at 7:15 PM, Cook, Rich <cook47 at llnl.gov<mailto:cook47 at llnl.gov>> wrote:
/collab/usr/global/tools/Kitware/Catalyst/bgqos_0/catalyst_backend_crossbuild/CMakeFiles/cmTryCompileExec2363604927-CMAKE_REQUIRE_LARGE_FILE_SUPPORT

Is the name of the executable to run.

set( CMAKE_REQUIRE_LARGE_FILE_SUPPORT
     "PLEASE_FILL_OUT-FAILED_TO_RUN"
     CACHE STRING "Result from TRY_RUN" FORCE)

Change that to:
set( CMAKE_REQUIRE_LARGE_FILE_SUPPORT
     0
     CACHE STRING "Result from TRY_RUN" FORCE)

if the return code of that executable is 0

set( CMAKE_REQUIRE_LARGE_FILE_SUPPORT__TRYRUN_OUTPUT
     "PLEASE_FILL_OUT-NOTFOUND"
     CACHE STRING "Output from TRY_RUN" FORCE)

Change that to:
set( CMAKE_REQUIRE_LARGE_FILE_SUPPORT__TRYRUN_OUTPUT
     "I am the very model of a modern major general"
     CACHE STRING "Output from TRY_RUN" FORCE)

If the console output of that executable was "I am the very model of a modern major general"

David E DeMarle
Kitware, Inc.
R&D Engineer
21 Corporate Drive
Clifton Park, NY 12065-8662
Phone: 518-881-4909

--
✐Richard Cook
✇ Lawrence Livermore National Laboratory
Bldg-453 Rm-4024, Mail Stop L-557
7000 East Avenue,  Livermore, CA, 94550, USA
☎ (office) (925) 423-9605
☎ (fax) (925) 423-6961
---
Information Management & Graphics Grp., Services & Development Div., Integrated Computing & Communications Dept.
(opinions expressed herein are mine and not those of LLNL)



-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://public.kitware.com/pipermail/paraview/attachments/20150217/c4476671/attachment.html>