jowr / librefprop.so

Create a shared library from the Fortran sources provided by Refprop from NIST. This project provides an alternative to the refprop.dll that comes with the software. Please use the official instructions if possible

Home Page:https://github.com/usnistgov/REFPROP-cmake

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

OPENMP

ibell opened this issue · comments

I guess we have most of the issues resolved on OSX, but I tried to rebuild today and I couldn't. It worked fine if I commented out the -fopenmp. Was there some magic that we did to get it to build? I think it is ok to not have openmp support though. I guess the issue was with my gfortran, right?

Hm, no idea about gfortran and openmp. I guess we can just disable it.

Yeah, let's just disable it. If people are doing HPC stuff they can dig
into the makefile.

I still can't get the compiled version of REFPROP to work with MATLAB on
OSX. I think I am just going to give up. It works
MATLAB->CoolProp->REFPROP though.

On Wed, Mar 12, 2014 at 7:29 PM, Jorrit Wronski notifications@github.comwrote:

Hm, no idea about gfortran and openmp. I guess we can just disable it.

Reply to this email directly or view it on GitHubhttps://github.com//issues/8#issuecomment-37445781
.

Hm, I am very curious to hear what @nkampy did to solve this issue. I just borrowed a Mac and run into the same issue:

...
*$omp threadprivate(/FSHSAV/)                                           
     1
Error: Unclassifiable OpenMP directive at (1)
src/PASS_FTN_ALT.f:0: sorry, unimplemented: threadprivate variables not supported in this target
COMMONS.f:696.6:
    Included at src/PASS_FTN_ALT.f:44:

*$omp threadprivate(/FEQSAV/)                                           
     1
Error: Unclassifiable OpenMP directive at (1)
Fatal Error: Error count reached limit of 25.
...

Does anyone know what that error actually means? Are we doing anything wrong? Disabling OpenMP by commenting the -fopenmp flag solves the problem, but I guess we all prefer to use OpenMP.

Jowr

I didn't have that issue with the orginal code. Let me finish testing the current method now...
What step are you at? What version of mac OSX? what model mac? what gcc? what ver of matlab?

On Apr 28, 2014, at 2:39 PM, Jorrit Wronski notifications@github.com wrote:

Hm, I am very curious to hear what @nkampy did to solve this issue. I just borrowed a Mac and run into the same issue:

...
*$omp threadprivate(/FSHSAV/)
1
Error: Unclassifiable OpenMP directive at (1)
src/PASS_FTN_ALT.f:0: sorry, unimplemented: threadprivate variables not supported in this target
COMMONS.f:696.6:
Included at src/PASS_FTN_ALT.f:44:

*$omp threadprivate(/FEQSAV/)
1
Error: Unclassifiable OpenMP directive at (1)
Fatal Error: Error count reached limit of 25.
...
Does anyone know what that error actually means? Are we doing anything wrong? Disabling OpenMP by commenting the -fopenmp flag solves the problem, but I guess we all prefer to use OpenMP.


Reply to this email directly or view it on GitHub.

It might be that you use gfortran packages that do not come from the same sources. When I was working on refpropm mex file compilation on OS X some years ago, it was just really tricky to have people install only the right "flavor" of gfortran. At the time, that version was coming from the R-cran team. It's not online anymore but I still have the DMG file in some hard drives, if you're interested. What "flavor" of gfortran have you installed?

Thanks for the comments. I use an older MacBook running 10.6.8 with the latest compatible gfortran version from HPC.

MacBook-Pro:librefprop.so.git jorrit$ sw_vers
ProductName:    Mac OS X
ProductVersion: 10.6.8
BuildVersion:   10K549
MacBook-Pro:librefprop.so.git jorrit$ uname -a
Darwin MacBook-Pro.local 10.8.0 Darwin Kernel Version 10.8.0: Tue Jun  7 16:33:36 PDT 2011; root:xnu-1504.15.3~1/RELEASE_I386 i386
MacBook-Pro:librefprop.so.git jorrit$ gfortran --version
GNU Fortran (GCC) 4.2.1 (Apple Inc. build 5664)
Copyright (C) 2007 Free Software Foundation, Inc.

I've got a mid 2009 Mac Pro running 10.9.2 with gcc 4.8.2 and have no problems compiling with the openmp option. It's probably a Version issue. Is upgrading the operating system an option to enable a higher version of gcc?

Sent from my iPhone

On Apr 28, 2014, at 11:51 PM, Jorrit Wronski notifications@github.com wrote:

Thanks for the comments. I use an older MacBook running 10.6.8 with the latest compatible gfortran version from HPC.

MacBook-Pro:librefprop.so.git jorrit$ sw_vers
ProductName: Mac OS X
ProductVersion: 10.6.8
BuildVersion: 10K549
MacBook-Pro:librefprop.so.git jorrit$ uname -a
Darwin MacBook-Pro.local 10.8.0 Darwin Kernel Version 10.8.0: Tue Jun 7 16:33:36 PDT 2011; root:xnu-1504.15.3~1/RELEASE_I386 i386
MacBook-Pro:librefprop.so.git jorrit$ gfortran --version
GNU Fortran (GCC) 4.2.1 (Apple Inc. build 5664)
Copyright (C) 2007 Free Software Foundation, Inc.

Reply to this email directly or view it on GitHub.

No, upgrading is not an option due to a lack of disk space. I included a switch in the makefile and added a known problems section in the readme.