Menu
▾
▴
pfunit-support
|
From: Jesse L. <lo...@st...> - 2013-07-18 19:49:49
|
Hello all, I've just cloned pFUnit from the master branch on the sourceforge repository and I've been unable to successfully compile the code with a variety of compilers/platforms. 1. ifort 12.1.0 on linux https://gist.github.com/yosoyjay/6032388 2. gfortran 4.4.6 on linux (I know, it's ancient) https://gist.github.com/yosoyjay/6032415 3. gfortran 4.7.8 on os x 10.7 https://gist.github.com/yosoyjay/6032437 Any hints on how to get it compiled and working would be much appreciated. Thanks, -Jesse |
|
From: Tom C. <Tho...@na...> - 2013-07-18 20:22:03
|
Hi Jesse, For ifort, you probably need a more recent compiler, though gfortran 4.7.8 ought to have worked. We recommend ifort 13.x and gfortran 4.7/4.8 If you can provide a log of what goes wrong with cases 1 and 2 we'll help trouble shoot. (Case 2 simply will not work.) One obvious thing that others have pointed out (and should go in the documentation once we push it to the repository): you need to define the variable F90 to be ifort/gfortran. Or you can use F90_VENDOR=Intel/GNU. Cheers, - Tom On Jul 18, 2013, at 3:49 PM, Jesse Lopez <lo...@st...> wrote: > Hello all, > > I've just cloned pFUnit from the master branch on the sourceforge repository and I've been unable to successfully compile the code with a variety of compilers/platforms. > > 1. ifort 12.1.0 on linux > https://gist.github.com/yosoyjay/6032388 > > 2. gfortran 4.4.6 on linux (I know, it's ancient) > https://gist.github.com/yosoyjay/6032415 > > 3. gfortran 4.7.8 on os x 10.7 > https://gist.github.com/yosoyjay/6032437 > > Any hints on how to get it compiled and working would be much appreciated. > > Thanks, > -Jesse > ------------------------------------------------------------------------------ > See everything from the browser to the database with AppDynamics > Get end-to-end visibility with application monitoring from AppDynamics > Isolate bottlenecks and diagnose root cause in seconds. > Start your free trial of AppDynamics Pro today! > http://pubads.g.doubleclick.net/gampad/clk?id=48808831&iu=/4140/ostg.clktrk_______________________________________________ > Pfunit-support mailing list > Pfu...@li... > https://lists.sourceforge.net/lists/listinfo/pfunit-support Thomas Clune, Ph. D. <Tho...@na...> Chief, Software Systems Support Office Code 610.3 NASA GSFC 301-286-4635 MS 610.8 B33-C128 <http://ssso.gsfc.nasa.gov> Greenbelt, MD 20771 |
|
From: Jesse L. <lo...@st...> - 2013-07-18 20:41:47
|
Hi Tom, Thanks for the prompt response. I had in fact specified the F90*, so I was good there. I did successfully compile with ifort 13.0 as you suggested. But, I ran into some trouble with 'make tests MPI=YES' with complaints from gfortran! https://gist.github.com/yosoyjay/6032859 Any idea about the gfortran 4.7.3 error on OS X? Thanks again for your help. -Jesse On Thu, Jul 18, 2013 at 3:21 PM, Tom Clune <Tho...@na...> wrote: > Hi Jesse, > > For ifort, you probably need a more recent compiler, though gfortran 4.7.8 > ought to have worked. We recommend ifort 13.x and gfortran 4.7/4.8 > > If you can provide a log of what goes wrong with cases 1 and 2 we'll help > trouble shoot. (Case 2 simply will not work.) One obvious thing that > others have pointed out (and should go in the documentation once we push it > to the repository): you need to define the variable F90 to be > ifort/gfortran. Or you can use F90_VENDOR=Intel/GNU. > > Cheers, > > - Tom > > > On Jul 18, 2013, at 3:49 PM, Jesse Lopez <lo...@st...> wrote: > > Hello all, > > I've just cloned pFUnit from the master branch on the sourceforge > repository and I've been unable to successfully compile the code with a > variety of compilers/platforms. > > 1. ifort 12.1.0 on linux > https://gist.github.com/yosoyjay/6032388 > > 2. gfortran 4.4.6 on linux (I know, it's ancient) > https://gist.github.com/yosoyjay/6032415 > > 3. gfortran 4.7.8 on os x 10.7 > https://gist.github.com/yosoyjay/6032437 > > Any hints on how to get it compiled and working would be much appreciated. > > Thanks, > -Jesse > > ------------------------------------------------------------------------------ > See everything from the browser to the database with AppDynamics > Get end-to-end visibility with application monitoring from AppDynamics > Isolate bottlenecks and diagnose root cause in seconds. > Start your free trial of AppDynamics Pro today! > > http://pubads.g.doubleclick.net/gampad/clk?id=48808831&iu=/4140/ostg.clktrk_______________________________________________ > Pfunit-support mailing list > Pfu...@li... > https://lists.sourceforge.net/lists/listinfo/pfunit-support > > > Thomas Clune, Ph. D. <Tho...@na...> > Chief, Software Systems Support Office Code 610.3 > NASA GSFC 301-286-4635 > MS 610.8 B33-C128 <http://ssso.gsfc.nasa.gov> > Greenbelt, MD 20771 > > > > > > |
|
From: Tom C. <tho...@na...> - 2013-07-18 20:46:19
|
Hi Jesse, The error log suggests that you are somehow getting Ifort compile flags with gfortran. So either you don't have F90=gfortran for that case or you have F90_VENDOR set to Intel. Seems unrelated to MPI. - Tom On Jul 18, 2013, at 4:41 PM, Jesse Lopez <lo...@st...> wrote: > Hi Tom, > > Thanks for the prompt response. I had in fact specified the F90*, so I was good there. > I did successfully compile with ifort 13.0 as you suggested. But, I ran into some trouble > with 'make tests MPI=YES' with complaints from gfortran! > > https://gist.github.com/yosoyjay/6032859 > > Any idea about the gfortran 4.7.3 error on OS X? > > Thanks again for your help. > > -Jesse > > > On Thu, Jul 18, 2013 at 3:21 PM, Tom Clune <Tho...@na...> wrote: > Hi Jesse, > > For ifort, you probably need a more recent compiler, though gfortran 4.7.8 ought to have worked. We recommend ifort 13.x and gfortran 4.7/4.8 > > If you can provide a log of what goes wrong with cases 1 and 2 we'll help trouble shoot. (Case 2 simply will not work.) One obvious thing that others have pointed out (and should go in the documentation once we push it to the repository): you need to define the variable F90 to be ifort/gfortran. Or you can use F90_VENDOR=Intel/GNU. > > Cheers, > > - Tom > > > On Jul 18, 2013, at 3:49 PM, Jesse Lopez <lo...@st...> wrote: > >> Hello all, >> >> I've just cloned pFUnit from the master branch on the sourceforge repository and I've been unable to successfully compile the code with a variety of compilers/platforms. >> >> 1. ifort 12.1.0 on linux >> https://gist.github.com/yosoyjay/6032388 >> >> 2. gfortran 4.4.6 on linux (I know, it's ancient) >> https://gist.github.com/yosoyjay/6032415 >> >> 3. gfortran 4.7.8 on os x 10.7 >> https://gist.github.com/yosoyjay/6032437 >> >> Any hints on how to get it compiled and working would be much appreciated. >> >> Thanks, >> -Jesse >> ------------------------------------------------------------------------------ >> See everything from the browser to the database with AppDynamics >> Get end-to-end visibility with application monitoring from AppDynamics >> Isolate bottlenecks and diagnose root cause in seconds. >> Start your free trial of AppDynamics Pro today! >> http://pubads.g.doubleclick.net/gampad/clk?id=48808831&iu=/4140/ostg.clktrk_______________________________________________ >> Pfunit-support mailing list >> Pfu...@li... >> https://lists.sourceforge.net/lists/listinfo/pfunit-support > > Thomas Clune, Ph. D. <Tho...@na...> > Chief, Software Systems Support Office Code 610.3 > NASA GSFC 301-286-4635 > MS 610.8 B33-C128 <http://ssso.gsfc.nasa.gov> > Greenbelt, MD 20771 > > > > > > Thomas Clune, Ph. D. <Tho...@na...> Chief, Software Systems Support Office Code 610.3 NASA GSFC 301-286-4635 MS 610.8 B33-C128 <http://ssso.gsfc.nasa.gov> Greenbelt, MD 20771 |
|
From: Jesse L. <lo...@st...> - 2013-07-18 21:13:50
|
Hi Tom, I'm explicitly providing the compiler and vendor. make all F90=ifort F90_VENDOR=Intel [good] make tests F90=ifort F90_VENDOR=Intel [good] make tests MPI=YES F90=ifort F90_VENDOR=Intel [bad] mpif90 -c -assume realloc_lhs -g -O0 -traceback -check uninit -check bounds -check stack -check uninit -DSTRINGIFY_OPERATOR -I../include -DUSE_MPI -DIntel -DLinux -I/state/partition1/home/lopezj/pFUnit/include -DLONG_PTR -DSTRINGIFY_OPERATOR -DIntel -o MpiContext.o MpiContext.F90 gfortran: realloc_lhs: No such file or directory gfortran: uninit: No such file or directory .... I'm a bit perplexed by the gfortran complaints. I also tried make --debug=v MPI=YES F90=ifort F90_VENDOR=Intel But it doesn't reveal much beyond what I already know. It's using Intel.mk and it dies when with the command above. which -a mpif90 /opt/intel/impi/4.1.0.024/intel64/bin/mpif90 Any ideas? Thanks, -Jesse On Thu, Jul 18, 2013 at 3:46 PM, Tom Clune <tho...@na...> wrote: > Hi Jesse, > > The error log suggests that you are somehow getting Ifort compile flags > with gfortran. So either you don't have F90=gfortran for that case or you > have F90_VENDOR set to Intel. > > Seems unrelated to MPI. > > - Tom > > On Jul 18, 2013, at 4:41 PM, Jesse Lopez <lo...@st...> wrote: > > Hi Tom, > > Thanks for the prompt response. I had in fact specified the F90*, so I > was good there. > I did successfully compile with ifort 13.0 as you suggested. But, I ran > into some trouble > with 'make tests MPI=YES' with complaints from gfortran! > > https://gist.github.com/yosoyjay/6032859 > > Any idea about the gfortran 4.7.3 error on OS X? > > Thanks again for your help. > > -Jesse > > > On Thu, Jul 18, 2013 at 3:21 PM, Tom Clune <Tho...@na...>wrote: > >> Hi Jesse, >> >> For ifort, you probably need a more recent compiler, though gfortran >> 4.7.8 ought to have worked. We recommend ifort 13.x and gfortran 4.7/4.8 >> >> If you can provide a log of what goes wrong with cases 1 and 2 we'll help >> trouble shoot. (Case 2 simply will not work.) One obvious thing that >> others have pointed out (and should go in the documentation once we push it >> to the repository): you need to define the variable F90 to be >> ifort/gfortran. Or you can use F90_VENDOR=Intel/GNU. >> >> Cheers, >> >> - Tom >> >> >> On Jul 18, 2013, at 3:49 PM, Jesse Lopez <lo...@st...> wrote: >> >> Hello all, >> >> I've just cloned pFUnit from the master branch on the sourceforge >> repository and I've been unable to successfully compile the code with a >> variety of compilers/platforms. >> >> 1. ifort 12.1.0 on linux >> https://gist.github.com/yosoyjay/6032388 >> >> 2. gfortran 4.4.6 on linux (I know, it's ancient) >> https://gist.github.com/yosoyjay/6032415 >> >> 3. gfortran 4.7.8 on os x 10.7 >> https://gist.github.com/yosoyjay/6032437 >> >> Any hints on how to get it compiled and working would be much >> appreciated. >> >> Thanks, >> -Jesse >> >> ------------------------------------------------------------------------------ >> See everything from the browser to the database with AppDynamics >> Get end-to-end visibility with application monitoring from AppDynamics >> Isolate bottlenecks and diagnose root cause in seconds. >> Start your free trial of AppDynamics Pro today! >> >> http://pubads.g.doubleclick.net/gampad/clk?id=48808831&iu=/4140/ostg.clktrk_______________________________________________ >> Pfunit-support mailing list >> Pfu...@li... >> https://lists.sourceforge.net/lists/listinfo/pfunit-support >> >> >> Thomas Clune, Ph. D. <Tho...@na...> >> Chief, Software Systems Support Office Code 610.3 >> NASA GSFC 301-286-4635 >> MS 610.8 B33-C128 <http://ssso.gsfc.nasa.gov> >> Greenbelt, MD 20771 >> >> >> >> >> >> > > Thomas Clune, Ph. D. <Tho...@na...> > Chief, Software Systems Support Office Code 610.3 > NASA GSFC 301-286-4635 > MS 610.8 B33-C128 <http://ssso.gsfc.nasa.gov> > Greenbelt, MD 20771 > > > > > > |
|
From: Tom C. <tho...@na...> - 2013-07-18 21:21:01
|
Jesse, Ah - I had thought you were trying to use gfortran. Now I think I know what the problem is. Mac OS has a built in version of MPI and it is probably in your path. mpif90 is defaulting to gfortran. You need to make certain that mpif90 in your path points to the Intel compiler. (This probably means building OpenMPI if you've not done that already.) I _think_ that there is an environment variable (MPIF90) to override this as well, but I don't have time to go look it up right now as I'm about to head home. I'll check in again this evening in case this either did not help or you are encountering a different problem. - Tom On Jul 18, 2013, at 5:13 PM, Jesse Lopez <lo...@st...> wrote: > Hi Tom, > > I'm explicitly providing the compiler and vendor. > make all F90=ifort F90_VENDOR=Intel [good] > make tests F90=ifort F90_VENDOR=Intel [good] > make tests MPI=YES F90=ifort F90_VENDOR=Intel [bad] > > mpif90 -c -assume realloc_lhs -g -O0 -traceback -check uninit -check bounds -check stack -check uninit -DSTRINGIFY_OPERATOR -I../include -DUSE_MPI -DIntel -DLinux -I/state/partition1/home/lopezj/pFUnit/include -DLONG_PTR -DSTRINGIFY_OPERATOR -DIntel -o MpiContext.o MpiContext.F90 > gfortran: realloc_lhs: No such file or directory > gfortran: uninit: No such file or directory > .... > > I'm a bit perplexed by the gfortran complaints. I also tried > make --debug=v MPI=YES F90=ifort F90_VENDOR=Intel > > But it doesn't reveal much beyond what I already know. It's using Intel.mk and it dies when with the > command above. > > which -a mpif90 > /opt/intel/impi/4.1.0.024/intel64/bin/mpif90 > > Any ideas? > > Thanks, > -Jesse > > > On Thu, Jul 18, 2013 at 3:46 PM, Tom Clune <tho...@na...> wrote: > Hi Jesse, > > The error log suggests that you are somehow getting Ifort compile flags with gfortran. So either you don't have F90=gfortran for that case or you have F90_VENDOR set to Intel. > > Seems unrelated to MPI. > > - Tom > > On Jul 18, 2013, at 4:41 PM, Jesse Lopez <lo...@st...> wrote: > >> Hi Tom, >> >> Thanks for the prompt response. I had in fact specified the F90*, so I was good there. >> I did successfully compile with ifort 13.0 as you suggested. But, I ran into some trouble >> with 'make tests MPI=YES' with complaints from gfortran! >> >> https://gist.github.com/yosoyjay/6032859 >> >> Any idea about the gfortran 4.7.3 error on OS X? >> >> Thanks again for your help. >> >> -Jesse >> >> >> On Thu, Jul 18, 2013 at 3:21 PM, Tom Clune <Tho...@na...> wrote: >> Hi Jesse, >> >> For ifort, you probably need a more recent compiler, though gfortran 4.7.8 ought to have worked. We recommend ifort 13.x and gfortran 4.7/4.8 >> >> If you can provide a log of what goes wrong with cases 1 and 2 we'll help trouble shoot. (Case 2 simply will not work.) One obvious thing that others have pointed out (and should go in the documentation once we push it to the repository): you need to define the variable F90 to be ifort/gfortran. Or you can use F90_VENDOR=Intel/GNU. >> >> Cheers, >> >> - Tom >> >> >> On Jul 18, 2013, at 3:49 PM, Jesse Lopez <lo...@st...> wrote: >> >>> Hello all, >>> >>> I've just cloned pFUnit from the master branch on the sourceforge repository and I've been unable to successfully compile the code with a variety of compilers/platforms. >>> >>> 1. ifort 12.1.0 on linux >>> https://gist.github.com/yosoyjay/6032388 >>> >>> 2. gfortran 4.4.6 on linux (I know, it's ancient) >>> https://gist.github.com/yosoyjay/6032415 >>> >>> 3. gfortran 4.7.8 on os x 10.7 >>> https://gist.github.com/yosoyjay/6032437 >>> >>> Any hints on how to get it compiled and working would be much appreciated. >>> >>> Thanks, >>> -Jesse >>> ------------------------------------------------------------------------------ >>> See everything from the browser to the database with AppDynamics >>> Get end-to-end visibility with application monitoring from AppDynamics >>> Isolate bottlenecks and diagnose root cause in seconds. >>> Start your free trial of AppDynamics Pro today! >>> http://pubads.g.doubleclick.net/gampad/clk?id=48808831&iu=/4140/ostg.clktrk_______________________________________________ >>> Pfunit-support mailing list >>> Pfu...@li... >>> https://lists.sourceforge.net/lists/listinfo/pfunit-support >> >> Thomas Clune, Ph. D. <Tho...@na...> >> Chief, Software Systems Support Office Code 610.3 >> NASA GSFC 301-286-4635 >> MS 610.8 B33-C128 <http://ssso.gsfc.nasa.gov> >> Greenbelt, MD 20771 >> >> >> >> >> >> > > Thomas Clune, Ph. D. <Tho...@na...> > Chief, Software Systems Support Office Code 610.3 > NASA GSFC 301-286-4635 > MS 610.8 B33-C128 <http://ssso.gsfc.nasa.gov> > Greenbelt, MD 20771 > > > > > > Thomas Clune, Ph. D. <Tho...@na...> Chief, Software Systems Support Office Code 610.3 NASA GSFC 301-286-4635 MS 610.8 B33-C128 <http://ssso.gsfc.nasa.gov> Greenbelt, MD 20771 |
|
From: Tom C. <tho...@na...> - 2013-07-18 21:24:38
|
Oh - I did not read far enough. I now see you've already shown that mpif90 is pointing to the right compiler. I still think that somehow you are using the "wrong" mpif90, but I cannot explain why "which" is saying that you are not. Probably GNUmake is using slightly different environment variables. Please try export MPIF90=/opt/intel/impi/4.1.0.024/intel64/bin/mpif90 and see if that fixes the problem. I know that years ago I had a lot of trouble flushing the builtin mpif90 out of my path, but no longer remember the steps I took. - Tom On Jul 18, 2013, at 5:13 PM, Jesse Lopez <lo...@st...> wrote: > Hi Tom, > > I'm explicitly providing the compiler and vendor. > make all F90=ifort F90_VENDOR=Intel [good] > make tests F90=ifort F90_VENDOR=Intel [good] > make tests MPI=YES F90=ifort F90_VENDOR=Intel [bad] > > mpif90 -c -assume realloc_lhs -g -O0 -traceback -check uninit -check bounds -check stack -check uninit -DSTRINGIFY_OPERATOR -I../include -DUSE_MPI -DIntel -DLinux -I/state/partition1/home/lopezj/pFUnit/include -DLONG_PTR -DSTRINGIFY_OPERATOR -DIntel -o MpiContext.o MpiContext.F90 > gfortran: realloc_lhs: No such file or directory > gfortran: uninit: No such file or directory > .... > > I'm a bit perplexed by the gfortran complaints. I also tried > make --debug=v MPI=YES F90=ifort F90_VENDOR=Intel > > But it doesn't reveal much beyond what I already know. It's using Intel.mk and it dies when with the > command above. > > which -a mpif90 > /opt/intel/impi/4.1.0.024/intel64/bin/mpif90 > > Any ideas? > > Thanks, > -Jesse > > > On Thu, Jul 18, 2013 at 3:46 PM, Tom Clune <tho...@na...> wrote: > Hi Jesse, > > The error log suggests that you are somehow getting Ifort compile flags with gfortran. So either you don't have F90=gfortran for that case or you have F90_VENDOR set to Intel. > > Seems unrelated to MPI. > > - Tom > > On Jul 18, 2013, at 4:41 PM, Jesse Lopez <lo...@st...> wrote: > >> Hi Tom, >> >> Thanks for the prompt response. I had in fact specified the F90*, so I was good there. >> I did successfully compile with ifort 13.0 as you suggested. But, I ran into some trouble >> with 'make tests MPI=YES' with complaints from gfortran! >> >> https://gist.github.com/yosoyjay/6032859 >> >> Any idea about the gfortran 4.7.3 error on OS X? >> >> Thanks again for your help. >> >> -Jesse >> >> >> On Thu, Jul 18, 2013 at 3:21 PM, Tom Clune <Tho...@na...> wrote: >> Hi Jesse, >> >> For ifort, you probably need a more recent compiler, though gfortran 4.7.8 ought to have worked. We recommend ifort 13.x and gfortran 4.7/4.8 >> >> If you can provide a log of what goes wrong with cases 1 and 2 we'll help trouble shoot. (Case 2 simply will not work.) One obvious thing that others have pointed out (and should go in the documentation once we push it to the repository): you need to define the variable F90 to be ifort/gfortran. Or you can use F90_VENDOR=Intel/GNU. >> >> Cheers, >> >> - Tom >> >> >> On Jul 18, 2013, at 3:49 PM, Jesse Lopez <lo...@st...> wrote: >> >>> Hello all, >>> >>> I've just cloned pFUnit from the master branch on the sourceforge repository and I've been unable to successfully compile the code with a variety of compilers/platforms. >>> >>> 1. ifort 12.1.0 on linux >>> https://gist.github.com/yosoyjay/6032388 >>> >>> 2. gfortran 4.4.6 on linux (I know, it's ancient) >>> https://gist.github.com/yosoyjay/6032415 >>> >>> 3. gfortran 4.7.8 on os x 10.7 >>> https://gist.github.com/yosoyjay/6032437 >>> >>> Any hints on how to get it compiled and working would be much appreciated. >>> >>> Thanks, >>> -Jesse >>> ------------------------------------------------------------------------------ >>> See everything from the browser to the database with AppDynamics >>> Get end-to-end visibility with application monitoring from AppDynamics >>> Isolate bottlenecks and diagnose root cause in seconds. >>> Start your free trial of AppDynamics Pro today! >>> http://pubads.g.doubleclick.net/gampad/clk?id=48808831&iu=/4140/ostg.clktrk_______________________________________________ >>> Pfunit-support mailing list >>> Pfu...@li... >>> https://lists.sourceforge.net/lists/listinfo/pfunit-support >> >> Thomas Clune, Ph. D. <Tho...@na...> >> Chief, Software Systems Support Office Code 610.3 >> NASA GSFC 301-286-4635 >> MS 610.8 B33-C128 <http://ssso.gsfc.nasa.gov> >> Greenbelt, MD 20771 >> >> >> >> >> >> > > Thomas Clune, Ph. D. <Tho...@na...> > Chief, Software Systems Support Office Code 610.3 > NASA GSFC 301-286-4635 > MS 610.8 B33-C128 <http://ssso.gsfc.nasa.gov> > Greenbelt, MD 20771 > > > > > > Thomas Clune, Ph. D. <Tho...@na...> Chief, Software Systems Support Office Code 610.3 NASA GSFC 301-286-4635 MS 610.8 B33-C128 <http://ssso.gsfc.nasa.gov> Greenbelt, MD 20771 |
|
From: Jesse L. <lo...@st...> - 2013-07-18 22:20:48
|
Tom,
It was indeed a problem with mpif90 still referencing gfortran which is in
fact
the default behavior Intel's mpif90. If you conjure up this encantation:
export I_MPI_F90=ifort
make F90=ifort F90_VENDOR=Intel MPI=YES [good!!!]
So, I now have pFUnit working with the Intel compilers on Linux.
I still have the a problem on OS X using gfortran 4.7/4.8:
gfortran -c -g -O0 -fbacktrace -fbounds-check -fcheck=mem
-DSTRINGIFY_SIMPLE -I../include -DGNU -DDarwin
-I/Users/jesse/src/pFunit-git/include -o SurrogateTestCase.o
SurrogateTestCase.F90
SurrogateTestCase.F90:32.57:
class (SurrogateTestCase), intent(inout) :: this
1
Error: Deferred-length character component 'getname' at (1) is not yet
supported
make[1]: *** [SurrogateTestCase.o] Error 1
make: *** [all] Error 2
I have no idea how to track down this problem. I am using the gfortan
versions from MacPorts if that helps at all.
Thanks again for all of your help!
-Jesse
On Thu, Jul 18, 2013 at 4:24 PM, Tom Clune <tho...@na...> wrote:
> Oh - I did not read far enough. I now see you've already shown that
> mpif90 is pointing to the right compiler.
>
> I still think that somehow you are using the "wrong" mpif90, but I cannot
> explain why "which" is saying that you are not. Probably GNUmake is using
> slightly different environment variables.
>
> Please try
>
> export MPIF90=/opt/intel/impi/4.1.0.024/intel64/bin/mpif90
>
> and see if that fixes the problem. I know that years ago I had a lot of
> trouble flushing the builtin mpif90 out of my path, but no longer remember
> the steps I took.
>
> - Tom
>
>
> On Jul 18, 2013, at 5:13 PM, Jesse Lopez <lo...@st...> wrote:
>
> Hi Tom,
>
> I'm explicitly providing the compiler and vendor.
> make all F90=ifort F90_VENDOR=Intel [good]
> make tests F90=ifort F90_VENDOR=Intel [good]
> make tests MPI=YES F90=ifort F90_VENDOR=Intel [bad]
>
> mpif90 -c -assume realloc_lhs -g -O0 -traceback -check uninit -check
> bounds -check stack -check uninit -DSTRINGIFY_OPERATOR -I../include
> -DUSE_MPI -DIntel -DLinux -I/state/partition1/home/lopezj/pFUnit/include
> -DLONG_PTR -DSTRINGIFY_OPERATOR -DIntel -o MpiContext.o MpiContext.F90
> gfortran: realloc_lhs: No such file or directory
> gfortran: uninit: No such file or directory
> ....
>
> I'm a bit perplexed by the gfortran complaints. I also tried
> make --debug=v MPI=YES F90=ifort F90_VENDOR=Intel
>
> But it doesn't reveal much beyond what I already know. It's using
> Intel.mk and it dies when with the
> command above.
>
> which -a mpif90
> /opt/intel/impi/4.1.0.024/intel64/bin/mpif90
>
> Any ideas?
>
> Thanks,
> -Jesse
>
>
> On Thu, Jul 18, 2013 at 3:46 PM, Tom Clune <tho...@na...>wrote:
>
>> Hi Jesse,
>>
>> The error log suggests that you are somehow getting Ifort compile flags
>> with gfortran. So either you don't have F90=gfortran for that case or you
>> have F90_VENDOR set to Intel.
>>
>> Seems unrelated to MPI.
>>
>> - Tom
>>
>> On Jul 18, 2013, at 4:41 PM, Jesse Lopez <lo...@st...> wrote:
>>
>> Hi Tom,
>>
>> Thanks for the prompt response. I had in fact specified the F90*, so I
>> was good there.
>> I did successfully compile with ifort 13.0 as you suggested. But, I ran
>> into some trouble
>> with 'make tests MPI=YES' with complaints from gfortran!
>>
>> https://gist.github.com/yosoyjay/6032859
>>
>> Any idea about the gfortran 4.7.3 error on OS X?
>>
>> Thanks again for your help.
>>
>> -Jesse
>>
>>
>> On Thu, Jul 18, 2013 at 3:21 PM, Tom Clune <Tho...@na...>wrote:
>>
>>> Hi Jesse,
>>>
>>> For ifort, you probably need a more recent compiler, though gfortran
>>> 4.7.8 ought to have worked. We recommend ifort 13.x and gfortran 4.7/4.8
>>>
>>> If you can provide a log of what goes wrong with cases 1 and 2 we'll
>>> help trouble shoot. (Case 2 simply will not work.) One obvious thing that
>>> others have pointed out (and should go in the documentation once we push it
>>> to the repository): you need to define the variable F90 to be
>>> ifort/gfortran. Or you can use F90_VENDOR=Intel/GNU.
>>>
>>> Cheers,
>>>
>>> - Tom
>>>
>>>
>>> On Jul 18, 2013, at 3:49 PM, Jesse Lopez <lo...@st...> wrote:
>>>
>>> Hello all,
>>>
>>> I've just cloned pFUnit from the master branch on the sourceforge
>>> repository and I've been unable to successfully compile the code with a
>>> variety of compilers/platforms.
>>>
>>> 1. ifort 12.1.0 on linux
>>> https://gist.github.com/yosoyjay/6032388
>>>
>>> 2. gfortran 4.4.6 on linux (I know, it's ancient)
>>> https://gist.github.com/yosoyjay/6032415
>>>
>>> 3. gfortran 4.7.8 on os x 10.7
>>> https://gist.github.com/yosoyjay/6032437
>>>
>>> Any hints on how to get it compiled and working would be much
>>> appreciated.
>>>
>>> Thanks,
>>> -Jesse
>>>
>>> ------------------------------------------------------------------------------
>>> See everything from the browser to the database with AppDynamics
>>> Get end-to-end visibility with application monitoring from AppDynamics
>>> Isolate bottlenecks and diagnose root cause in seconds.
>>> Start your free trial of AppDynamics Pro today!
>>>
>>> http://pubads.g.doubleclick.net/gampad/clk?id=48808831&iu=/4140/ostg.clktrk_______________________________________________
>>> Pfunit-support mailing list
>>> Pfu...@li...
>>> https://lists.sourceforge.net/lists/listinfo/pfunit-support
>>>
>>>
>>> Thomas Clune, Ph. D. <Tho...@na...>
>>> Chief, Software Systems Support Office Code 610.3
>>> NASA GSFC 301-286-4635
>>> MS 610.8 B33-C128 <http://ssso.gsfc.nasa.gov>
>>> Greenbelt, MD 20771
>>>
>>>
>>>
>>>
>>>
>>>
>>
>> Thomas Clune, Ph. D. <Tho...@na...>
>> Chief, Software Systems Support Office Code 610.3
>> NASA GSFC 301-286-4635
>> MS 610.8 B33-C128 <http://ssso.gsfc.nasa.gov>
>> Greenbelt, MD 20771
>>
>>
>>
>>
>>
>>
>
> Thomas Clune, Ph. D. <Tho...@na...>
> Chief, Software Systems Support Office Code 610.3
> NASA GSFC 301-286-4635
> MS 610.8 B33-C128 <http://ssso.gsfc.nasa.gov>
> Greenbelt, MD 20771
>
>
>
>
>
>
|
|
From: Tom C. <tho...@na...> - 2013-07-19 00:26:14
|
Jess, On Jul 18, 2013, at 6:20 PM, Jesse Lopez <lo...@st...> wrote: > Tom, > > It was indeed a problem with mpif90 still referencing gfortran which is in fact > the default behavior Intel's mpif90. If you conjure up this encantation: > export I_MPI_F90=ifort > make F90=ifort F90_VENDOR=Intel MPI=YES [good!!!] Great. This is the first I've heard of someone installing Intel MPI on a Mac. It did not even occur to me that it would not default to their compiler. Glad you found the solution. > > So, I now have pFUnit working with the Intel compilers on Linux. > > I still have the a problem on OS X using gfortran 4.7/4.8: > > gfortran -c -g -O0 -fbacktrace -fbounds-check -fcheck=mem -DSTRINGIFY_SIMPLE -I../include -DGNU -DDarwin -I/Users/jesse/src/pFunit-git/include -o SurrogateTestCase.o SurrogateTestCase.F90 > SurrogateTestCase.F90:32.57: > > class (SurrogateTestCase), intent(inout) :: this > 1 > Error: Deferred-length character component 'getname' at (1) is not yet supported > make[1]: *** [SurrogateTestCase.o] Error 1 > make: *** [all] Error 2 > > I have no idea how to track down this problem. I am using the gfortan versions from MacPorts if that helps at all. Hmmm. We did have to put #ifdef around some of these gfortran limitations, but the current version is building and passing all the tests on my Mac. Checking 4.7 … I do reproduce your error with 4.7, so apparently my memory was wrong that 4.7 would still work. Sorry about that. I've not been running the regression tests on the older compilers. So now the question is whether or not the specific sub-version of 4.8 matters. What I have installed is: GNU Fortran (MacPorts gcc48 4.8.1_0). Using OpenMPI, but I don't think the MPI is what matters. > > Thanks again for all of your help! No problem. I'm really hoping that we can establish a sustainable community of people using unit-testing frameworks in Fortran. Keep asking until it works the way you want it to. Cheers, - Tom > > -Jesse > > > On Thu, Jul 18, 2013 at 4:24 PM, Tom Clune <tho...@na...> wrote: > Oh - I did not read far enough. I now see you've already shown that mpif90 is pointing to the right compiler. > > I still think that somehow you are using the "wrong" mpif90, but I cannot explain why "which" is saying that you are not. Probably GNUmake is using slightly different environment variables. > > Please try > > export MPIF90=/opt/intel/impi/4.1.0.024/intel64/bin/mpif90 > > and see if that fixes the problem. I know that years ago I had a lot of trouble flushing the builtin mpif90 out of my path, but no longer remember the steps I took. > > - Tom > > > On Jul 18, 2013, at 5:13 PM, Jesse Lopez <lo...@st...> wrote: > >> Hi Tom, >> >> I'm explicitly providing the compiler and vendor. >> make all F90=ifort F90_VENDOR=Intel [good] >> make tests F90=ifort F90_VENDOR=Intel [good] >> make tests MPI=YES F90=ifort F90_VENDOR=Intel [bad] >> >> mpif90 -c -assume realloc_lhs -g -O0 -traceback -check uninit -check bounds -check stack -check uninit -DSTRINGIFY_OPERATOR -I../include -DUSE_MPI -DIntel -DLinux -I/state/partition1/home/lopezj/pFUnit/include -DLONG_PTR -DSTRINGIFY_OPERATOR -DIntel -o MpiContext.o MpiContext.F90 >> gfortran: realloc_lhs: No such file or directory >> gfortran: uninit: No such file or directory >> .... >> >> I'm a bit perplexed by the gfortran complaints. I also tried >> make --debug=v MPI=YES F90=ifort F90_VENDOR=Intel >> >> But it doesn't reveal much beyond what I already know. It's using Intel.mk and it dies when with the >> command above. >> >> which -a mpif90 >> /opt/intel/impi/4.1.0.024/intel64/bin/mpif90 >> >> Any ideas? >> >> Thanks, >> -Jesse >> >> >> On Thu, Jul 18, 2013 at 3:46 PM, Tom Clune <tho...@na...> wrote: >> Hi Jesse, >> >> The error log suggests that you are somehow getting Ifort compile flags with gfortran. So either you don't have F90=gfortran for that case or you have F90_VENDOR set to Intel. >> >> Seems unrelated to MPI. >> >> - Tom >> >> On Jul 18, 2013, at 4:41 PM, Jesse Lopez <lo...@st...> wrote: >> >>> Hi Tom, >>> >>> Thanks for the prompt response. I had in fact specified the F90*, so I was good there. >>> I did successfully compile with ifort 13.0 as you suggested. But, I ran into some trouble >>> with 'make tests MPI=YES' with complaints from gfortran! >>> >>> https://gist.github.com/yosoyjay/6032859 >>> >>> Any idea about the gfortran 4.7.3 error on OS X? >>> >>> Thanks again for your help. >>> >>> -Jesse >>> >>> >>> On Thu, Jul 18, 2013 at 3:21 PM, Tom Clune <Tho...@na...> wrote: >>> Hi Jesse, >>> >>> For ifort, you probably need a more recent compiler, though gfortran 4.7.8 ought to have worked. We recommend ifort 13.x and gfortran 4.7/4.8 >>> >>> If you can provide a log of what goes wrong with cases 1 and 2 we'll help trouble shoot. (Case 2 simply will not work.) One obvious thing that others have pointed out (and should go in the documentation once we push it to the repository): you need to define the variable F90 to be ifort/gfortran. Or you can use F90_VENDOR=Intel/GNU. >>> >>> Cheers, >>> >>> - Tom >>> >>> >>> On Jul 18, 2013, at 3:49 PM, Jesse Lopez <lo...@st...> wrote: >>> >>>> Hello all, >>>> >>>> I've just cloned pFUnit from the master branch on the sourceforge repository and I've been unable to successfully compile the code with a variety of compilers/platforms. >>>> >>>> 1. ifort 12.1.0 on linux >>>> https://gist.github.com/yosoyjay/6032388 >>>> >>>> 2. gfortran 4.4.6 on linux (I know, it's ancient) >>>> https://gist.github.com/yosoyjay/6032415 >>>> >>>> 3. gfortran 4.7.8 on os x 10.7 >>>> https://gist.github.com/yosoyjay/6032437 >>>> >>>> Any hints on how to get it compiled and working would be much appreciated. >>>> >>>> Thanks, >>>> -Jesse >>>> ------------------------------------------------------------------------------ >>>> See everything from the browser to the database with AppDynamics >>>> Get end-to-end visibility with application monitoring from AppDynamics >>>> Isolate bottlenecks and diagnose root cause in seconds. >>>> Start your free trial of AppDynamics Pro today! >>>> http://pubads.g.doubleclick.net/gampad/clk?id=48808831&iu=/4140/ostg.clktrk_______________________________________________ >>>> Pfunit-support mailing list >>>> Pfu...@li... >>>> https://lists.sourceforge.net/lists/listinfo/pfunit-support >>> >>> Thomas Clune, Ph. D. <Tho...@na...> >>> Chief, Software Systems Support Office Code 610.3 >>> NASA GSFC 301-286-4635 >>> MS 610.8 B33-C128 <http://ssso.gsfc.nasa.gov> >>> Greenbelt, MD 20771 >>> >>> >>> >>> >>> >>> >> >> Thomas Clune, Ph. D. <Tho...@na...> >> Chief, Software Systems Support Office Code 610.3 >> NASA GSFC 301-286-4635 >> MS 610.8 B33-C128 <http://ssso.gsfc.nasa.gov> >> Greenbelt, MD 20771 >> >> >> >> >> >> > > Thomas Clune, Ph. D. <Tho...@na...> > Chief, Software Systems Support Office Code 610.3 > NASA GSFC 301-286-4635 > MS 610.8 B33-C128 <http://ssso.gsfc.nasa.gov> > Greenbelt, MD 20771 > > > > > > Thomas Clune, Ph. D. <Tho...@na...> Chief, Software Systems Support Office Code 610.3 NASA GSFC 301-286-4635 MS 610.8 B33-C128 <http://ssso.gsfc.nasa.gov> Greenbelt, MD 20771 |
|
From: Jesse L. <lo...@st...> - 2013-07-19 02:21:41
|
On Thu, Jul 18, 2013 at 7:25 PM, Tom Clune <tho...@na...> wrote: > > Hmmm. We did have to put #ifdef around some of these gfortran > limitations, but the current version is building and passing all the tests > on my Mac. Checking 4.7 … I do reproduce your error with 4.7, so > apparently my memory was wrong that 4.7 would still work. Sorry about > that. I've not been running the regression tests on the older compilers. > > So now the question is whether or not the specific sub-version of 4.8 > matters. What I have installed is: GNU Fortran (MacPorts gcc48 4.8.1_0). > Using OpenMPI, but I don't think the MPI is what matters. > Good news! It's working. It turns out I didn't switch gcc versions when I installed gcc48. I missed the command: sudo port select --set gcc mp-gcc48 Once I had the latest and greatest gfortan, the make and tests completed without any problems. Now I'm in business and I can get busy doing some refactoring. Thanks, -Jesse |
|
From: Tom C. <tho...@na...> - 2013-07-19 12:41:56
|
Glad to hear it. Cheers, - Tom On Jul 18, 2013, at 10:21 PM, Jesse Lopez <lo...@st...> wrote: > > On Thu, Jul 18, 2013 at 7:25 PM, Tom Clune <tho...@na...> wrote: > > Hmmm. We did have to put #ifdef around some of these gfortran limitations, but the current version is building and passing all the tests on my Mac. Checking 4.7 … I do reproduce your error with 4.7, so apparently my memory was wrong that 4.7 would still work. Sorry about that. I've not been running the regression tests on the older compilers. > > So now the question is whether or not the specific sub-version of 4.8 matters. What I have installed is: GNU Fortran (MacPorts gcc48 4.8.1_0). Using OpenMPI, but I don't think the MPI is what matters. > > Good news! It's working. It turns out I didn't switch gcc versions when I installed gcc48. I missed the command: > > sudo port select --set gcc mp-gcc48 > > Once I had the latest and greatest gfortan, the make and tests completed without any problems. Now I'm in business and I can get busy doing some refactoring. > > Thanks, > -Jesse > Thomas Clune, Ph. D. <Tho...@na...> Chief, Software Systems Support Office Code 610.3 NASA GSFC 301-286-4635 MS 610.8 B33-C128 <http://ssso.gsfc.nasa.gov> Greenbelt, MD 20771 |
Thanks for helping keep SourceForge clean.
X