[mvapich-discuss] benchmark compile problem

Justin Cinkelj justin.cinkelj at xlab.si
Fri Mar 25 11:25:18 EDT 2016


Yes, everything works now - I just needed to account for OpenMPI 
installed in unusual directory.

On 03/25/2016 03:58 PM, Jonathan Perkins wrote:
> Hi Justin.  Are things working well for you now?  If not I suggest
> trying to use the following method when configuring the
> OSU-Micro-Benchmarks.
>
> Example:
>          ./configure CC=/path/to/oshcc CXX=/path/to/oshcxx
>          make
>          make install
>
> If OpenMPI also has these wrapper scripts for OpenSHMEM you can be sure
> that our configure tests will detect the available support more
> accurately.
>
> If you don't set CC and/or CXX the configure script will first try to
> detect mpicc (and mpicxx) and if found will use that instead of oshcc or
> upcc.
>
> On Fri, Mar 25, 2016 at 10:13:06AM +0100, Justin Cinkelj wrote:
>> ./configure CFLAGS=-I$HOME/openmpi-bin/include
>> LDFLAGS=-L$HOME/openmpi-bin/lib LIBS="-lmpi -lopen-rte -lopen-pal"
>> helped to get to compile stage.
>>
>> On 03/25/2016 09:56 AM, Justin Cinkelj wrote:
>>> Hi
>>>
>>> I want to run microbenchmark from
>>> http://mvapich.cse.ohio-state.edu/benchmarks/ to test/compare shmem
>>> communication performance on two different platforms. And I'm using
>>> OpenMPI. During ./configure I get
>>> ...
>>> checking for mpicxx... mpicxx
>>> checking whether we are using the GNU C++ compiler... yes
>>> checking whether mpicxx accepts -g... yes
>>> ...
>>> checking for MPI_Init... no
>>> checking for MPI_Accumulate... no
>>> checking for MPI_Get_accumulate... no
>>> checking for shmem_barrier_all... no
>>> checking for upc_memput... no
>>>
>>> ./make then does noting:
>>> [xlab at mike-c7 osu-micro-benchmarks-5.2]$ make
>>> make[1]: Entering directory
>>> `/home/justin_cinkelj/devel/mikelangelo/osu-micro-benchmarks-5.2'
>>> make[1]: Nothing to be done for `all-am'.
>>> make[1]: Leaving directory
>>> `/home/justin_cinkelj/devel/mikelangelo/osu-micro-benchmarks-5.2'
>>>
>>> Is the above benchmark only for MVAPICH MPI implementation, or I'm I doing
>>> something wrong?
>> _______________________________________________
>> mvapich-discuss mailing list
>> mvapich-discuss at cse.ohio-state.edu
>> http://mailman.cse.ohio-state.edu/mailman/listinfo/mvapich-discuss



More information about the mvapich-discuss mailing list