Re: [Starpu-devel] [MACOSX] problem with examples

[Nathalie Furmento <nathalie.furmento@labri.fr>]

Hi,

The last commit (#3500) should solve your problem. It checks the maximum 
size available on the CUDA device before doing the allocation.

Regards,

Nathalie

On 08/04/2011 11:51, Nathalie Furmento wrote:
> Hi,
>
> Here we are trying to malloc 32 G on the device. It looks like there 
> is not enough memory.
>
> I am going to look at how to update this code to deal with this kind 
> of error.
>
> In the meantime, you can try to decrease the value of SIZE in the file 
> src/core/perfmodel/perfmodel_bus.c , line 43
>
> #define SIZE    (32*1024*1024*sizeof(char))
>
> Regards,
>
> Nathalie
>
> It looks like On 08/04/2011 11:44, tog wrote:
> > I am back :-)
> >
> > Here is the output with the requested options:
> >
> > obelix:basic_examples alleon$ ./hello_world
> > No performance model for the bus,
> > calibrating...[starpu][benchmark_all_gpu_devices] Benchmarking the
> > speed of the bus
> > Assertion failed: (d_buffer), function
> > measure_bandwidth_between_host_and_dev_on_cpu_with_cuda, file
> > core/perfmodel/perfmodel_bus.c, line 105.
> > Abort trap
> >
> >
> > I have updated my system so I might be able to switch to repo
> >
> > Best Regards
> > Guillaume
> >
> > obelix:basic_examples alleon$ automake --version
> > automake (GNU automake) 1.11.1
> > Copyright (C) 2009 Free Software Foundation, Inc.
> > License GPLv2+: GNU GPL version 2 or later
> > <http://gnu.org/licenses/gpl-2.0.html>
> > This is free software: you are free to change and redistribute it.
> > There is NO WARRANTY, to the extent permitted by law.
> >
> > Written by Tom Tromey<tromey@redhat.com>
> >         and Alexandre Duret-Lutz<adl@gnu.org>.
> >
> > obelix:basic_examples alleon$ autoconf --version
> > autoconf (GNU Autoconf) 2.63
> > Copyright (C) 2008 Free Software Foundation, Inc.
> > License GPLv2+: GNU GPL version 2 or later
> > <http://gnu.org/licenses/old-licenses/gpl-2.0.html>
> > This is free software: you are free to change and redistribute it.
> > There is NO WARRANTY, to the extent permitted by law.
> >
> > Written by David J. MacKenzie and Akim Demaille.
> >
> >
> > On Thu, Apr 7, 2011 at 5:07 PM, Nathalie Furmento
> > <nathalie.furmento@labri.fr>  wrote:
> > > Could you try to recompile with
> > >
> > > ./configure --enable-debug --enable-verbose
> > >
> > > Thanks,
> > >
> > > Nathalie
> > >
> > > On 07/04/2011 12:46, tog wrote:
> > > > ./configure NVCCFLAGS=-m64 is working fine
> > > >
> > > > Nevertheless all the examples I run lead to a Bus error
> > > > obelix:basic_examples alleon$ ./hello_world
> > > > No performance model for the bus, calibrating...Bus error
> > > >
> > > >
> > > > Regards
> > > > Guillaume
> > > >
> > > > On Thu, Apr 7, 2011 at 3:48 PM, Nathalie Furmento
> > > > <nathalie.furmento@labri.fr>    wrote:
> > > >
> > > > > You can use the environment variable NVCCFLAGS to pass specific 
> > > > > options
> > > > > to
> > > > > nvcc.
> > > > >
> > > > > Regards,
> > > > >
> > > > > Nathalie
> > > > >
> > > > > On 07/04/2011 12:11, tog wrote:
> > > > >
> > > > > > Ok, this is better. We are now in the problem I was anticipating 
> > > > > > earlier
> > > > > > ...
> > > > > > mpicc is generating 64 bits while nvcc is generationg 32 bits
> > > > > >
> > > > > > Is there a way to trigger nvcc compiler flags ?
> > > > > >
> > > > > > My automake version:
> > > > > > oggy:oce alleon$ automake --version
> > > > > > automake (GNU automake) 1.9.6
> > > > > > Written by Tom Tromey<tromey@redhat.com>.
> > > > > >
> > > > > > Best Regards
> > > > > > Guillaume
> > > > > >
> > > > > > ....
> > > > > >    CCLD   tests/mpi_irecv
> > > > > >    CC     mpi_isend_detached.o
> > > > > >    CCLD   tests/mpi_isend_detached
> > > > > >    CC     mpi_irecv_detached.o
> > > > > >    CCLD   tests/mpi_irecv_detached
> > > > > >    CC     mpi_detached_tag.o
> > > > > >    CCLD   tests/mpi_detached_tag
> > > > > >    CC     ring.o
> > > > > > /usr/local/cuda/bin/nvcc tests/ring_kernel.cu -c -o
> > > > > > tests/ring_kernel.o --compiler-options -fno-strict-aliasing
> > > > > > -I/usr/local/cuda/include -I../include/ -I../include  -I../include/
> > > > > > -I../include/
> > > > > >    CCLD   tests/ring
> > > > > > ld: warning: in tests/ring_kernel.o, file was built for i386 
> > > > > > which is
> > > > > > not the architecture being linked (x86_64)
> > > > > > Undefined symbols:
> > > > > >    "_increment_cuda", referenced from:
> > > > > >        _increment_cl in ring.o
> > > > > > ld: symbol(s) not found
> > > > > > collect2: ld returned 1 exit status
> > > > > > make[2]: *** [tests/ring] Error 1
> > > > > > make[1]: *** [all] Error 2
> > > > > > make: *** [all-recursive] Error 1
>
>
> _______________________________________________
> Starpu-devel mailing list
> Starpu-devel@lists.gforge.inria.fr
> http://lists.gforge.inria.fr/cgi-bin/mailman/listinfo/starpu-devel