Re: [Starpu-devel] [daxpy with StarPU-MPI]

[Maxim Abalenkov <maxim.abalenkov@gmail.com>]

Hello Nathalie,

False alarm. I have got it working on a Mac too. There was an error related to saving the vector segments into files.

Maybe, I can contribute this example to the StarPU-MPI examples collection in the Git repository?

—

Best wishes,

Maxim

Maxim Abalenkov \\ maxim.abalenkov@gmail.com
+44 7 486 486 505 \\ http://mabalenk.gitlab.io

On 17 Oct 2018, at 15:51, Maxim Abalenkov <maxim.abalenkov@gmail.com> wrote:

Dear Nathalie,

Thank you very much for your help. I have corrected the mistakes you noticed. The code runs and computes the result correctly on Linux. However, it still crashes with the SIGABRT error on Mac. Please find the debug log attached. Unfortunately, I can’t get a more meaningful output from it.

—

Best wishes,

Maxim

<daxpy_mpi.c>

<daxpy_mpi.h>

<gdb.log>

<Makefile>

<tst_daxpy_mpi.c>

Maxim Abalenkov \\ maxim.abalenkov@gmail.com
+44 7 486 486 505 \\ http://mabalenk.gitlab.io

On 17 Oct 2018, at 14:09, Nathalie Furmento <nathalie.furmento@labri.fr> wrote:

Here some mistakes i found in your code.

* when allocating hx and hy, you should specify sizeof(starpu_data_handle_t) and not sizeof(starpu_data_handle_t*)

// allocate data handles
starpu_data_handle_t *hx = (starpu_data_handle_t *) malloc((size_t)NT*sizeof(starpu_data_handle_t));
starpu_data_handle_t *hy = (starpu_data_handle_t *) malloc((size_t)NT*sizeof(starpu_data_handle_t));

* you need to check if hx[i] and hy[i] are valid before calling starpu_data_unregister()

// unregister vector handles
for (int i = 0; i < NT; i++)
{
if (hx[i]) starpu_data_unregister(hx[i]);
if (hy[i]) starpu_data_unregister(hy[i]);
}

Everything else seems ok, let us know if it keeps failing, and more important, a backtrace of your segfault would be really useful. We cannot guess otherwise how it fails.

Cheers,

Nathalie

On oct. 17, 12:54, Maxim Abalenkov wrote:

Dear Samuel et al.,

I hope all is well with you. Would you please be so kind to check the logic of my StarPU-MPI “daxpy” code? I’m following the Handbook code snippet on StarPU-MPI. I use two vectors x, y of length N=10 split into NT=5 segments nb=2 elements each. Running on a single processor with mpiexec -np 1 ./tst_daxpy_mpi the first three segments are calculated correctly, while the last two are unchanged. It seems there is a synchronisation issue. In case I launch the code on two CPUs it crashes with a segmentation fault. Thank you and have a good day ahead!

—