 Cédric Augonnet
|
dbbf7a2a93
Simplify the cublas helper functions by the means of the
|
%!s(int64=16) %!d(string=hai) anos |
|
Cédric Augonnet
|
c35710a810
do not hardcode the number of memory nodes anymore
|
%!s(int64=16) %!d(string=hai) anos |
|
Cédric Augonnet
|
b2a72aa358
implement the ring example with asynchronous calls
|
%!s(int64=16) %!d(string=hai) anos |
|
Cédric Augonnet
|
52c311463b
fix asynchronous request submission (aka modprobe brain)
|
%!s(int64=16) %!d(string=hai) anos |
|
Cédric Augonnet
|
744f2a8624
- Bug fix: actually use mpi_isend.c in the mpi_isend test :)
|
%!s(int64=16) %!d(string=hai) anos |
|
Cédric Augonnet
|
2b9b3e948b
implement isend, irecv, wait and test
|
%!s(int64=16) %!d(string=hai) anos |
|
Cédric Augonnet
|
42379b35dc
bug fix: the number of children was not set properly in the existing
|
%!s(int64=16) %!d(string=hai) anos |
|
Cédric Augonnet
|
e5d825e471
Clean up other filters.
|
%!s(int64=16) %!d(string=hai) anos |
|
Cédric Augonnet
|
51fe1b1897
Various cleanups in the data management code, start to outline an API to
|
%!s(int64=16) %!d(string=hai) anos |
|
Cédric Augonnet
|
bbea7f8d4b
Support the "block" interface with our mpi lib.
|
%!s(int64=16) %!d(string=hai) anos |
|
Cédric Augonnet
|
cf48d2f6c6
In spite of the MPI norm, some MPI implementation seems to require that the
|
%!s(int64=16) %!d(string=hai) anos |
|
Cédric Augonnet
|
9e9015d718
implement starpu_get_block_elemsize
|
%!s(int64=16) %!d(string=hai) anos |
|
Cédric Augonnet
|
480ee416d5
- Cleanup the code managing performance models and per model "debug" files. We
|
%!s(int64=16) %!d(string=hai) anos |
|
Cédric Augonnet
|
9711586171
fix previous commit
|
%!s(int64=16) %!d(string=hai) anos |
|
Cédric Augonnet
|
89fe8a7605
Support 2D matrices ("blas" interface)
|
%!s(int64=16) %!d(string=hai) anos |
|
Cédric Augonnet
|
fda086c7ae
Implement starpu_get_blas_elemsize
|
%!s(int64=16) %!d(string=hai) anos |
|
Cédric Augonnet
|
1470ba3508
- Provide the status returned by the MPI_Irecv function.
|
%!s(int64=16) %!d(string=hai) anos |
|
Cédric Augonnet
|
bf8bb00b8a
Start to implement starpu_mpi_handle_to_datatype.
|
%!s(int64=16) %!d(string=hai) anos |
|
Cédric Augonnet
|
22e39e050a
starpu_get_handle_interface_id returns the identifier of the interface of a
|
%!s(int64=16) %!d(string=hai) anos |
|
Cédric Augonnet
|
e51aadfda3
- The different data interfaces are now indexed between 0 and
|
%!s(int64=16) %!d(string=hai) anos |
|
Cédric Augonnet
|
869320b7a2
Small hack to avoid side effects from third party libs that also try to bind
|
%!s(int64=16) %!d(string=hai) anos |
|
Cédric Augonnet
|
5763bd2d0c
Implement starpu_get_vector_elemsize that was somehow missing.
|
%!s(int64=16) %!d(string=hai) anos |
|
Cédric Augonnet
|
2a2758ac3e
Make the "ring" example more challenging as we now have both MPI and CUDA
|
%!s(int64=16) %!d(string=hai) anos |
|
Cédric Augonnet
|
c3c1b0de02
Introduce the MPI helper lib (enabled with --with-mpi at configure time).
|
%!s(int64=16) %!d(string=hai) anos |
|
Cédric Augonnet
|
c94a373e97
Fix the starpu_sync_data_with_mem API, it is now required to call
|
%!s(int64=16) %!d(string=hai) anos |
|
Cédric Augonnet
|
fff603e54c
Make sure that the directory that should contain the performance models is
|
%!s(int64=16) %!d(string=hai) anos |
|
Cédric Augonnet
|
0363a81ed4
fix compilation in case CUDA is not used
|
%!s(int64=16) %!d(string=hai) anos |
|
Cédric Augonnet
|
0934f728e3
We don't hardcode the performance of the bus anymore, so that tool is useless
|
%!s(int64=16) %!d(string=hai) anos |
|
Cédric Augonnet
|
c054754a75
Implement bus performance sampling: this is no longer hardcoded.
|
%!s(int64=16) %!d(string=hai) anos |
|
Cédric Augonnet
|
a3be10b313
The performande models associated to the codelets are now put in
|
%!s(int64=16) %!d(string=hai) anos |
