diff --git a/doc/user/chapters/using.org b/doc/user/chapters/using.org
index 490a90268ad41b174f30af60e20cfc0ae2caed55..72fe8616ce086ffbb172743c0eb7f3533249238b 100644
--- a/doc/user/chapters/using.org
+++ b/doc/user/chapters/using.org
@@ -687,34 +687,34 @@
In this program we use the tile data layout from PLASMA so that the call
#+begin_example
- CHAMELEON_Desc_Create_User(&descA, NULL, ChamRealDouble,
- NB, NB, NB*NB, N, N,
- 0, 0, N, N,
- GRID_P, GRID_Q,
- chameleon_getaddr_ccrb,
- chameleon_getblkldd_ccrb,
- chameleon_getrankof_2d, NULL);
+ CHAMELEON_Desc_Create(&descA, NULL, ChamRealDouble,
+ NB, NB, NB*NB, N, N,
+ 0, 0, N, N,
+ GRID_P, GRID_Q);
#+end_example
is equivalent to the following call
-
#+begin_example
- CHAMELEON_Desc_Create(&descA, NULL, ChamRealDouble,
- NB, NB, NB*NB, N, N,
- 0, 0, N, N,
- GRID_P, GRID_Q);
+ CHAMELEON_Desc_Create_User(&descA, NULL, ChamRealDouble,
+ NB, NB, NB*NB, N, N,
+ 0, 0, N, N,
+ GRID_P, GRID_Q,
+ chameleon_getaddr_ccrb,
+ chameleon_getblkldd_ccrb,
+ chameleon_getrankof_2d, NULL);
#+end_example
- functions *chameleon_getaddr_ccrb*, *chameleon_getblkldd_ccrb*,
- *chameleon_getrankof_2d* being used in *Desc_Create*. It is interesting
- to notice that the code is almost the same as Step5. The only
- additional information to give is the way tiles are distributed
- through the third function given to *CHAMELEON_Desc_Create_User*.
- Here, because we have made experiments only with a 2-D
- block-cyclic distribution, we have parameters P and Q in the
- interface of *Desc_Create* but they have sense only for 2-D
- block-cyclic distribution and then using *chameleon_getrankof_2d*
- function. Of course it could be used with other distributions,
- being no more the parameters of a 2-D block-cyclic grid but of
- another distribution.
+ functions *chameleon_getaddr_ccrb*, *chameleon_getblkldd_ccrb*, *chameleon_getrankof_2d* being used in *Desc_Create*. It
+ is interesting to notice that the code is almost the same as Step5. The only additional information to give is the
+ way tiles are distributed through the third function given to *CHAMELEON_Desc_Create_User*. Here, because we have
+ made experiments only with a 2-D block-cyclic distribution, we have parameters P and Q in the interface of
+ *Desc_Create* but they have sense only for 2-D block-cyclic distribution and then using *chameleon_getrankof_2d*
+ function. Of course it could be used with other distributions, being no more the parameters of a 2-D block-cyclic
+ grid but of another distribution. And the last parameter ~void* get_rankof_arg~ of CHAMELEON_Desc_Create_User can be
+ used to get custom data in the *get_rankof* function.
+
+ This example shows also that the user can initialize Chameleon with a custom MPI communicator. By default Chameleon
+ is initialized for using *MPI_COMM_WORLD*. This feature can be tested with ~--scom=1~ (Chameleon will operate only on
+ even MPI ranks). Notice the use of *CHAMELEON_InitParComm( NCPU, NGPU, 1, comm );* to initialize Chameleon with the
+ custom communicator.
**** Step7
@@ -733,23 +733,20 @@
that the function used to fill the tiles is provided by the user,
and therefore this approach is much more flexible.
- The new function to understand is *CHAMELEON_dbuild_Tile*, e.g.
+ The new function to understand is *CHAMELEON_map_Tile*, e.g.
#+begin_example
- struct data_pl data_A={(double)N, 51, N};
- CHAMELEON_dbuild_Tile(ChamUpperLower, descA, (void*)&data_A,
- Cham_build_callback_plgsy);
+ struct data_pl data_A={(double)N, 51};
+ CHAMELEON_map_Tile(ChamW, ChamUpperLower, descA, Cham_map_plgsy, (void*)&data_A);
#+end_example
- The idea here is to let Chameleon fill the matrix data in a
- task-based fashion (parallel) by using a function given by the
- user. First, the user should define if all the blocks must be
- entirelly filled or just the upper/lower part with, /e.g./
- ChamUpperLower. We still relies on the same structure
- *CHAM_desc_t* which must be initialized with the proper
- parameters, by calling for example *CHAMELEON_Desc_Create*. Then, an
- opaque pointer is used to let the user give some extra data used
- by his function. The last parameter is the pointer to the user's
- function.
+ The idea here is to let Chameleon fill the matrix data in a task-based fashion (parallel) by using a function given
+ by the user. First, the user has to give the access mode to the matrix between: *ChamR, *ChamW*, *ChamRW* depending on
+ the kind of operations the callback function needs to do on the tiles. In our example here we fill the matrix with
+ random values for the first time so that we use the access mode *ChamW*. Second, the user should define if all the
+ blocks must be entirelly filled or just the upper/lower part with, /e.g./ ChamUpperLower. We still relies on the
+ same structure *CHAM_desc_t* which must be initialized with the proper parameters, by calling for example
+ *CHAMELEON_Desc_Create*. Then comes the pointer to the user's function. And finally the last parameter is an opaque
+ pointer is used to let the user give some extra data used by his function.
*** Using custom data distributions
:PROPERTIES:
diff --git a/example/lapack_to_chameleon/step6.c b/example/lapack_to_chameleon/step6.c
index 721ce14b323ea86eca7c025c779aca6ccce00a2d..633e9a64d1f17770a7d4e4aeef2f9fe954f7c0a5 100644
--- a/example/lapack_to_chameleon/step6.c
+++ b/example/lapack_to_chameleon/step6.c
@@ -28,7 +28,9 @@
* The data distribution used here is 2D block cyclic, see for example
* http://www.netlib.org/scalapack/slug/node75.html for explanation.
* The user can enter the parameters of the distribution grid at execution with
- * --p= and --q=
+ * --p=. This example shows also that the user can initialize Chameleon with a
+ * custom MPI communicator (default is MPI_COMM_WORLD), try this feature with
+ * --scom=1 (Chameleon will operate only on even MPI ranks).
*/
int main(int argc, char *argv[]) {
size_t N; // matrix order