Actually, the MXM component is an MTL, not a PML, so Jeff's option
wouldn't work.

I'm not sure this is any better than saying "--mca pml cm" versus "--mca
pml ob1", other than there's at least some rational names involved.
However, it does seem awful silly to thread the enabled through the
module; just check it and return not available.

Brian

On 1/30/12 6:06 AM, "Jeff Squyres" <jsquy...@cisco.com> wrote:

>Mellanox -- 
>
>Isn't this redundant with the "pml" MCA parameter?  For example, you can
>disable MXM via:
>
>    mpirun --mca pml ^mxm ...
>
>(I said the same thing on #2988, the CMR to move this commit to v1.5.5)
>
>
>On Jan 30, 2012, at 6:14 AM, mi...@osl.iu.edu wrote:
>
>> Author: miked
>> Date: 2012-01-30 06:14:20 EST (Mon, 30 Jan 2012)
>> New Revision: 25813
>> URL: https://svn.open-mpi.org/trac/ompi/changeset/25813
>> 
>> Log:
>> add mca param to enable/disable mxm
>> Text files modified:
>>   trunk/ompi/mca/mtl/mxm/mtl_mxm_component.c |    11 +++++++++++
>>                 
>>   trunk/ompi/mca/mtl/mxm/mtl_mxm_types.h     |     1 +
>>                 
>>   2 files changed, 12 insertions(+), 0 deletions(-)
>> 
>> Modified: trunk/ompi/mca/mtl/mxm/mtl_mxm_component.c
>> 
>>=========================================================================
>>=====
>> --- trunk/ompi/mca/mtl/mxm/mtl_mxm_component.c    (original)
>> +++ trunk/ompi/mca/mtl/mxm/mtl_mxm_component.c    2012-01-30 06:14:20
>>EST (Mon, 30 Jan 2012)
>> @@ -72,6 +72,12 @@
>>                            0,
>>                            &ompi_mtl_mxm.verbose);
>> 
>> +    mca_base_param_reg_int(&mca_mtl_mxm_component.super.mtl_version,
>>"enable",
>> +                           "Enable/Disable MXM component",
>> +                           false, false,
>> +                           1,
>> +                           &ompi_mtl_mxm.enabled);
>> +
>>     return OMPI_SUCCESS;
>> }
>> 
>> @@ -82,6 +88,11 @@
>>     mca_mtl_mxm_output = opal_output_open(NULL);
>>     opal_output_set_verbosity(mca_mtl_mxm_output, ompi_mtl_mxm.verbose);
>> 
>> +    if (0 == ompi_mtl_mxm.enabled) {
>> +        MXM_VERBOSE(1, "MTL/MXM was disabled on user request");
>> +        return OPAL_ERR_NOT_AVAILABLE;
>> +    }
>> +
>>     mxm_fill_context_opts(&ompi_mtl_mxm.mxm_opts);
>>     err = mxm_init(&ompi_mtl_mxm.mxm_opts, &ompi_mtl_mxm.mxm_context);
>>     if (MXM_OK != err) {
>> 
>> Modified: trunk/ompi/mca/mtl/mxm/mtl_mxm_types.h
>> 
>>=========================================================================
>>=====
>> --- trunk/ompi/mca/mtl/mxm/mtl_mxm_types.h    (original)
>> +++ trunk/ompi/mca/mtl/mxm/mtl_mxm_types.h    2012-01-30 06:14:20 EST
>>(Mon, 30 Jan 2012)
>> @@ -27,6 +27,7 @@
>> typedef struct mca_mtl_mxm_module_t {
>>     mca_mtl_base_module_t super; /**< base MTL interface */
>>     int                   verbose;
>> +    int                   enabled;
>>     mxm_h                 mxm_context;
>>     mxm_context_opts_t    mxm_opts;
>>     mxm_ep_h              ep;
>> _______________________________________________
>> svn-full mailing list
>> svn-f...@open-mpi.org
>> http://www.open-mpi.org/mailman/listinfo.cgi/svn-full
>
>
>-- 
>Jeff Squyres
>jsquy...@cisco.com
>For corporate legal information go to:
>http://www.cisco.com/web/about/doing_business/legal/cri/
>
>
>_______________________________________________
>devel mailing list
>de...@open-mpi.org
>http://www.open-mpi.org/mailman/listinfo.cgi/devel
>
>


-- 
  Brian W. Barrett
  Dept. 1423: Scalable System Software
  Sandia National Laboratories






Reply via email to