2013-07-19 22:13:58 +00:00
|
|
|
/*
|
|
|
|
|
* Copyright (c) 2004-2008 The Trustees of Indiana University and Indiana
|
|
|
|
|
* University Research and Technology
|
|
|
|
|
* Corporation. All rights reserved.
|
|
|
|
|
* Copyright (c) 2004-2011 The University of Tennessee and The University
|
|
|
|
|
* of Tennessee Research Foundation. All rights
|
|
|
|
|
* reserved.
|
|
|
|
|
* Copyright (c) 2004-2005 High Performance Computing Center Stuttgart,
|
|
|
|
|
* University of Stuttgart. All rights reserved.
|
|
|
|
|
* Copyright (c) 2004-2005 The Regents of the University of California.
|
|
|
|
|
* All rights reserved.
|
|
|
|
|
* Copyright (c) 2006 Sandia National Laboratories. All rights
|
|
|
|
|
* reserved.
|
2017-01-05 00:57:06 +00:00
|
|
|
* Copyright (c) 2008-2017 Cisco Systems, Inc. All rights reserved.
|
2013-07-19 22:13:58 +00:00
|
|
|
* Copyright (c) 2012 Los Alamos National Security, LLC. All rights
|
2014-07-30 20:56:15 +00:00
|
|
|
* reserved.
|
2013-07-19 22:13:58 +00:00
|
|
|
* $COPYRIGHT$
|
|
|
|
|
*
|
|
|
|
|
* Additional copyrights may follow
|
|
|
|
|
*
|
|
|
|
|
* $HEADER$
|
|
|
|
|
*/
|
|
|
|
|
|
George did the work and deserves all the credit for it. Ralph did the merge, and deserves whatever blame results from errors in it :-)
WHAT: Open our low-level communication infrastructure by moving all necessary components (btl/rcache/allocator/mpool) down in OPAL
All the components required for inter-process communications are currently deeply integrated in the OMPI layer. Several groups/institutions have express interest in having a more generic communication infrastructure, without all the OMPI layer dependencies. This communication layer should be made available at a different software level, available to all layers in the Open MPI software stack. As an example, our ORTE layer could replace the current OOB and instead use the BTL directly, gaining access to more reactive network interfaces than TCP. Similarly, external software libraries could take advantage of our highly optimized AM (active message) communication layer for their own purpose. UTK with support from Sandia, developped a version of Open MPI where the entire communication infrastucture has been moved down to OPAL (btl/rcache/allocator/mpool). Most of the moved components have been updated to match the new schema, with few exceptions (mainly BTLs where I have no way of compiling/testing them). Thus, the completion of this RFC is tied to being able to completing this move for all BTLs. For this we need help from the rest of the Open MPI community, especially those supporting some of the BTLs. A non-exhaustive list of BTLs that qualify here is: mx, portals4, scif, udapl, ugni, usnic.
This commit was SVN r32317.
2014-07-26 00:47:28 +00:00
|
|
|
#include "opal_config.h"
|
2013-07-19 22:13:58 +00:00
|
|
|
|
|
|
|
|
#include "opal_stdint.h"
|
|
|
|
|
|
George did the work and deserves all the credit for it. Ralph did the merge, and deserves whatever blame results from errors in it :-)
WHAT: Open our low-level communication infrastructure by moving all necessary components (btl/rcache/allocator/mpool) down in OPAL
All the components required for inter-process communications are currently deeply integrated in the OMPI layer. Several groups/institutions have express interest in having a more generic communication infrastructure, without all the OMPI layer dependencies. This communication layer should be made available at a different software level, available to all layers in the Open MPI software stack. As an example, our ORTE layer could replace the current OOB and instead use the BTL directly, gaining access to more reactive network interfaces than TCP. Similarly, external software libraries could take advantage of our highly optimized AM (active message) communication layer for their own purpose. UTK with support from Sandia, developped a version of Open MPI where the entire communication infrastucture has been moved down to OPAL (btl/rcache/allocator/mpool). Most of the moved components have been updated to match the new schema, with few exceptions (mainly BTLs where I have no way of compiling/testing them). Thus, the completion of this RFC is tied to being able to completing this move for all BTLs. For this we need help from the rest of the Open MPI community, especially those supporting some of the BTLs. A non-exhaustive list of BTLs that qualify here is: mx, portals4, scif, udapl, ugni, usnic.
This commit was SVN r32317.
2014-07-26 00:47:28 +00:00
|
|
|
#include "opal/constants.h"
|
2014-12-02 13:09:46 -08:00
|
|
|
|
|
|
|
|
#if BTL_IN_OPAL
|
George did the work and deserves all the credit for it. Ralph did the merge, and deserves whatever blame results from errors in it :-)
WHAT: Open our low-level communication infrastructure by moving all necessary components (btl/rcache/allocator/mpool) down in OPAL
All the components required for inter-process communications are currently deeply integrated in the OMPI layer. Several groups/institutions have express interest in having a more generic communication infrastructure, without all the OMPI layer dependencies. This communication layer should be made available at a different software level, available to all layers in the Open MPI software stack. As an example, our ORTE layer could replace the current OOB and instead use the BTL directly, gaining access to more reactive network interfaces than TCP. Similarly, external software libraries could take advantage of our highly optimized AM (active message) communication layer for their own purpose. UTK with support from Sandia, developped a version of Open MPI where the entire communication infrastucture has been moved down to OPAL (btl/rcache/allocator/mpool). Most of the moved components have been updated to match the new schema, with few exceptions (mainly BTLs where I have no way of compiling/testing them). Thus, the completion of this RFC is tied to being able to completing this move for all BTLs. For this we need help from the rest of the Open MPI community, especially those supporting some of the BTLs. A non-exhaustive list of BTLs that qualify here is: mx, portals4, scif, udapl, ugni, usnic.
This commit was SVN r32317.
2014-07-26 00:47:28 +00:00
|
|
|
#include "opal/mca/btl/btl.h"
|
|
|
|
|
#include "opal/mca/btl/base/base.h"
|
2014-12-02 13:09:46 -08:00
|
|
|
#else
|
|
|
|
|
#include "ompi/mca/btl/btl.h"
|
|
|
|
|
#include "ompi/mca/btl/base/base.h"
|
|
|
|
|
#endif
|
2013-07-19 22:13:58 +00:00
|
|
|
|
|
|
|
|
#include "btl_usnic.h"
|
|
|
|
|
#include "btl_usnic_frag.h"
|
|
|
|
|
#include "btl_usnic_util.h"
|
|
|
|
|
#include "btl_usnic_send.h"
|
|
|
|
|
#include "btl_usnic_ack.h"
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
/*
|
2017-01-05 00:57:06 +00:00
|
|
|
* This function is called when a send of a segment completes that is
|
|
|
|
|
* the one-and-only segment of an MPI message. Return the WQE and
|
|
|
|
|
* also return the segment if no ACK pending.
|
2013-07-19 22:13:58 +00:00
|
|
|
*/
|
|
|
|
|
void
|
George did the work and deserves all the credit for it. Ralph did the merge, and deserves whatever blame results from errors in it :-)
WHAT: Open our low-level communication infrastructure by moving all necessary components (btl/rcache/allocator/mpool) down in OPAL
All the components required for inter-process communications are currently deeply integrated in the OMPI layer. Several groups/institutions have express interest in having a more generic communication infrastructure, without all the OMPI layer dependencies. This communication layer should be made available at a different software level, available to all layers in the Open MPI software stack. As an example, our ORTE layer could replace the current OOB and instead use the BTL directly, gaining access to more reactive network interfaces than TCP. Similarly, external software libraries could take advantage of our highly optimized AM (active message) communication layer for their own purpose. UTK with support from Sandia, developped a version of Open MPI where the entire communication infrastucture has been moved down to OPAL (btl/rcache/allocator/mpool). Most of the moved components have been updated to match the new schema, with few exceptions (mainly BTLs where I have no way of compiling/testing them). Thus, the completion of this RFC is tied to being able to completing this move for all BTLs. For this we need help from the rest of the Open MPI community, especially those supporting some of the BTLs. A non-exhaustive list of BTLs that qualify here is: mx, portals4, scif, udapl, ugni, usnic.
This commit was SVN r32317.
2014-07-26 00:47:28 +00:00
|
|
|
opal_btl_usnic_frag_send_complete(opal_btl_usnic_module_t *module,
|
|
|
|
|
opal_btl_usnic_send_segment_t *sseg)
|
2013-07-19 22:13:58 +00:00
|
|
|
{
|
George did the work and deserves all the credit for it. Ralph did the merge, and deserves whatever blame results from errors in it :-)
WHAT: Open our low-level communication infrastructure by moving all necessary components (btl/rcache/allocator/mpool) down in OPAL
All the components required for inter-process communications are currently deeply integrated in the OMPI layer. Several groups/institutions have express interest in having a more generic communication infrastructure, without all the OMPI layer dependencies. This communication layer should be made available at a different software level, available to all layers in the Open MPI software stack. As an example, our ORTE layer could replace the current OOB and instead use the BTL directly, gaining access to more reactive network interfaces than TCP. Similarly, external software libraries could take advantage of our highly optimized AM (active message) communication layer for their own purpose. UTK with support from Sandia, developped a version of Open MPI where the entire communication infrastucture has been moved down to OPAL (btl/rcache/allocator/mpool). Most of the moved components have been updated to match the new schema, with few exceptions (mainly BTLs where I have no way of compiling/testing them). Thus, the completion of this RFC is tied to being able to completing this move for all BTLs. For this we need help from the rest of the Open MPI community, especially those supporting some of the BTLs. A non-exhaustive list of BTLs that qualify here is: mx, portals4, scif, udapl, ugni, usnic.
This commit was SVN r32317.
2014-07-26 00:47:28 +00:00
|
|
|
opal_btl_usnic_send_frag_t *frag;
|
2013-07-19 22:13:58 +00:00
|
|
|
|
|
|
|
|
frag = sseg->ss_parent_frag;
|
|
|
|
|
|
|
|
|
|
/* Reap a frag that was sent */
|
|
|
|
|
--sseg->ss_send_posted;
|
|
|
|
|
--frag->sf_seg_post_cnt;
|
|
|
|
|
|
|
|
|
|
/* checks for returnability made inside */
|
2017-01-05 00:58:59 +00:00
|
|
|
opal_btl_usnic_endpoint_t *ep = frag->sf_endpoint;
|
George did the work and deserves all the credit for it. Ralph did the merge, and deserves whatever blame results from errors in it :-)
WHAT: Open our low-level communication infrastructure by moving all necessary components (btl/rcache/allocator/mpool) down in OPAL
All the components required for inter-process communications are currently deeply integrated in the OMPI layer. Several groups/institutions have express interest in having a more generic communication infrastructure, without all the OMPI layer dependencies. This communication layer should be made available at a different software level, available to all layers in the Open MPI software stack. As an example, our ORTE layer could replace the current OOB and instead use the BTL directly, gaining access to more reactive network interfaces than TCP. Similarly, external software libraries could take advantage of our highly optimized AM (active message) communication layer for their own purpose. UTK with support from Sandia, developped a version of Open MPI where the entire communication infrastucture has been moved down to OPAL (btl/rcache/allocator/mpool). Most of the moved components have been updated to match the new schema, with few exceptions (mainly BTLs where I have no way of compiling/testing them). Thus, the completion of this RFC is tied to being able to completing this move for all BTLs. For this we need help from the rest of the Open MPI community, especially those supporting some of the BTLs. A non-exhaustive list of BTLs that qualify here is: mx, portals4, scif, udapl, ugni, usnic.
This commit was SVN r32317.
2014-07-26 00:47:28 +00:00
|
|
|
opal_btl_usnic_send_frag_return_cond(module, frag);
|
2013-07-19 22:13:58 +00:00
|
|
|
|
2017-01-05 00:58:59 +00:00
|
|
|
// In a short frag segment, the sseg is embedded in the frag. So
|
|
|
|
|
// there's no need to return the sseg (because we already returned
|
|
|
|
|
// the frag).
|
|
|
|
|
|
2013-07-19 22:13:58 +00:00
|
|
|
/* do bookkeeping */
|
2017-01-05 00:58:59 +00:00
|
|
|
++ep->endpoint_send_credits;
|
2013-07-19 22:13:58 +00:00
|
|
|
|
|
|
|
|
/* see if this endpoint needs to be made ready-to-send */
|
2017-01-05 00:58:59 +00:00
|
|
|
opal_btl_usnic_check_rts(ep);
|
2015-07-10 10:09:15 -07:00
|
|
|
|
|
|
|
|
++module->mod_channels[sseg->ss_channel].credits;
|
2013-07-19 22:13:58 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/*
|
2017-01-05 00:57:06 +00:00
|
|
|
* This function is called when a send segment completes that is part
|
|
|
|
|
* of a larger MPI message (ie., there may still be other chunk
|
|
|
|
|
* segments that have not yet completed sending). Return the WQE and
|
|
|
|
|
* also return the segment if no ACK pending.
|
2013-07-19 22:13:58 +00:00
|
|
|
*/
|
|
|
|
|
void
|
George did the work and deserves all the credit for it. Ralph did the merge, and deserves whatever blame results from errors in it :-)
WHAT: Open our low-level communication infrastructure by moving all necessary components (btl/rcache/allocator/mpool) down in OPAL
All the components required for inter-process communications are currently deeply integrated in the OMPI layer. Several groups/institutions have express interest in having a more generic communication infrastructure, without all the OMPI layer dependencies. This communication layer should be made available at a different software level, available to all layers in the Open MPI software stack. As an example, our ORTE layer could replace the current OOB and instead use the BTL directly, gaining access to more reactive network interfaces than TCP. Similarly, external software libraries could take advantage of our highly optimized AM (active message) communication layer for their own purpose. UTK with support from Sandia, developped a version of Open MPI where the entire communication infrastucture has been moved down to OPAL (btl/rcache/allocator/mpool). Most of the moved components have been updated to match the new schema, with few exceptions (mainly BTLs where I have no way of compiling/testing them). Thus, the completion of this RFC is tied to being able to completing this move for all BTLs. For this we need help from the rest of the Open MPI community, especially those supporting some of the BTLs. A non-exhaustive list of BTLs that qualify here is: mx, portals4, scif, udapl, ugni, usnic.
This commit was SVN r32317.
2014-07-26 00:47:28 +00:00
|
|
|
opal_btl_usnic_chunk_send_complete(opal_btl_usnic_module_t *module,
|
|
|
|
|
opal_btl_usnic_send_segment_t *sseg)
|
2013-07-19 22:13:58 +00:00
|
|
|
{
|
George did the work and deserves all the credit for it. Ralph did the merge, and deserves whatever blame results from errors in it :-)
WHAT: Open our low-level communication infrastructure by moving all necessary components (btl/rcache/allocator/mpool) down in OPAL
All the components required for inter-process communications are currently deeply integrated in the OMPI layer. Several groups/institutions have express interest in having a more generic communication infrastructure, without all the OMPI layer dependencies. This communication layer should be made available at a different software level, available to all layers in the Open MPI software stack. As an example, our ORTE layer could replace the current OOB and instead use the BTL directly, gaining access to more reactive network interfaces than TCP. Similarly, external software libraries could take advantage of our highly optimized AM (active message) communication layer for their own purpose. UTK with support from Sandia, developped a version of Open MPI where the entire communication infrastucture has been moved down to OPAL (btl/rcache/allocator/mpool). Most of the moved components have been updated to match the new schema, with few exceptions (mainly BTLs where I have no way of compiling/testing them). Thus, the completion of this RFC is tied to being able to completing this move for all BTLs. For this we need help from the rest of the Open MPI community, especially those supporting some of the BTLs. A non-exhaustive list of BTLs that qualify here is: mx, portals4, scif, udapl, ugni, usnic.
This commit was SVN r32317.
2014-07-26 00:47:28 +00:00
|
|
|
opal_btl_usnic_send_frag_t *frag;
|
2013-07-19 22:13:58 +00:00
|
|
|
|
|
|
|
|
frag = sseg->ss_parent_frag;
|
|
|
|
|
|
|
|
|
|
/* Reap a frag that was sent */
|
|
|
|
|
--sseg->ss_send_posted;
|
|
|
|
|
--frag->sf_seg_post_cnt;
|
|
|
|
|
|
|
|
|
|
if (sseg->ss_send_posted == 0 && !sseg->ss_ack_pending) {
|
George did the work and deserves all the credit for it. Ralph did the merge, and deserves whatever blame results from errors in it :-)
WHAT: Open our low-level communication infrastructure by moving all necessary components (btl/rcache/allocator/mpool) down in OPAL
All the components required for inter-process communications are currently deeply integrated in the OMPI layer. Several groups/institutions have express interest in having a more generic communication infrastructure, without all the OMPI layer dependencies. This communication layer should be made available at a different software level, available to all layers in the Open MPI software stack. As an example, our ORTE layer could replace the current OOB and instead use the BTL directly, gaining access to more reactive network interfaces than TCP. Similarly, external software libraries could take advantage of our highly optimized AM (active message) communication layer for their own purpose. UTK with support from Sandia, developped a version of Open MPI where the entire communication infrastucture has been moved down to OPAL (btl/rcache/allocator/mpool). Most of the moved components have been updated to match the new schema, with few exceptions (mainly BTLs where I have no way of compiling/testing them). Thus, the completion of this RFC is tied to being able to completing this move for all BTLs. For this we need help from the rest of the Open MPI community, especially those supporting some of the BTLs. A non-exhaustive list of BTLs that qualify here is: mx, portals4, scif, udapl, ugni, usnic.
This commit was SVN r32317.
2014-07-26 00:47:28 +00:00
|
|
|
opal_btl_usnic_release_send_segment(module, frag, sseg);
|
2013-07-19 22:13:58 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* done with whole fragment? */
|
|
|
|
|
/* checks for returnability made inside */
|
George did the work and deserves all the credit for it. Ralph did the merge, and deserves whatever blame results from errors in it :-)
WHAT: Open our low-level communication infrastructure by moving all necessary components (btl/rcache/allocator/mpool) down in OPAL
All the components required for inter-process communications are currently deeply integrated in the OMPI layer. Several groups/institutions have express interest in having a more generic communication infrastructure, without all the OMPI layer dependencies. This communication layer should be made available at a different software level, available to all layers in the Open MPI software stack. As an example, our ORTE layer could replace the current OOB and instead use the BTL directly, gaining access to more reactive network interfaces than TCP. Similarly, external software libraries could take advantage of our highly optimized AM (active message) communication layer for their own purpose. UTK with support from Sandia, developped a version of Open MPI where the entire communication infrastucture has been moved down to OPAL (btl/rcache/allocator/mpool). Most of the moved components have been updated to match the new schema, with few exceptions (mainly BTLs where I have no way of compiling/testing them). Thus, the completion of this RFC is tied to being able to completing this move for all BTLs. For this we need help from the rest of the Open MPI community, especially those supporting some of the BTLs. A non-exhaustive list of BTLs that qualify here is: mx, portals4, scif, udapl, ugni, usnic.
This commit was SVN r32317.
2014-07-26 00:47:28 +00:00
|
|
|
opal_btl_usnic_send_frag_return_cond(module, frag);
|
2013-07-19 22:13:58 +00:00
|
|
|
|
|
|
|
|
/* do bookkeeping */
|
|
|
|
|
++frag->sf_endpoint->endpoint_send_credits;
|
|
|
|
|
|
|
|
|
|
/* see if this endpoint needs to be made ready-to-send */
|
George did the work and deserves all the credit for it. Ralph did the merge, and deserves whatever blame results from errors in it :-)
WHAT: Open our low-level communication infrastructure by moving all necessary components (btl/rcache/allocator/mpool) down in OPAL
All the components required for inter-process communications are currently deeply integrated in the OMPI layer. Several groups/institutions have express interest in having a more generic communication infrastructure, without all the OMPI layer dependencies. This communication layer should be made available at a different software level, available to all layers in the Open MPI software stack. As an example, our ORTE layer could replace the current OOB and instead use the BTL directly, gaining access to more reactive network interfaces than TCP. Similarly, external software libraries could take advantage of our highly optimized AM (active message) communication layer for their own purpose. UTK with support from Sandia, developped a version of Open MPI where the entire communication infrastucture has been moved down to OPAL (btl/rcache/allocator/mpool). Most of the moved components have been updated to match the new schema, with few exceptions (mainly BTLs where I have no way of compiling/testing them). Thus, the completion of this RFC is tied to being able to completing this move for all BTLs. For this we need help from the rest of the Open MPI community, especially those supporting some of the BTLs. A non-exhaustive list of BTLs that qualify here is: mx, portals4, scif, udapl, ugni, usnic.
This commit was SVN r32317.
2014-07-26 00:47:28 +00:00
|
|
|
opal_btl_usnic_check_rts(frag->sf_endpoint);
|
2015-07-10 10:09:15 -07:00
|
|
|
|
|
|
|
|
++module->mod_channels[sseg->ss_channel].credits;
|
2013-07-19 22:13:58 +00:00
|
|
|
}
|
|
|
|
|
|
2013-11-04 22:52:03 +00:00
|
|
|
/* Responsible for completing non-fastpath parts of a put or send operation,
|
|
|
|
|
* including initializing any large frag bookkeeping fields and enqueuing the
|
|
|
|
|
* frag on the endpoint.
|
|
|
|
|
*
|
|
|
|
|
* This routine lives in this file to help prevent automatic inlining by the
|
|
|
|
|
* compiler.
|
|
|
|
|
*
|
2015-02-06 14:41:19 -08:00
|
|
|
* The "tag" only applies to sends.
|
|
|
|
|
*/
|
2013-07-19 22:13:58 +00:00
|
|
|
int
|
George did the work and deserves all the credit for it. Ralph did the merge, and deserves whatever blame results from errors in it :-)
WHAT: Open our low-level communication infrastructure by moving all necessary components (btl/rcache/allocator/mpool) down in OPAL
All the components required for inter-process communications are currently deeply integrated in the OMPI layer. Several groups/institutions have express interest in having a more generic communication infrastructure, without all the OMPI layer dependencies. This communication layer should be made available at a different software level, available to all layers in the Open MPI software stack. As an example, our ORTE layer could replace the current OOB and instead use the BTL directly, gaining access to more reactive network interfaces than TCP. Similarly, external software libraries could take advantage of our highly optimized AM (active message) communication layer for their own purpose. UTK with support from Sandia, developped a version of Open MPI where the entire communication infrastucture has been moved down to OPAL (btl/rcache/allocator/mpool). Most of the moved components have been updated to match the new schema, with few exceptions (mainly BTLs where I have no way of compiling/testing them). Thus, the completion of this RFC is tied to being able to completing this move for all BTLs. For this we need help from the rest of the Open MPI community, especially those supporting some of the BTLs. A non-exhaustive list of BTLs that qualify here is: mx, portals4, scif, udapl, ugni, usnic.
This commit was SVN r32317.
2014-07-26 00:47:28 +00:00
|
|
|
opal_btl_usnic_finish_put_or_send(
|
|
|
|
|
opal_btl_usnic_module_t *module,
|
|
|
|
|
opal_btl_usnic_endpoint_t *endpoint,
|
|
|
|
|
opal_btl_usnic_send_frag_t *frag,
|
2013-07-19 22:13:58 +00:00
|
|
|
mca_btl_base_tag_t tag)
|
|
|
|
|
{
|
|
|
|
|
int rc;
|
George did the work and deserves all the credit for it. Ralph did the merge, and deserves whatever blame results from errors in it :-)
WHAT: Open our low-level communication infrastructure by moving all necessary components (btl/rcache/allocator/mpool) down in OPAL
All the components required for inter-process communications are currently deeply integrated in the OMPI layer. Several groups/institutions have express interest in having a more generic communication infrastructure, without all the OMPI layer dependencies. This communication layer should be made available at a different software level, available to all layers in the Open MPI software stack. As an example, our ORTE layer could replace the current OOB and instead use the BTL directly, gaining access to more reactive network interfaces than TCP. Similarly, external software libraries could take advantage of our highly optimized AM (active message) communication layer for their own purpose. UTK with support from Sandia, developped a version of Open MPI where the entire communication infrastucture has been moved down to OPAL (btl/rcache/allocator/mpool). Most of the moved components have been updated to match the new schema, with few exceptions (mainly BTLs where I have no way of compiling/testing them). Thus, the completion of this RFC is tied to being able to completing this move for all BTLs. For this we need help from the rest of the Open MPI community, especially those supporting some of the BTLs. A non-exhaustive list of BTLs that qualify here is: mx, portals4, scif, udapl, ugni, usnic.
This commit was SVN r32317.
2014-07-26 00:47:28 +00:00
|
|
|
opal_btl_usnic_small_send_frag_t *sfrag;
|
|
|
|
|
opal_btl_usnic_send_segment_t *sseg;
|
2013-07-19 22:13:58 +00:00
|
|
|
|
|
|
|
|
/*
|
|
|
|
|
* If this is small, need to do the copyin now.
|
|
|
|
|
* We don't do this earlier in case we got lucky and were
|
|
|
|
|
* able to do an inline send. We did not, so here we are...
|
|
|
|
|
*/
|
George did the work and deserves all the credit for it. Ralph did the merge, and deserves whatever blame results from errors in it :-)
WHAT: Open our low-level communication infrastructure by moving all necessary components (btl/rcache/allocator/mpool) down in OPAL
All the components required for inter-process communications are currently deeply integrated in the OMPI layer. Several groups/institutions have express interest in having a more generic communication infrastructure, without all the OMPI layer dependencies. This communication layer should be made available at a different software level, available to all layers in the Open MPI software stack. As an example, our ORTE layer could replace the current OOB and instead use the BTL directly, gaining access to more reactive network interfaces than TCP. Similarly, external software libraries could take advantage of our highly optimized AM (active message) communication layer for their own purpose. UTK with support from Sandia, developped a version of Open MPI where the entire communication infrastucture has been moved down to OPAL (btl/rcache/allocator/mpool). Most of the moved components have been updated to match the new schema, with few exceptions (mainly BTLs where I have no way of compiling/testing them). Thus, the completion of this RFC is tied to being able to completing this move for all BTLs. For this we need help from the rest of the Open MPI community, especially those supporting some of the BTLs. A non-exhaustive list of BTLs that qualify here is: mx, portals4, scif, udapl, ugni, usnic.
This commit was SVN r32317.
2014-07-26 00:47:28 +00:00
|
|
|
if (frag->sf_base.uf_type == OPAL_BTL_USNIC_FRAG_SMALL_SEND) {
|
2013-07-19 22:13:58 +00:00
|
|
|
|
George did the work and deserves all the credit for it. Ralph did the merge, and deserves whatever blame results from errors in it :-)
WHAT: Open our low-level communication infrastructure by moving all necessary components (btl/rcache/allocator/mpool) down in OPAL
All the components required for inter-process communications are currently deeply integrated in the OMPI layer. Several groups/institutions have express interest in having a more generic communication infrastructure, without all the OMPI layer dependencies. This communication layer should be made available at a different software level, available to all layers in the Open MPI software stack. As an example, our ORTE layer could replace the current OOB and instead use the BTL directly, gaining access to more reactive network interfaces than TCP. Similarly, external software libraries could take advantage of our highly optimized AM (active message) communication layer for their own purpose. UTK with support from Sandia, developped a version of Open MPI where the entire communication infrastucture has been moved down to OPAL (btl/rcache/allocator/mpool). Most of the moved components have been updated to match the new schema, with few exceptions (mainly BTLs where I have no way of compiling/testing them). Thus, the completion of this RFC is tied to being able to completing this move for all BTLs. For this we need help from the rest of the Open MPI community, especially those supporting some of the BTLs. A non-exhaustive list of BTLs that qualify here is: mx, portals4, scif, udapl, ugni, usnic.
This commit was SVN r32317.
2014-07-26 00:47:28 +00:00
|
|
|
sfrag = (opal_btl_usnic_small_send_frag_t *)frag;
|
2013-07-19 22:13:58 +00:00
|
|
|
sseg = &sfrag->ssf_segment;
|
|
|
|
|
|
2013-11-04 22:52:03 +00:00
|
|
|
/* Copy in user data if there is any, collapsing 2 segments into 1.
|
|
|
|
|
* We already packed via the convertor if necessary, so we only need to
|
|
|
|
|
* handle the simple memcpy case here.
|
2013-07-19 22:13:58 +00:00
|
|
|
*/
|
2014-12-02 13:09:46 -08:00
|
|
|
if (frag->sf_base.uf_base.USNIC_SEND_LOCAL_COUNT > 1) {
|
2013-11-04 22:52:03 +00:00
|
|
|
/* no convertor */
|
2014-07-10 17:18:03 +00:00
|
|
|
assert(NULL != frag->sf_base.uf_local_seg[1].seg_addr.pval);
|
2013-07-19 22:13:58 +00:00
|
|
|
|
2014-07-10 17:18:03 +00:00
|
|
|
memcpy(((char *)(intptr_t)frag->sf_base.uf_local_seg[0].seg_addr.lval +
|
|
|
|
|
frag->sf_base.uf_local_seg[0].seg_len),
|
|
|
|
|
frag->sf_base.uf_local_seg[1].seg_addr.pval,
|
|
|
|
|
frag->sf_base.uf_local_seg[1].seg_len);
|
2013-07-19 22:13:58 +00:00
|
|
|
|
|
|
|
|
/* update 1st segment length */
|
2014-12-02 13:09:46 -08:00
|
|
|
frag->sf_base.uf_base.USNIC_SEND_LOCAL_COUNT = 1;
|
2014-07-10 17:18:03 +00:00
|
|
|
frag->sf_base.uf_local_seg[0].seg_len +=
|
|
|
|
|
frag->sf_base.uf_local_seg[1].seg_len;
|
2013-07-19 22:13:58 +00:00
|
|
|
}
|
|
|
|
|
|
2014-12-02 13:09:46 -08:00
|
|
|
sseg->ss_len = sizeof(opal_btl_usnic_btl_header_t) + frag->sf_size;
|
2013-07-19 22:13:58 +00:00
|
|
|
|
|
|
|
|
/* use standard channel */
|
|
|
|
|
sseg->ss_channel = USNIC_DATA_CHANNEL;
|
2013-09-17 07:20:44 +00:00
|
|
|
sseg->ss_base.us_btl_header->tag = tag;
|
|
|
|
|
} else {
|
George did the work and deserves all the credit for it. Ralph did the merge, and deserves whatever blame results from errors in it :-)
WHAT: Open our low-level communication infrastructure by moving all necessary components (btl/rcache/allocator/mpool) down in OPAL
All the components required for inter-process communications are currently deeply integrated in the OMPI layer. Several groups/institutions have express interest in having a more generic communication infrastructure, without all the OMPI layer dependencies. This communication layer should be made available at a different software level, available to all layers in the Open MPI software stack. As an example, our ORTE layer could replace the current OOB and instead use the BTL directly, gaining access to more reactive network interfaces than TCP. Similarly, external software libraries could take advantage of our highly optimized AM (active message) communication layer for their own purpose. UTK with support from Sandia, developped a version of Open MPI where the entire communication infrastucture has been moved down to OPAL (btl/rcache/allocator/mpool). Most of the moved components have been updated to match the new schema, with few exceptions (mainly BTLs where I have no way of compiling/testing them). Thus, the completion of this RFC is tied to being able to completing this move for all BTLs. For this we need help from the rest of the Open MPI community, especially those supporting some of the BTLs. A non-exhaustive list of BTLs that qualify here is: mx, portals4, scif, udapl, ugni, usnic.
This commit was SVN r32317.
2014-07-26 00:47:28 +00:00
|
|
|
opal_btl_usnic_large_send_frag_t *lfrag;
|
2013-09-17 07:20:44 +00:00
|
|
|
|
2013-11-04 22:52:03 +00:00
|
|
|
/* Save info about the frag so that future invocations of
|
|
|
|
|
* usnic_handle_large_send can generate segments to put on the wire. */
|
George did the work and deserves all the credit for it. Ralph did the merge, and deserves whatever blame results from errors in it :-)
WHAT: Open our low-level communication infrastructure by moving all necessary components (btl/rcache/allocator/mpool) down in OPAL
All the components required for inter-process communications are currently deeply integrated in the OMPI layer. Several groups/institutions have express interest in having a more generic communication infrastructure, without all the OMPI layer dependencies. This communication layer should be made available at a different software level, available to all layers in the Open MPI software stack. As an example, our ORTE layer could replace the current OOB and instead use the BTL directly, gaining access to more reactive network interfaces than TCP. Similarly, external software libraries could take advantage of our highly optimized AM (active message) communication layer for their own purpose. UTK with support from Sandia, developped a version of Open MPI where the entire communication infrastucture has been moved down to OPAL (btl/rcache/allocator/mpool). Most of the moved components have been updated to match the new schema, with few exceptions (mainly BTLs where I have no way of compiling/testing them). Thus, the completion of this RFC is tied to being able to completing this move for all BTLs. For this we need help from the rest of the Open MPI community, especially those supporting some of the BTLs. A non-exhaustive list of BTLs that qualify here is: mx, portals4, scif, udapl, ugni, usnic.
This commit was SVN r32317.
2014-07-26 00:47:28 +00:00
|
|
|
lfrag = (opal_btl_usnic_large_send_frag_t *)frag;
|
2013-09-17 07:20:44 +00:00
|
|
|
lfrag->lsf_tag = tag;
|
2013-09-17 07:27:39 +00:00
|
|
|
lfrag->lsf_cur_offset = 0;
|
2013-11-04 22:52:03 +00:00
|
|
|
lfrag->lsf_cur_ptr = lfrag->lsf_des_src[0].seg_addr.pval;
|
2013-09-17 07:27:39 +00:00
|
|
|
lfrag->lsf_cur_sge = 0;
|
2013-11-04 22:52:03 +00:00
|
|
|
lfrag->lsf_bytes_left_in_sge = lfrag->lsf_des_src[0].seg_len;
|
|
|
|
|
lfrag->lsf_bytes_left = frag->sf_size;
|
|
|
|
|
|
|
|
|
|
if (lfrag->lsf_pack_on_the_fly) {
|
|
|
|
|
lfrag->lsf_pack_bytes_left = frag->sf_size;
|
|
|
|
|
} else {
|
|
|
|
|
/* we pre-packed the convertor into a chain in prepare_src */
|
|
|
|
|
lfrag->lsf_pack_bytes_left = 0;
|
2013-09-17 07:27:39 +00:00
|
|
|
}
|
2013-07-19 22:13:58 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
|
|
/* queue this fragment into the send engine */
|
George did the work and deserves all the credit for it. Ralph did the merge, and deserves whatever blame results from errors in it :-)
WHAT: Open our low-level communication infrastructure by moving all necessary components (btl/rcache/allocator/mpool) down in OPAL
All the components required for inter-process communications are currently deeply integrated in the OMPI layer. Several groups/institutions have express interest in having a more generic communication infrastructure, without all the OMPI layer dependencies. This communication layer should be made available at a different software level, available to all layers in the Open MPI software stack. As an example, our ORTE layer could replace the current OOB and instead use the BTL directly, gaining access to more reactive network interfaces than TCP. Similarly, external software libraries could take advantage of our highly optimized AM (active message) communication layer for their own purpose. UTK with support from Sandia, developped a version of Open MPI where the entire communication infrastucture has been moved down to OPAL (btl/rcache/allocator/mpool). Most of the moved components have been updated to match the new schema, with few exceptions (mainly BTLs where I have no way of compiling/testing them). Thus, the completion of this RFC is tied to being able to completing this move for all BTLs. For this we need help from the rest of the Open MPI community, especially those supporting some of the BTLs. A non-exhaustive list of BTLs that qualify here is: mx, portals4, scif, udapl, ugni, usnic.
This commit was SVN r32317.
2014-07-26 00:47:28 +00:00
|
|
|
rc = opal_btl_usnic_endpoint_enqueue_frag(endpoint, frag);
|
2013-07-19 22:13:58 +00:00
|
|
|
return rc;
|
|
|
|
|
}
|