ports/openmpi
ports/openmpi
1
1
Форкнуть 0
Код Релизы Активность

If the gm_get is not supported do not even compile the long rendez-vous protocol. In same time if gm_get

Просмотр исходного кода 
is not supported remove some of the MCA parameters.

This commit was SVN r5320.
Этот коммит содержится в:
George Bosilca 2005-04-13 20:15:17 +00:00
родитель dfabcbde43
Коммит 86ac9eab99
4 изменённых файлов: 45 добавлений и 31 удалений
Показать статистику Скачать Patch файл Скачать Diff файл Показать все файлы Свернуть все файлы

2
src/mca/ptl/gm/configure.stub
Просмотреть файл

@ -160,7 +160,7 @@ AC_DEFUN([MCA_CONFIGURE_STUB],[
AC_MSG_RESULT([$MSG])
# Now test to see if the targetted GM is a broken one broken gm builds
AC_MSG_CHECKING(for broken GM 2.x RDMA gets build)
AC_MSG_CHECKING([for broken GM 2.x RDMA gets build])
AC_TRY_COMPILE([
#include <gm.h>
], [

2
src/mca/ptl/gm/src/ptl_gm.h
Просмотреть файл

@ -58,8 +58,10 @@ extern "C" {
uint32_t gm_eager_limit; /**< number of bytes before the rendez-vous protocol. If the
**< size of the message is less than this number then GM
**< use a eager protocol. */
#if OMPI_MCA_PTL_GM_HAVE_RDMA_GET
uint32_t gm_rndv_burst_limit;
uint32_t gm_rdma_frag_size; /**< maximum fragment size used to transfer data over RDMA */
#endif /* OMPI_MCA_PTL_GM_HAVE_RDMA_GET */
char* gm_port_name; /**< the name used to get the port */
struct mca_ptl_gm_proc_t* gm_local;

5
src/mca/ptl/gm/src/ptl_gm_component.c
Просмотреть файл

@ -132,12 +132,9 @@ mca_ptl_gm_component_open(void)
#if OMPI_MCA_PTL_GM_HAVE_RDMA_GET
mca_ptl_gm_component.gm_rndv_burst_limit =
mca_ptl_gm_param_register_int( "rndv_burst_limit", 512 * 1024 );
#else
mca_ptl_gm_component.gm_rndv_burst_limit =
mca_ptl_gm_param_register_int( "rndv_burst_limit", mca_ptl_gm_module.super.ptl_max_frag_size );
#endif /* OMPI_MCA_PTL_GM_HAVE_RDMA_GET */
mca_ptl_gm_component.gm_rdma_frag_size =
mca_ptl_gm_param_register_int ("rdma_frag_size", 128 * 1024);
#endif /* OMPI_MCA_PTL_GM_HAVE_RDMA_GET */
mca_ptl_gm_component.gm_free_list_num =
mca_ptl_gm_param_register_int ("free_list_num", 256);

67
src/mca/ptl/gm/src/ptl_gm_priv.c
Просмотреть файл

@ -97,7 +97,6 @@ static void mca_ptl_gm_get_callback( struct gm_port *port, void * context, gm_st
ompi_output( 0, "mca_ptl_gm_get_callback other error %d\n", status );
}
}
#endif /* OMPI_MCA_PTL_GM_HAVE_RDMA_GET */
static inline
int mca_ptl_gm_receiver_advance_pipeline( mca_ptl_gm_recv_frag_t* frag, int onlyifget )
@ -114,12 +113,10 @@ int mca_ptl_gm_receiver_advance_pipeline( mca_ptl_gm_recv_frag_t* frag, int only
get_line = &(frag->pipeline.lines[frag->pipeline.pos_transfert]);
if( (PTL_GM_PIPELINE_TRANSFERT & get_line->flags) == PTL_GM_PIPELINE_TRANSFERT ) {
peer->get_started = true;
#if OMPI_MCA_PTL_GM_HAVE_RDMA_GET
gm_get( peer->peer_ptl->gm_port, get_line->remote_memory.lval,
get_line->local_memory.pval, get_line->length,
GM_LOW_PRIORITY, peer->peer_addr.local_id, peer->peer_addr.port_id,
mca_ptl_gm_get_callback, frag );
#endif /* OMPI_MCA_PTL_GM_HAVE_RDMA_GET */
get_line->flags ^= PTL_GM_PIPELINE_REMOTE;
DO_DEBUG( count += sprintf( buffer + count, " start get %lld (%d)", get_line->length, frag->pipeline.pos_transfert ); );
frag->pipeline.pos_transfert = (frag->pipeline.pos_transfert + 1) % GM_PIPELINE_DEPTH;
@ -272,6 +269,7 @@ int mca_ptl_gm_sender_advance_pipeline( mca_ptl_gm_send_frag_t* frag )
DO_DEBUG( ompi_output( 0, "sender %d %s", orte_process_info.my_name->vpid, buffer ); )
return OMPI_SUCCESS;
}
#endif /* OMPI_MCA_PTL_GM_HAVE_RDMA_GET */
static inline
int mca_ptl_gm_send_internal_rndv_header( mca_ptl_gm_peer_t *ptl_peer,
@ -385,8 +383,10 @@ int mca_ptl_gm_peer_send_continue( mca_ptl_gm_peer_t *ptl_peer,
uint64_t remaining_bytes, burst_length;
ompi_list_item_t *item;
int rc = 0;
#if OMPI_MCA_PTL_GM_HAVE_RDMA_GET
gm_status_t status;
mca_ptl_gm_pipeline_line_t* pipeline;
#endif /* OMPI_MCA_PTL_GM_HAVE_RDMA_GET */
fragment->frag_offset = offset;
@ -444,7 +444,7 @@ int mca_ptl_gm_peer_send_continue( mca_ptl_gm_peer_t *ptl_peer,
* with the others informations. When we reach this point the rendez-vous protocol
* has already been realized so we know that the receiver expect our message.
*/
#if OMPI_MCA_PTL_GM_HAVE_RDMA_GET == 0
#if OMPI_MCA_PTL_GM_HAVE_RDMA_GET
/* Trigger the long rendez-vous protocol only if gm_get is supported */
if( remaining_bytes > mca_ptl_gm_component.gm_rndv_burst_limit )
flags |= PTL_FLAG_GM_REQUIRE_LOCK;
@ -453,6 +453,7 @@ int mca_ptl_gm_peer_send_continue( mca_ptl_gm_peer_t *ptl_peer,
if( !(PTL_FLAG_GM_REQUIRE_LOCK & flags) )
return OMPI_SUCCESS;
#if OMPI_MCA_PTL_GM_HAVE_RDMA_GET
pipeline = &(fragment->pipeline.lines[0]);
pipeline->length = fragment->frag_send.frag_base.frag_size - fragment->frag_bytes_processed;
if( pipeline->length > mca_ptl_gm_component.gm_rdma_frag_size ) {
@ -476,6 +477,9 @@ int mca_ptl_gm_peer_send_continue( mca_ptl_gm_peer_t *ptl_peer,
/* Now we are waiting for the ack message. Meanwhile we can register the sender first piece
* of data. In this way we have a recovery between the expensive registration on both sides.
*/
#else
assert( 0 );
#endif /* OMPI_MCA_PTL_GM_HAVE_RDMA_GET */
return OMPI_SUCCESS;
}
@ -700,7 +704,6 @@ mca_ptl_gm_recv_frag_frag( struct mca_ptl_gm_module_t* ptl,
uint32_t iov_count, max_data = 0, header_length;
int32_t freeAfter, rc;
mca_ptl_gm_recv_frag_t* frag;
mca_ptl_gm_pipeline_line_t* pipeline;
header_length = sizeof(mca_ptl_base_frag_header_t);
if( hdr->hdr_frag.hdr_common.hdr_flags & PTL_FLAG_GM_HAS_FRAGMENT ) {
@ -783,28 +786,36 @@ mca_ptl_gm_recv_frag_frag( struct mca_ptl_gm_module_t* ptl,
return NULL;
}
/* There is a kind of rendez-vous protocol used internally by the GM driver. If the amount of data
* to transfert is large enough, then the sender will start sending a frag message with the
* remote_memory set to NULL (but with the length set to the length of the first fragment).
* It will allow the receiver to start to register it's own memory. Later when the receiver
* get a fragment with the remote_memory field not NULL it can start getting the data.
*/
if( NULL == hdr->registered_memory.pval ) { /* first round of the local rendez-vous protocol */
pipeline = &(frag->pipeline.lines[0]);
pipeline->hdr_flags = hdr->hdr_frag.hdr_common.hdr_flags;
pipeline->offset = frag->frag_offset + frag->frag_bytes_processed;
pipeline->length = 0; /* we can lie about this one */
mca_ptl_gm_receiver_advance_pipeline( frag, 0 );
} else {
pipeline = &(frag->pipeline.lines[frag->pipeline.pos_remote]);
DO_DEBUG( ompi_output( 0, "receiver %d %p get remote memory length %lld (%d)\n",
orte_process_info.my_name->vpid, frag, hdr->hdr_frag.hdr_frag_length, frag->pipeline.pos_remote ); );
frag->pipeline.pos_remote = (frag->pipeline.pos_remote + 1) % GM_PIPELINE_DEPTH;
assert( (pipeline->flags & PTL_GM_PIPELINE_REMOTE) == 0 );
pipeline->remote_memory = hdr->registered_memory;
pipeline->flags |= PTL_GM_PIPELINE_REMOTE;
mca_ptl_gm_receiver_advance_pipeline( frag, 0 );
#if OMPI_MCA_PTL_GM_HAVE_RDMA_GET
{
mca_ptl_gm_pipeline_line_t* pipeline;
/* There is a kind of rendez-vous protocol used internally by the GM driver. If the amount of data
* to transfert is large enough, then the sender will start sending a frag message with the
* remote_memory set to NULL (but with the length set to the length of the first fragment).
* It will allow the receiver to start to register it's own memory. Later when the receiver
* get a fragment with the remote_memory field not NULL it can start getting the data.
*/
if( NULL == hdr->registered_memory.pval ) { /* first round of the local rendez-vous protocol */
pipeline = &(frag->pipeline.lines[0]);
pipeline->hdr_flags = hdr->hdr_frag.hdr_common.hdr_flags;
pipeline->offset = frag->frag_offset + frag->frag_bytes_processed;
pipeline->length = 0; /* we can lie about this one */
mca_ptl_gm_receiver_advance_pipeline( frag, 0 );
} else {
pipeline = &(frag->pipeline.lines[frag->pipeline.pos_remote]);
DO_DEBUG( ompi_output( 0, "receiver %d %p get remote memory length %lld (%d)\n",
orte_process_info.my_name->vpid, frag, hdr->hdr_frag.hdr_frag_length, frag->pipeline.pos_remote ); );
frag->pipeline.pos_remote = (frag->pipeline.pos_remote + 1) % GM_PIPELINE_DEPTH;
assert( (pipeline->flags & PTL_GM_PIPELINE_REMOTE) == 0 );
pipeline->remote_memory = hdr->registered_memory;
pipeline->flags |= PTL_GM_PIPELINE_REMOTE;
mca_ptl_gm_receiver_advance_pipeline( frag, 0 );
}
}
#else
assert( 0 );
#endif /* OMPI_MCA_PTL_GM_HAVE_RDMA_GET */
return NULL;
}
@ -821,6 +832,7 @@ mca_ptl_gm_recv_frag_fin( struct mca_ptl_gm_module_t* ptl,
frag->frag_send.frag_base.frag_header.hdr_ack.hdr_dst_match = hdr->hdr_ack.hdr_dst_match;
frag->frag_send.frag_request->req_peer_match = hdr->hdr_ack.hdr_dst_match;
if( PTL_FLAG_GM_REQUIRE_LOCK & hdr->hdr_common.hdr_flags ) {
#if OMPI_MCA_PTL_GM_HAVE_RDMA_GET
if( 0 == hdr->hdr_ack.hdr_dst_size ) {
DO_DEBUG( ompi_output( 0, "sender %d %p get FIN message (initial)", orte_process_info.my_name->vpid, frag ); );
/* I just receive the ack for the first fragment => setup the pipeline */
@ -832,6 +844,9 @@ mca_ptl_gm_recv_frag_fin( struct mca_ptl_gm_module_t* ptl,
}
/* continue the pipeline ... send the next segment */
mca_ptl_gm_sender_advance_pipeline( frag );
#else
assert( 0 );
#endif /* OMPI_MCA_PTL_GM_HAVE_RDMA_GET */
} else {
DO_DEBUG( ompi_output( 0, "sender %d burst data after rendez-vous protocol", orte_process_info.my_name->vpid ); );
/* do a burst but with the remote fragment as we just get it from the message */