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

Decrease the latency. Just keep everything minimal.

Просмотр исходного кода 
This commit was SVN r3844.
Этот коммит содержится в:
George Bosilca 2004-12-17 09:27:10 +00:00
родитель 17f37130d2
Коммит 5d50b025b8
2 изменённых файлов: 10 добавлений и 20 удалений
Показать статистику Скачать Patch файл Скачать Diff файл Показать все файлы Свернуть все файлы

6
src/mca/ptl/gm/src/ptl_gm.c
Просмотреть файл

@ -393,17 +393,17 @@ mca_ptl_gm_matched( mca_ptl_base_module_t * ptl,
ompi_convertor_copy( peer->peer_proc->proc_ompi->proc_convertor,
&frag->frag_base.frag_convertor);
ompi_convertor_init_for_recv( &frag->frag_base.frag_convertor,
0,
ompi_convertor_init_for_recv( &frag->frag_base.frag_convertor, 0,
request->req_base.req_datatype,
request->req_base.req_count,
request->req_base.req_addr,
0 /* TO DO which offset ? */, NULL );
0, NULL );
out_size = 1;
max_data = iov.iov_len;
rc = ompi_convertor_unpack( &frag->frag_base.frag_convertor, &(iov),
&out_size, &max_data, &freeAfter );
assert( rc >= 0 );
recv_frag->frag_bytes_processed += max_data;
}
/* update progress*/

24
src/mca/ptl/gm/src/ptl_gm_priv.c
Просмотреть файл

@ -217,13 +217,6 @@ int mca_ptl_gm_peer_send( mca_ptl_gm_peer_t *ptl_peer,
/* At this point the header is already filled up with informations as a match header */
if( (flags & MCA_PTL_FLAGS_ACK) || (0 == offset) ) {
hdr->hdr_common.hdr_flags = flags;
hdr->hdr_match.hdr_contextid = sendreq->req_base.req_comm->c_contextid;
hdr->hdr_match.hdr_src = sendreq->req_base.req_comm->c_my_rank;
hdr->hdr_match.hdr_dst = sendreq->req_base.req_peer;
hdr->hdr_match.hdr_tag = sendreq->req_base.req_tag;
hdr->hdr_match.hdr_msg_length = sendreq->req_bytes_packed;
hdr->hdr_match.hdr_msg_seq = sendreq->req_base.req_sequence;
if( flags & MCA_PTL_FLAGS_ACK ) {
header_length = sizeof(mca_ptl_base_rendezvous_header_t);
hdr->hdr_common.hdr_type = MCA_PTL_HDR_TYPE_RNDV;
@ -231,13 +224,11 @@ int mca_ptl_gm_peer_send( mca_ptl_gm_peer_t *ptl_peer,
hdr->hdr_rndv.hdr_src_ptr.lval = 0;
hdr->hdr_rndv.hdr_src_ptr.pval = fragment;
} else {
hdr->hdr_common.hdr_type = MCA_PTL_HDR_TYPE_MATCH;
header_length = sizeof(mca_ptl_base_match_header_t);
}
} else {
header_length = sizeof(mca_ptl_base_frag_header_t);
hdr->hdr_frag.hdr_common.hdr_type = MCA_PTL_HDR_TYPE_FRAG;
hdr->hdr_frag.hdr_common.hdr_flags = flags;
hdr->hdr_frag.hdr_frag_length = size_in;
hdr->hdr_frag.hdr_frag_offset = offset;
hdr->hdr_frag.hdr_src_ptr.lval = 0; /* for VALGRIND/PURIFY - REPLACE WITH MACRO */
@ -245,8 +236,6 @@ int mca_ptl_gm_peer_send( mca_ptl_gm_peer_t *ptl_peer,
hdr->hdr_frag.hdr_dst_ptr = sendreq->req_peer_match;
}
iov.iov_len = 0; /* nothing yet */
if( size_in > 0 ) {
/* first fragment (eager send) and first fragment of long protocol
* can use the convertor initialized on the request. The remaining
@ -279,9 +268,16 @@ int mca_ptl_gm_peer_send( mca_ptl_gm_peer_t *ptl_peer,
fragment->send_frag.frag_base.frag_addr = ((char*)fragment->send_buf) + header_length;
fragment->send_frag.frag_base.frag_size = max_data;
/* must update the offset after actual fragment size is determined
* before attempting to send the fragment
*/
mca_pml_base_send_request_offset( sendreq,
fragment->send_frag.frag_base.frag_size );
} else {
fragment->send_frag.frag_base.frag_addr = NULL;
fragment->send_frag.frag_base.frag_size = 0;
iov.iov_len = 0; /* no data will be transmitted */
}
/* adjust size and request offset to reflect actual number of bytes
@ -289,12 +285,6 @@ int mca_ptl_gm_peer_send( mca_ptl_gm_peer_t *ptl_peer,
*/
size_out = iov.iov_len + header_length;
/* must update the offset after actual fragment size is determined
* before attempting to send the fragment
*/
mca_pml_base_send_request_offset( sendreq,
fragment->send_frag.frag_base.frag_size );
/* Send the first fragment */
gm_send_to_peer_with_callback( ptl_peer->peer_ptl->gm_port, fragment->send_buf,
GM_SIZE, size_out, GM_LOW_PRIORITY, ptl_peer->local_id,