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

Don't keep the data attached to a fragment segmented when we have

Просмотр исходного кода 
to move it into the unexpected queue. Instead pack the data in
only one buffer. Now the code look more optimized and clear, but
I have a doubt about who's using this functionality. I think that
all BTLs always return only one memory segment attached to the
matching fragment (i.e. there is no unexpected iov type receive).

This commit was SVN r14416.
Этот коммит содержится в:
George Bosilca 2007-04-18 15:52:11 +00:00
родитель 037b01ce9e
Коммит 51fc2474f1
2 изменённых файлов: 40 добавлений и 49 удалений
Показать статистику Скачать Patch файл Скачать Diff файл Показать все файлы Свернуть все файлы

2
ompi/mca/pml/ob1/pml_ob1_comm.h
Просмотреть файл

@ -74,7 +74,7 @@ OBJ_CLASS_DECLARATION(mca_pml_ob1_comm_t);
* @return OMPI_SUCCESS or error status on failure.
*/
OMPI_DECLSPEC extern int mca_pml_ob1_comm_init_size(mca_pml_ob1_comm_t* comm, size_t size);
extern int mca_pml_ob1_comm_init_size(mca_pml_ob1_comm_t* comm, size_t size);
#if defined(c_plusplus) || defined(__cplusplus)
}

63
ompi/mca/pml/ob1/pml_ob1_recvfrag.h
Просмотреть файл

@ -62,51 +62,44 @@ do { \
#define MCA_PML_OB1_RECV_FRAG_INIT(frag, hdr, segs, cnt, btl ) \
do { \
size_t i; \
size_t i, _size; \
mca_btl_base_segment_t* macro_segments = frag->segments; \
mca_pml_ob1_buffer_t* buffers = frag->buffers; \
\
unsigned char* _ptr = (unsigned char*)frag->addr; \
/* init recv_frag */ \
frag->btl = btl; \
frag->hdr = *(mca_pml_ob1_hdr_t*)hdr; \
frag->num_segments = cnt; \
/* copy over data */ \
if(cnt == 1 && segs[0].seg_len <= mca_pml_ob1.unexpected_limit ) { \
macro_segments[0].seg_addr.pval = frag->addr; \
macro_segments[0].seg_len = segs[0].seg_len; \
memcpy(frag->addr, \
segs[0].seg_addr.pval, \
segs[0].seg_len); \
} else { \
for(i=0; i<cnt; i++) { \
buffers[i].len = segs[i].seg_len; \
buffers[i].addr = (char*) mca_pml_ob1.allocator->alc_alloc( \
mca_pml_ob1.allocator, \
segs[i].seg_len, \
0, \
NULL); \
macro_segments[i].seg_addr.pval = buffers[i].addr; \
macro_segments[i].seg_len = segs[i].seg_len; \
memcpy(buffers[i].addr, \
segs[i].seg_addr.pval, \
segs[i].seg_len); \
frag->num_segments = 1; \
_size = segs[0].seg_len; \
for( i = 1; i < cnt; i++ ) { \
_size += segs[i].seg_len; \
} \
/* copy over data */ \
if(_size <= mca_pml_ob1.unexpected_limit ) { \
macro_segments[0].seg_addr.pval = frag->addr; \
} else { \
buffers[0].len = _size; \
buffers[0].addr = (char*) \
mca_pml_ob1.allocator->alc_alloc( mca_pml_ob1.allocator, \
buffers[0].len, \
0, NULL); \
_ptr = (unsigned char*)(buffers[0].addr); \
macro_segments[0].seg_addr.pval = buffers[0].addr; \
} \
macro_segments[0].seg_len = _size; \
for( i = 0; i < cnt; i++ ) { \
memcpy( _ptr, segs[i].seg_addr.pval, segs[i].seg_len); \
_ptr += segs[i].seg_len; \
} \
} while(0)
#define MCA_PML_OB1_RECV_FRAG_RETURN(frag) \
do { \
size_t i; \
if(!(frag->num_segments == 1 && frag->segments[0].seg_len <= \
mca_pml_ob1.unexpected_limit)) { \
\
if( frag->segments[0].seg_len > mca_pml_ob1.unexpected_limit ) { \
/* return buffers */ \
for(i=0; i<frag->num_segments; i++) { \
mca_pml_ob1.allocator->alc_free( \
mca_pml_ob1.allocator, \
frag->buffers[i].addr ); \
} \
mca_pml_ob1.allocator->alc_free( mca_pml_ob1.allocator, \
frag->buffers[0].addr ); \
} \
frag->num_segments = 0; \
\
@ -120,12 +113,10 @@ do { \
* Callback from BTL on receipt of a recv_frag.
*/
OMPI_DECLSPEC void mca_pml_ob1_recv_frag_callback(
mca_btl_base_module_t *btl,
extern void mca_pml_ob1_recv_frag_callback( mca_btl_base_module_t *btl,
mca_btl_base_tag_t tag,
mca_btl_base_descriptor_t* descriptor,
void* cbdata
);
void* cbdata );
#if defined(c_plusplus) || defined(__cplusplus)
}