mirror of
https://github.com/amd/blis.git
synced 2026-04-20 07:38:53 +00:00
ed5010d65b
Standardize format of AMD copyright notice. AMD-Internal: [CPUPL-3519] Change-Id: I98530e58138765e5cd5bc0c97500506801eb0bf0
439 lines
13 KiB
C
439 lines
13 KiB
C
/*
|
|
|
|
BLIS
|
|
An object-based framework for developing high-performance BLAS-like
|
|
libraries.
|
|
|
|
Copyright (C) 2014, The University of Texas at Austin
|
|
Copyright (C) 2022 - 2023, Advanced Micro Devices, Inc. All rights reserved.
|
|
|
|
Redistribution and use in source and binary forms, with or without
|
|
modification, are permitted provided that the following conditions are
|
|
met:
|
|
- Redistributions of source code must retain the above copyright
|
|
notice, this list of conditions and the following disclaimer.
|
|
- Redistributions in binary form must reproduce the above copyright
|
|
notice, this list of conditions and the following disclaimer in the
|
|
documentation and/or other materials provided with the distribution.
|
|
- Neither the name(s) of the copyright holder(s) nor the names of its
|
|
contributors may be used to endorse or promote products derived
|
|
from this software without specific prior written permission.
|
|
|
|
THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
|
|
"AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
|
|
LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
|
|
A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
|
|
HOLDER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
|
|
SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
|
|
LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
|
|
DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
|
|
THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
|
|
(INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
|
|
OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
|
|
|
|
*/
|
|
|
|
#include "blis.h"
|
|
|
|
#undef GENTFUNC
|
|
#define GENTFUNC( ctype, ch, opname ) \
|
|
\
|
|
void PASTEMAC(ch,opname) \
|
|
( \
|
|
bool will_pack, \
|
|
packbuf_t pack_buf_type, \
|
|
dim_t m, \
|
|
dim_t k, \
|
|
dim_t mr, \
|
|
cntx_t* restrict cntx, \
|
|
rntm_t* restrict rntm, \
|
|
mem_t* restrict mem, \
|
|
thrinfo_t* restrict thread \
|
|
) \
|
|
{ \
|
|
AOCL_DTL_TRACE_ENTRY(AOCL_DTL_LEVEL_TRACE_5); \
|
|
/* Inspect whether we are going to be packing matrix A. */ \
|
|
if ( will_pack == FALSE ) \
|
|
{ \
|
|
} \
|
|
else /* if ( will_pack == TRUE ) */ \
|
|
{ \
|
|
/* NOTE: This "rounding up" of the last upanel is actually optional
|
|
for the rrc/crc cases, but absolutely necessary for the other cases
|
|
since we NEED that last micropanel to have the same ldim (cs_p) as
|
|
the other micropanels. Why? So that millikernels can use the same
|
|
upanel ldim for all iterations of the ir loop. */ \
|
|
const dim_t m_pack = ( m / mr + ( m % mr ? 1 : 0 ) ) * mr; \
|
|
const dim_t k_pack = k; \
|
|
\
|
|
/* Barrier to make sure all threads are caught up and ready to begin
|
|
the packm stage. */ \
|
|
bli_thread_barrier( thread ); \
|
|
\
|
|
/* Compute the size of the memory block eneded. */ \
|
|
siz_t size_needed = sizeof( ctype ) * m_pack * k_pack; \
|
|
\
|
|
/* Check the mem_t entry provided by the caller. If it is unallocated,
|
|
then we need to acquire a block from the memory broker. */ \
|
|
if ( bli_mem_is_unalloc( mem ) ) \
|
|
{ \
|
|
if ( bli_thread_am_ochief( thread ) ) \
|
|
{ \
|
|
/* Acquire directly to the chief thread's mem_t that was
|
|
passed in. It needs to be that mem_t struct, and not a
|
|
local (temporary) mem_t, since there is no barrier until
|
|
after packing is finished, which could allow a race
|
|
condition whereby the chief thread exits the current
|
|
function before the other threads have a chance to copy
|
|
from it. (A barrier would fix that race condition, but
|
|
then again, I prefer to keep barriers to a minimum.) */ \
|
|
bli_pba_acquire_m \
|
|
( \
|
|
rntm, \
|
|
size_needed, \
|
|
pack_buf_type, \
|
|
mem \
|
|
); \
|
|
} \
|
|
\
|
|
/* Broadcast the address of the chief thread's passed-in mem_t
|
|
to all threads. */ \
|
|
mem_t* mem_p = bli_thread_broadcast( thread, mem ); \
|
|
\
|
|
/* Non-chief threads: Copy the contents of the chief thread's
|
|
passed-in mem_t to the passed-in mem_t for this thread. (The
|
|
chief thread already has the mem_t, so it does not need to
|
|
perform any copy.) */ \
|
|
if ( !bli_thread_am_ochief( thread ) ) \
|
|
{ \
|
|
*mem = *mem_p; \
|
|
} \
|
|
} \
|
|
else /* if ( bli_mem_is_alloc( mem ) ) */ \
|
|
{ \
|
|
/* If the mem_t entry provided by the caller does NOT contain a NULL
|
|
buffer, then a block has already been acquired from the memory
|
|
broker and cached by the caller. */ \
|
|
\
|
|
/* As a sanity check, we should make sure that the mem_t object isn't
|
|
associated with a block that is too small compared to the size of
|
|
the packed matrix buffer that is needed, according to the value
|
|
computed above. */ \
|
|
siz_t mem_size = bli_mem_size( mem ); \
|
|
\
|
|
if ( mem_size < size_needed ) \
|
|
{ \
|
|
if ( bli_thread_am_ochief( thread ) ) \
|
|
{ \
|
|
/* The chief thread releases the existing block associated
|
|
with the mem_t, and then re-acquires a new block, saving
|
|
the associated mem_t to its passed-in mem_t. (See coment
|
|
above for why the acquisition needs to be directly to
|
|
the chief thread's passed-in mem_t and not a local
|
|
(temporary) mem_t. */ \
|
|
bli_pba_release \
|
|
( \
|
|
rntm, \
|
|
mem \
|
|
); \
|
|
bli_pba_acquire_m \
|
|
( \
|
|
rntm, \
|
|
size_needed, \
|
|
pack_buf_type, \
|
|
mem \
|
|
); \
|
|
} \
|
|
\
|
|
/* Broadcast the address of the chief thread's passed-in mem_t
|
|
to all threads. */ \
|
|
mem_t* mem_p = bli_thread_broadcast( thread, mem ); \
|
|
\
|
|
/* Non-chief threads: Copy the contents of the chief thread's
|
|
passed-in mem_t to the passed-in mem_t for this thread. (The
|
|
chief thread already has the mem_t, so it does not need to
|
|
perform any copy.) */ \
|
|
if ( !bli_thread_am_ochief( thread ) ) \
|
|
{ \
|
|
*mem = *mem_p; \
|
|
} \
|
|
} \
|
|
else \
|
|
{ \
|
|
/* If the mem_t entry is already allocated and sufficiently large,
|
|
then we use it as-is. No action is needed. */ \
|
|
} \
|
|
} \
|
|
} \
|
|
AOCL_DTL_TRACE_EXIT(AOCL_DTL_LEVEL_TRACE_5);\
|
|
}
|
|
|
|
INSERT_GENTFUNC_BASIC0( packm_sup_init_mem_a )
|
|
|
|
|
|
#undef GENTFUNC
|
|
#define GENTFUNC( ctype, ch, opname ) \
|
|
\
|
|
void PASTEMAC(ch,opname) \
|
|
( \
|
|
bool did_pack, \
|
|
rntm_t* restrict rntm, \
|
|
mem_t* restrict mem, \
|
|
thrinfo_t* restrict thread \
|
|
) \
|
|
{ \
|
|
AOCL_DTL_TRACE_ENTRY(AOCL_DTL_LEVEL_TRACE_5);\
|
|
/* Inspect whether we previously packed matrix A. */ \
|
|
if ( did_pack == FALSE ) \
|
|
{ \
|
|
/* If we didn't pack matrix A, there's nothing to be done. */ \
|
|
} \
|
|
else /* if ( did_pack == TRUE ) */ \
|
|
{ \
|
|
if ( thread != NULL ) \
|
|
if ( bli_thread_am_ochief( thread ) ) \
|
|
{ \
|
|
/* Check the mem_t entry provided by the caller. Only proceed if it
|
|
is allocated, which it should be. */ \
|
|
if ( bli_mem_is_alloc( mem ) ) \
|
|
{ \
|
|
bli_pba_release \
|
|
( \
|
|
rntm, \
|
|
mem \
|
|
); \
|
|
} \
|
|
} \
|
|
} \
|
|
AOCL_DTL_TRACE_EXIT(AOCL_DTL_LEVEL_TRACE_5);\
|
|
}
|
|
|
|
INSERT_GENTFUNC_BASIC0( packm_sup_finalize_mem_a )
|
|
|
|
|
|
#undef GENTFUNC
|
|
#define GENTFUNC( ctype, ch, opname ) \
|
|
\
|
|
void PASTEMAC(ch,opname) \
|
|
( \
|
|
bool will_pack, \
|
|
stor3_t stor_id, \
|
|
pack_t* restrict schema, \
|
|
dim_t m, \
|
|
dim_t k, \
|
|
dim_t mr, \
|
|
dim_t* restrict m_max, \
|
|
dim_t* restrict k_max, \
|
|
ctype* x, inc_t rs_x, inc_t cs_x, \
|
|
ctype** p, inc_t* restrict rs_p, inc_t* restrict cs_p, \
|
|
dim_t* restrict pd_p, inc_t* restrict ps_p, \
|
|
cntx_t* restrict cntx, \
|
|
mem_t* restrict mem, \
|
|
thrinfo_t* restrict thread \
|
|
) \
|
|
{ \
|
|
AOCL_DTL_TRACE_ENTRY(AOCL_DTL_LEVEL_TRACE_5);\
|
|
/* Inspect whether we are going to be packing matrix A. */ \
|
|
if ( will_pack == FALSE ) \
|
|
{ \
|
|
*m_max = m; \
|
|
*k_max = k; \
|
|
\
|
|
/* Set the parameters for use with no packing of A (ie: using the
|
|
source matrix A directly). */ \
|
|
{ \
|
|
/* Use the strides of the source matrix as the final values. */ \
|
|
*rs_p = rs_x; \
|
|
*cs_p = cs_x; \
|
|
\
|
|
*pd_p = mr; \
|
|
*ps_p = mr * rs_x; \
|
|
\
|
|
/* Set the schema to "not packed" to indicate that packing will be
|
|
skipped. */ \
|
|
*schema = BLIS_NOT_PACKED; \
|
|
} \
|
|
\
|
|
/* Since we won't be packing, simply update the buffer address provided
|
|
by the caller to point to source matrix. */ \
|
|
*p = x; \
|
|
} \
|
|
else /* if ( will_pack == TRUE ) */ \
|
|
{ \
|
|
/* NOTE: This is "rounding up" of the last upanel is actually optional
|
|
for the rrc/crc cases, but absolutely necessary for the other cases
|
|
since we NEED that last micropanel to have the same ldim (cs_p) as
|
|
the other micropanels. Why? So that millikernels can use the same
|
|
upanel ldim for all iterations of the ir loop. */ \
|
|
*m_max = ( m / mr + ( m % mr ? 1 : 0 ) ) * mr; \
|
|
*k_max = k; \
|
|
\
|
|
/* Determine the dimensions and strides for the packed matrix A. */ \
|
|
if ( stor_id == BLIS_RRC || \
|
|
stor_id == BLIS_CRC ) \
|
|
{ \
|
|
/* stor3_t id values _RRC and _CRC: pack A to plain row storage. */ \
|
|
*rs_p = k; \
|
|
*cs_p = 1; \
|
|
\
|
|
*pd_p = mr; \
|
|
*ps_p = mr * k; \
|
|
\
|
|
/* Set the schema to "row packed" to indicate packing to plain
|
|
row storage. */ \
|
|
*schema = BLIS_PACKED_ROWS; \
|
|
} \
|
|
else \
|
|
{ \
|
|
/* All other stor3_t ids: pack A to column-stored row-panels. */ \
|
|
*rs_p = 1; \
|
|
*cs_p = mr; \
|
|
\
|
|
*pd_p = mr; \
|
|
*ps_p = mr * k; \
|
|
\
|
|
/* Set the schema to "packed row panels" to indicate packing to
|
|
conventional column-stored row panels. */ \
|
|
*schema = BLIS_PACKED_ROW_PANELS; \
|
|
} \
|
|
\
|
|
/* Set the buffer address provided by the caller to point to the
|
|
memory associated with the mem_t entry acquired from the memory
|
|
broker. */ \
|
|
*p = bli_mem_buffer( mem ); \
|
|
} \
|
|
AOCL_DTL_TRACE_EXIT(AOCL_DTL_LEVEL_TRACE_5);\
|
|
}
|
|
|
|
INSERT_GENTFUNC_BASIC0( packm_sup_init_a )
|
|
|
|
|
|
//
|
|
// Define BLAS-like interfaces to the variant chooser.
|
|
//
|
|
|
|
#undef GENTFUNC
|
|
#define GENTFUNC( ctype, ch, opname ) \
|
|
\
|
|
void PASTEMAC(ch,opname) \
|
|
( \
|
|
bool will_pack, \
|
|
packbuf_t pack_buf_type, \
|
|
stor3_t stor_id, \
|
|
trans_t transc, \
|
|
dim_t m_alloc, \
|
|
dim_t k_alloc, \
|
|
dim_t m, \
|
|
dim_t k, \
|
|
dim_t mr, \
|
|
ctype* restrict kappa, \
|
|
ctype* restrict a, inc_t rs_a, inc_t cs_a, \
|
|
ctype** restrict p, inc_t* restrict rs_p, inc_t* restrict cs_p, \
|
|
inc_t* restrict ps_p, \
|
|
cntx_t* restrict cntx, \
|
|
rntm_t* restrict rntm, \
|
|
mem_t* restrict mem, \
|
|
thrinfo_t* restrict thread \
|
|
) \
|
|
{ \
|
|
pack_t schema; \
|
|
dim_t m_max; \
|
|
dim_t k_max; \
|
|
dim_t pd_p; \
|
|
\
|
|
AOCL_DTL_TRACE_ENTRY(AOCL_DTL_LEVEL_TRACE_5);\
|
|
/* Prepare the packing destination buffer. If packing is not requested,
|
|
this function will reduce to a no-op. */ \
|
|
PASTEMAC(ch,packm_sup_init_mem_a) \
|
|
( \
|
|
will_pack, \
|
|
pack_buf_type, \
|
|
m_alloc, k_alloc, mr, \
|
|
cntx, \
|
|
rntm, \
|
|
mem, \
|
|
thread \
|
|
); \
|
|
\
|
|
/* Determine the packing buffer and related parameters for matrix A. If A
|
|
will not be packed, then a_use will be set to point to a and the _a_use
|
|
strides will be set accordingly. */ \
|
|
PASTEMAC(ch,packm_sup_init_a) \
|
|
( \
|
|
will_pack, \
|
|
stor_id, \
|
|
&schema, \
|
|
m, k, mr, \
|
|
&m_max, &k_max, \
|
|
a, rs_a, cs_a, \
|
|
p, rs_p, cs_p, \
|
|
&pd_p, ps_p, \
|
|
cntx, \
|
|
mem, \
|
|
thread \
|
|
); \
|
|
\
|
|
/* Inspect whether we are going to be packing matrix A. */ \
|
|
if ( will_pack == FALSE ) \
|
|
{ \
|
|
/* If we aren't going to pack matrix A, then there's nothing to do. */ \
|
|
\
|
|
/*
|
|
printf( "blis_ packm_sup_a: not packing A.\n" ); \
|
|
*/ \
|
|
} \
|
|
else /* if ( will_pack == TRUE ) */ \
|
|
{ \
|
|
if ( schema == BLIS_PACKED_ROWS ) \
|
|
{ \
|
|
/*
|
|
printf( "blis_ packm_sup_a: packing A to rows.\n" ); \
|
|
*/ \
|
|
\
|
|
/* For plain packing by rows, use var2. */ \
|
|
PASTEMAC(ch,packm_sup_var2) \
|
|
( \
|
|
transc, \
|
|
schema, \
|
|
m, \
|
|
k, \
|
|
kappa, \
|
|
a, rs_a, cs_a, \
|
|
*p, *rs_p, *cs_p, \
|
|
cntx, \
|
|
thread \
|
|
); \
|
|
} \
|
|
else /* if ( schema == BLIS_PACKED_ROW_PANELS ) */ \
|
|
{ \
|
|
/*
|
|
printf( "blis_ packm_sup_a: packing A to row panels.\n" ); \
|
|
*/ \
|
|
\
|
|
/* For packing to column-stored row panels, use var1. */ \
|
|
PASTEMAC(ch,packm_sup_var1) \
|
|
( \
|
|
transc, \
|
|
schema, \
|
|
m, \
|
|
k, \
|
|
m_max, \
|
|
k_max, \
|
|
kappa, \
|
|
a, rs_a, cs_a, \
|
|
*p, *rs_p, *cs_p, \
|
|
pd_p, *ps_p, \
|
|
cntx, \
|
|
thread \
|
|
); \
|
|
} \
|
|
\
|
|
/* Barrier so that packing is done before computation. */ \
|
|
bli_thread_barrier( thread ); \
|
|
} \
|
|
AOCL_DTL_TRACE_EXIT(AOCL_DTL_LEVEL_TRACE_5);\
|
|
}
|
|
|
|
INSERT_GENTFUNC_BASIC0( packm_sup_a )
|
|
|