Omnibus PR - Oct 2023 (#678)

Details:
- This is an "omnibus" commit, consisting of multiple medium-sized
  commits that affect non-trivial aspects of BLIS. The major highlights:
  - Relocated the pba, sba pool (from the rntm_t), and mem_t (from the
    cntl_t) to the thrinfo_t object. This allows the rntm_t to be
    effectively const (although it is sometimes copied internally and
    modified to reflect different ways of parallelism). Moving the mem_t
    sets the stage for sharing a global control tree amongst all
    threads.
  - De-templatized the macrokernels for gemmt, trmm, and trsm to match
    the macrokernel for gemm, which has been de-templatized since
    54fa28b.
  - Reimplemented bli_l3_determine_kc() by separating out the logic for
    adjusting KC based on MR/NR for triangular A and/or B into a new
    function, bli_l3_adjust_kc(). For now, this function is still called
    from bli_l3_determine_kc(), but in the future we plan to have it
    called once when constructing the control tree.
  - Refactored the level-3 thread decorator into two parts:
    - One part deals only with launching threads, each one calling a 
      generic thread entry function. This code resides in frame/thread  
      and constitutes the definition of bli_thread_launch(). Note that 
      it is specific to the threading implementation (OpenMP, pthreads, 
      single, etc.)
    - The other part deals with passing the matrix operands and related
      information into bli_thread_launch(). This is the "l3 decorator" 
      and now resides in frame/3. It is agnostic to the threading
      implementation.
  - Modified the "level" of the thread control tree passed in at each 
    operation. Previously, each operation (e.g. bli_gemm_blk_var1()) was 
    passed in a communicator representing the active thread teams which
    would share the available work. Now, the *parent* thread comm is 
    passed in. The operation then grabs the child comm and uses it to 
    partition the work. The difference is in bli_trsm_blk_var1(), where
    there are now two children nodes for this single operation (i.e. the 
    thread control tree is split one level above where the control tree 
    is). The sub-prenode is used for the trsm subproblem while the 
    normal sub-node is used for the gemm part. Importantly, the parent 
    comm is used for the barrier between them. 
- Removed cntl_t* arguments from bli_*_front() functions. These will be
  added back in the future when the control tree's creation is moved so
  that it happens much sooner (provided that bli_*_front() have not been
  absorbed into their respective bli_*_ex() functions).
- Renamed various bli_thread_*() query functions to bli_thrinfo_*(),
  for consistency. This includes _num_threads(), _thread_id(), _n_way(), 
  _work_id(), _sba_pool(), _pba(), _mem(), _barrier(), _broadcast(), and
  _am_chief().
- Removed extraneous barrier from _blk_var3() of gemm and trsm.
- Fixed a typo in bli_type_defs.h where BLIS_BLAS_INT_TYPE_SIZE was
  misspelled.

Author: devinamatthews

addon/gemmd/attic/bao_gemmd_bp_var2.c

@@ -386,8 +386,8 @@ void PASTECH2(bao_,ch,varname) \
 				/* Query the number of threads and thread ids for the JR loop.
 				   NOTE: These values are only needed when computing the next
 				   micropanel of B. */ \
-				const dim_t jr_nt  = bli_thread_n_way( thread_jr ); \
-				const dim_t jr_tid = bli_thread_work_id( thread_jr ); \
+				const dim_t jr_nt  = bli_thrinfo_n_way( thread_jr ); \
+				const dim_t jr_tid = bli_thrinfo_work_id( thread_jr ); \
 \
 				/* Compute number of primary and leftover components of the JR loop. */ \
 				dim_t jr_iter = ( nc_cur + NR - 1 ) / NR; \
@@ -416,8 +416,8 @@ void PASTECH2(bao_,ch,varname) \
 					/* Query the number of threads and thread ids for the IR loop.
 					   NOTE: These values are only needed when computing the next
 					   micropanel of A. */ \
-					const dim_t ir_nt  = bli_thread_n_way( thread_ir ); \
-					const dim_t ir_tid = bli_thread_work_id( thread_ir ); \
+					const dim_t ir_nt  = bli_thrinfo_n_way( thread_ir ); \
+					const dim_t ir_tid = bli_thrinfo_work_id( thread_ir ); \
 \
 					/* Compute number of primary and leftover components of the IR loop. */ \
 					dim_t ir_iter = ( mc_cur + MR - 1 ) / MR; \
@@ -476,7 +476,7 @@ void PASTECH2(bao_,ch,varname) \
 			/* This barrier is needed to prevent threads from starting to pack
 			   the next row panel of B before the current row panel is fully
 			   computed upon. */ \
-			bli_thread_barrier( thread_pb ); \
+			bli_thrinfo_barrier( thread_pb ); \
 		} \
 	} \
 \

addon/gemmd/bao_gemmd_bp_var1.c

@@ -370,8 +370,8 @@ void PASTECH2(bao_,ch,varname) \
 				/* Query the number of threads and thread ids for the JR loop.
 				   NOTE: These values are only needed when computing the next
 				   micropanel of B. */ \
-				const dim_t jr_nt  = bli_thread_n_way( thread_jr ); \
-				const dim_t jr_tid = bli_thread_work_id( thread_jr ); \
+				const dim_t jr_nt  = bli_thrinfo_n_way( thread_jr ); \
+				const dim_t jr_tid = bli_thrinfo_work_id( thread_jr ); \
 \
 				/* Compute number of primary and leftover components of the JR loop. */ \
 				dim_t jr_iter = ( nc_cur + NR - 1 ) / NR; \
@@ -400,8 +400,8 @@ void PASTECH2(bao_,ch,varname) \
 					/* Query the number of threads and thread ids for the IR loop.
 					   NOTE: These values are only needed when computing the next
 					   micropanel of A. */ \
-					const dim_t ir_nt  = bli_thread_n_way( thread_ir ); \
-					const dim_t ir_tid = bli_thread_work_id( thread_ir ); \
+					const dim_t ir_nt  = bli_thrinfo_n_way( thread_ir ); \
+					const dim_t ir_tid = bli_thrinfo_work_id( thread_ir ); \
 \
 					/* Compute number of primary and leftover components of the IR loop. */ \
 					dim_t ir_iter = ( mc_cur + MR - 1 ) / MR; \
@@ -458,7 +458,7 @@ void PASTECH2(bao_,ch,varname) \
 			/* This barrier is needed to prevent threads from starting to pack
 			   the next row panel of B before the current row panel is fully
 			   computed upon. */ \
-			bli_thread_barrier( rntm, thread_pb ); \
+			bli_thrinfo_barrier( thread_pb ); \
 		} \
 	} \
 \

addon/gemmd/bao_l3_packm_a.c

@@ -61,7 +61,7 @@ void PASTECH2(bao_,ch,opname) \
 \
 	/* Barrier to make sure all threads are caught up and ready to begin the
 	   packm stage. */ \
-	bli_thread_barrier( rntm, thread ); \
+	bli_thrinfo_barrier( thread ); \
 \
 	/* Compute the size of the memory block eneded. */ \
 	siz_t size_needed = sizeof( ctype ) * m_pack * k_pack; \
@@ -90,7 +90,7 @@ void PASTECH2(bao_,ch,opname) \
 \
 		/* Broadcast the address of the chief thread's passed-in mem_t to all
 		   threads. */ \
-		mem_t* mem_p = bli_thread_broadcast( rntm, thread, mem ); \
+		mem_t* mem_p = bli_thrinfo_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
@@ -139,7 +139,7 @@ void PASTECH2(bao_,ch,opname) \
 \
 			/* Broadcast the address of the chief thread's passed-in mem_t
 			   to all threads. */ \
-			mem_t* mem_p = bli_thread_broadcast( rntm, thread, mem ); \
+			mem_t* mem_p = bli_thrinfo_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
@@ -313,13 +313,13 @@ void PASTECH2(bao_,ch,opname) \
 	  d,  incd, \
 	  a,  rs_a,  cs_a, \
 	  *p, *rs_p, *cs_p, \
-		  pd_p,  *ps_p, \
+	       pd_p, *ps_p, \
 	  cntx, \
 	  thread  \
 	); \
 \
 	/* Barrier so that packing is done before computation. */ \
-	bli_thread_barrier( rntm, thread ); \
+	bli_thrinfo_barrier( thread ); \
 }
 
 //INSERT_GENTFUNC_BASIC0( packm_a )

addon/gemmd/bao_l3_packm_b.c

@@ -61,7 +61,7 @@ void PASTECH2(bao_,ch,opname) \
 \
 	/* Barrier to make sure all threads are caught up and ready to begin the
 	   packm stage. */ \
-	bli_thread_barrier( rntm, thread ); \
+	bli_thrinfo_barrier( thread ); \
 \
 	/* Compute the size of the memory block eneded. */ \
 	siz_t size_needed = sizeof( ctype ) * k_pack * n_pack; \
@@ -90,7 +90,7 @@ void PASTECH2(bao_,ch,opname) \
 \
 		/* Broadcast the address of the chief thread's passed-in mem_t to all
 		   threads. */ \
-		mem_t* mem_p = bli_thread_broadcast( rntm, thread, mem ); \
+		mem_t* mem_p = bli_thrinfo_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
@@ -139,7 +139,7 @@ void PASTECH2(bao_,ch,opname) \
 \
 			/* Broadcast the address of the chief thread's passed-in mem_t
 			   to all threads. */ \
-			mem_t* mem_p = bli_thread_broadcast( rntm, thread, mem ); \
+			mem_t* mem_p = bli_thrinfo_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
@@ -313,13 +313,13 @@ void PASTECH2(bao_,ch,opname) \
 	  d,  incd, \
 	  b,  rs_b,  cs_b, \
 	  *p, *rs_p, *cs_p, \
-		  pd_p,  *ps_p, \
+	       pd_p, *ps_p, \
 	  cntx, \
 	  thread  \
 	); \
 \
 	/* Barrier so that packing is done before computation. */ \
-	bli_thread_barrier( rntm, thread ); \
+	bli_thrinfo_barrier( thread ); \
 }
 
 //INSERT_GENTFUNC_BASIC0( packm_b )

addon/gemmd/bao_l3_packm_var1.c

@@ -127,8 +127,8 @@ void PASTECH2(bao_,ch,varname) \
 \
 	/* Query the number of threads and thread ids from the current thread's
 	   packm thrinfo_t node. */ \
-	const dim_t nt  = bli_thread_n_way( thread ); \
-	const dim_t tid = bli_thread_work_id( thread ); \
+	const dim_t nt  = bli_thrinfo_n_way( thread ); \
+	const dim_t tid = bli_thrinfo_work_id( thread ); \
 \
 	/* Suppress warnings in case tid isn't used (ie: as in slab partitioning). */ \
 	( void )nt; \

addon/gemmd/bao_l3_packm_var2.c

@@ -127,8 +127,8 @@ void PASTECH2(bao_,ch,varname) \
 \
 	/* Query the number of threads and thread ids from the current thread's
 	   packm thrinfo_t node. */ \
-	const dim_t nt  = bli_thread_n_way( thread ); \
-	const dim_t tid = bli_thread_work_id( thread ); \
+	const dim_t nt  = bli_thrinfo_n_way( thread ); \
+	const dim_t tid = bli_thrinfo_work_id( thread ); \
 \
 	/* Suppress warnings in case tid isn't used (ie: as in slab partitioning). */ \
 	( void )nt; \