Made barrier after packing implicit.

This also fixed a bug where barriers in the blocked variants were inserted after the inner packing routines,
but not the outer packing routines.
This allowed, for instance, the block of B to not be finished being packed before computation to occur.

frame/1m/packm/bli_packm_int.c

@@ -123,5 +123,8 @@ void bli_packm_int( obj_t*   a,
 	f( a,
 	   p,
        thread );
+
+    // Barrier so that packing is done before computation
+    thread_obarrier( thread );
 }
 

frame/3/gemm/bli_gemm_blk_var1f.c

@@ -119,9 +119,6 @@ void bli_gemm_blk_var1f( obj_t*  a,
 		bli_packm_int( &c1, c1_pack,
 		               cntl_sub_packm_c( cntl ),
                        gemm_thread_sub_ipackm( thread ) );
-        
-        // Packing must be done before computation.
-        thread_ibarrier( thread );
 
 		// Perform gemm subproblem.
 		bli_gemm_int( &BLIS_ONE,

frame/3/gemm/bli_gemm_blk_var2f.c

@@ -119,9 +119,6 @@ void bli_gemm_blk_var2f( obj_t*  a,
 		               cntl_sub_packm_c( cntl ),
                        gemm_thread_sub_ipackm( thread ) );
 
-        // Packing must be done before computation
-        thread_ibarrier( thread );
-
 		// Perform gemm subproblem.
 		bli_gemm_int( &BLIS_ONE,
 		              a_pack,

frame/3/gemm/bli_gemm_blk_var3f.c

@@ -115,9 +115,6 @@ void bli_gemm_blk_var3f( obj_t*  a,
 		bli_packm_int( &b1, b1_pack,
 		               cntl_sub_packm_b( cntl ),
                        gemm_thread_sub_ipackm( thread ) );
-        
-        // Packing must be done before computation.
-        thread_ibarrier( thread );
 
 		// Perform gemm subproblem.
 		bli_gemm_int( &BLIS_ONE,

frame/3/herk/bli_herk_blk_var1f.c

@@ -116,9 +116,6 @@ void bli_herk_blk_var1f( obj_t*  a,
 		               cntl_sub_packm_c( cntl ),
                        herk_thread_sub_ipackm( thread ) );
 
-        // Packing must be done before computation
-        thread_ibarrier( thread );
-
 		// Perform herk subproblem.
 		bli_herk_int( &BLIS_ONE,
 		              a1_pack,

frame/3/herk/bli_herk_blk_var3f.c

@@ -120,9 +120,6 @@ void bli_herk_blk_var3f( obj_t*  a,
 		// internal beta scalar with BLIS_ONE once it has been used in the
 		// first iteration.
 		if ( i != 0 && thread_am_ichief( thread ) ) bli_obj_scalar_reset( c_pack );
-
-        // Packing must be done before computation
-        thread_ibarrier( thread );
         
 		// Perform herk subproblem.
 		bli_herk_int( &BLIS_ONE,

frame/3/trmm/bli_trmm_blk_var1f.c

@@ -128,9 +128,6 @@ void bli_trmm_blk_var1f( obj_t*  a,
 		               cntl_sub_packm_c( cntl ),
                        trmm_thread_sub_ipackm( thread ) );
         
-        // Packing must be finished before computation
-        thread_ibarrier( thread );
-
 		// Perform trmm subproblem.
 		bli_trmm_int( &BLIS_ONE,
 		              a1_pack,

frame/3/trmm/bli_trmm_blk_var2b.c

@@ -116,9 +116,6 @@ void bli_trmm_blk_var2b( obj_t*  a,
 		bli_packm_int( &c1, c1_pack,
 		               cntl_sub_packm_c( cntl ),
                        trmm_thread_sub_ipackm( thread ) );
-        
-        // Packing must be finished before computation
-        thread_ibarrier( thread );
 
 		// Perform trmm subproblem.
 		bli_trmm_int( &BLIS_ONE,

frame/3/trmm/bli_trmm_blk_var2f.c

@@ -116,9 +116,6 @@ void bli_trmm_blk_var2f( obj_t*  a,
 		bli_packm_int( &c1, c1_pack,
 		               cntl_sub_packm_c( cntl ),
                        trmm_thread_sub_ipackm( thread ) );
-        
-        // Packing must be finished before computation
-        thread_ibarrier( thread );
 
 		// Perform trmm subproblem.
 		bli_trmm_int( &BLIS_ONE,

frame/3/trmm/bli_trmm_blk_var3b.c

@@ -113,9 +113,6 @@ void bli_trmm_blk_var3b( obj_t*  a,
 		               cntl_sub_packm_b( cntl ),
                        trmm_thread_sub_ipackm( thread ) );
 
-        // Packing must be done before computation
-        thread_ibarrier( thread );
-
 		// Perform trmm subproblem.
 		bli_trmm_int( &BLIS_ONE,
 		              a1_pack,

frame/3/trmm/bli_trmm_blk_var3f.c

@@ -113,9 +113,6 @@ void bli_trmm_blk_var3f( obj_t*  a,
 		               cntl_sub_packm_b( cntl ),
                        trmm_thread_sub_ipackm( thread ) );
 
-        // Packing must be done before computation
-        thread_ibarrier( thread );
-
 		// Perform trmm subproblem.
 		bli_trmm_int( &BLIS_ONE,
 		              a1_pack,

frame/3/trsm/bli_trsm_blk_var2b.c

@@ -117,9 +117,6 @@ void bli_trsm_blk_var2b( obj_t*  a,
 		               cntl_sub_packm_c( cntl ),
                        trsm_thread_sub_ipackm( thread ) );
 
-        // Packing must be done before computation
-        thread_ibarrier( thread );
-
 		// Perform trsm subproblem.
 		bli_trsm_int( &BLIS_ONE,
 		              a_pack,

frame/3/trsm/bli_trsm_blk_var2f.c

@@ -118,9 +118,6 @@ void bli_trsm_blk_var2f( obj_t*  a,
 		               cntl_sub_packm_c( cntl ),
                        trsm_thread_sub_ipackm( thread ) );
 
-        // Packing must be done before computation
-        thread_ibarrier( thread );
-
 		// Perform trsm subproblem.
 		bli_trsm_int( &BLIS_ONE,
 		              a_pack,

frame/3/trsm/bli_trsm_blk_var3b.c

@@ -114,9 +114,6 @@ void bli_trsm_blk_var3b( obj_t*  a,
 		               cntl_sub_packm_b( cntl ),
                        trsm_thread_sub_ipackm( thread ) );
 
-        // Packing must be done before computation
-        thread_ibarrier( thread );
-
 		// Perform trsm subproblem.
 		bli_trsm_int( &BLIS_ONE,
 		              a1_pack,

frame/3/trsm/bli_trsm_blk_var3f.c

@@ -114,9 +114,6 @@ void bli_trsm_blk_var3f( obj_t*  a,
 		               cntl_sub_packm_b( cntl ),
                        trsm_thread_sub_ipackm( thread ) );
 
-        // Packing must be done before computation
-        thread_ibarrier( thread );
-
 		// Perform trsm subproblem.
 		bli_trsm_int( &BLIS_ONE,
 		              a1_pack,