libflame  revision_anchor
Functions
FLA_Gemm_ch_blk_var1.c File Reference

(r)

Functions

FLA_Error FLA_Gemm_ch_blk_var1 (FLA_Obj alpha, FLA_Obj A, FLA_Obj B, FLA_Obj beta, FLA_Obj C, fla_gemm_t *cntl)
 

Function Documentation

◆ FLA_Gemm_ch_blk_var1()

FLA_Error FLA_Gemm_ch_blk_var1 ( FLA_Obj  alpha,
FLA_Obj  A,
FLA_Obj  B,
FLA_Obj  beta,
FLA_Obj  C,
fla_gemm_t cntl 
)
14 {
15  FLA_Obj AT, A0,
16  AB, A1,
17  A2;
18 
19  FLA_Obj CT, C0,
20  CB, C1,
21  C2;
22 
23  dim_t b;
24 
25  FLA_Part_2x1( A, &AT,
26  &AB, 0, FLA_TOP );
27 
28  FLA_Part_2x1( C, &CT,
29  &CB, 0, FLA_TOP );
30 
31  while ( FLA_Obj_length( AT ) < FLA_Obj_length( A ) ){
32 
33  b = FLA_Determine_blocksize( AB, FLA_BOTTOM, FLA_Cntl_blocksize( cntl ) );
34 
35  FLA_Repart_2x1_to_3x1( AT, &A0,
36  /* ** */ /* ** */
37  &A1,
38  AB, &A2, b, FLA_BOTTOM );
39 
40  FLA_Repart_2x1_to_3x1( CT, &C0,
41  /* ** */ /* ** */
42  &C1,
43  CB, &C2, b, FLA_BOTTOM );
44 
45  /*------------------------------------------------------------*/
46 
47  /* C1 = alpha * A1 * B' + C1; */
48  FLA_Gemm_internal( FLA_CONJ_NO_TRANSPOSE, FLA_CONJ_TRANSPOSE,
49  alpha, A1, B, beta, C1,
50  FLA_Cntl_sub_gemm( cntl ) );
51 
52  /*------------------------------------------------------------*/
53 
54  FLA_Cont_with_3x1_to_2x1( &AT, A0,
55  A1,
56  /* ** */ /* ** */
57  &AB, A2, FLA_TOP );
58 
59  FLA_Cont_with_3x1_to_2x1( &CT, C0,
60  C1,
61  /* ** */ /* ** */
62  &CB, C2, FLA_TOP );
63 
64  }
65 
66  return FLA_SUCCESS;
67 }
FLA_Error FLA_Gemm_internal(FLA_Trans transa, FLA_Trans transb, FLA_Obj alpha, FLA_Obj A, FLA_Obj B, FLA_Obj beta, FLA_Obj C, fla_gemm_t *cntl)
Definition: FLA_Gemm_internal.c:16
FLA_Error FLA_Cont_with_3x1_to_2x1(FLA_Obj *AT, FLA_Obj A0, FLA_Obj A1, FLA_Obj *AB, FLA_Obj A2, FLA_Side side)
Definition: FLA_View.c:428
FLA_Error FLA_Repart_2x1_to_3x1(FLA_Obj AT, FLA_Obj *A0, FLA_Obj *A1, FLA_Obj AB, FLA_Obj *A2, dim_t mb, FLA_Side side)
Definition: FLA_View.c:226
FLA_Error FLA_Part_2x1(FLA_Obj A, FLA_Obj *A1, FLA_Obj *A2, dim_t mb, FLA_Side side)
Definition: FLA_View.c:76
dim_t FLA_Obj_length(FLA_Obj obj)
Definition: FLA_Query.c:116
dim_t FLA_Determine_blocksize(FLA_Obj A_unproc, FLA_Quadrant to_dir, fla_blocksize_t *cntl_blocksizes)
Definition: FLA_Blocksize.c:234
unsigned long dim_t
Definition: FLA_type_defs.h:71
Definition: FLA_type_defs.h:159

References FLA_Cont_with_3x1_to_2x1(), FLA_Determine_blocksize(), FLA_Gemm_internal(), FLA_Obj_length(), FLA_Part_2x1(), and FLA_Repart_2x1_to_3x1().

Referenced by FLA_Gemm_ch().