libflame revision_anchor
FLA_Gemm_hc_blk_var6.c File Reference

(r)

Functions

FLA_Error FLA_Gemm_hc_blk_var6 (FLA_Obj alpha, FLA_Obj A, FLA_Obj B, FLA_Obj beta, FLA_Obj C, fla_gemm_t *cntl)
 

Function Documentation

◆ FLA_Gemm_hc_blk_var6()

FLA_Error FLA_Gemm_hc_blk_var6 ( FLA_Obj alpha,
FLA_Obj A,
FLA_Obj B,
FLA_Obj beta,
FLA_Obj C,
fla_gemm_t * cntl )
16{
17 FLA_Obj AT, A0,
18 AB, A1,
19 A2;
20
21 FLA_Obj BT, B0,
22 BB, B1,
23 B2;
24
25 dim_t b;
26
27 FLA_Scal_internal( beta, C,
28 FLA_Cntl_sub_scal( cntl ) );
29
30 FLA_Part_2x1( A, &AT,
31 &AB, 0, FLA_BOTTOM );
32
33 FLA_Part_2x1( B, &BT,
34 &BB, 0, FLA_BOTTOM );
35
36 while ( FLA_Obj_length( AB ) < FLA_Obj_length( A ) ){
37
38 b = FLA_Determine_blocksize( AT, FLA_TOP, FLA_Cntl_blocksize( cntl ) );
39
40 FLA_Repart_2x1_to_3x1( AT, &A0,
41 &A1,
42 /* ** */ /* ** */
43 AB, &A2, b, FLA_TOP );
44
45 FLA_Repart_2x1_to_3x1( BT, &B0,
46 &B1,
47 /* ** */ /* ** */
48 BB, &B2, b, FLA_TOP );
49
50 /*------------------------------------------------------------*/
51
52 /* C = alpha * A1' * B1 + C; */
53 FLA_Gemm_internal( FLA_CONJ_TRANSPOSE, FLA_CONJ_NO_TRANSPOSE,
54 alpha, A1, B1, FLA_ONE, C,
55 FLA_Cntl_sub_gemm( cntl ) );
56
57 /*------------------------------------------------------------*/
58
60 /* ** */ /* ** */
61 A1,
62 &AB, A2, FLA_BOTTOM );
63
65 /* ** */ /* ** */
66 B1,
67 &BB, B2, FLA_BOTTOM );
68
69 }
70
71 return FLA_SUCCESS;
72}
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_Scal_internal(FLA_Obj alpha, FLA_Obj A, fla_scal_t *cntl)
Definition FLA_Scal_internal.c:16
FLA_Obj FLA_ONE
Definition FLA_Init.c:18
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_ONE, FLA_Part_2x1(), FLA_Repart_2x1_to_3x1(), and FLA_Scal_internal().

Referenced by FLA_Gemm_hc().