libflame revision_anchor
Functions
FLA_Hemm_ll_blk_var4.c File Reference

(r)

Functions

FLA_Error FLA_Hemm_ll_blk_var4 (FLA_Obj alpha, FLA_Obj A, FLA_Obj B, FLA_Obj beta, FLA_Obj C, fla_hemm_t *cntl)
 

Function Documentation

◆ FLA_Hemm_ll_blk_var4()

FLA_Error FLA_Hemm_ll_blk_var4 ( FLA_Obj  alpha,
FLA_Obj  A,
FLA_Obj  B,
FLA_Obj  beta,
FLA_Obj  C,
fla_hemm_t cntl 
)
14{
15 FLA_Obj ATL, ATR, A00, A01, A02,
16 ABL, ABR, A10, A11, A12,
17 A20, A21, A22;
18
19 FLA_Obj BT, B0,
20 BB, B1,
21 B2;
22
23 FLA_Obj CT, C0,
24 CB, C1,
25 C2;
26
27 dim_t b;
28
30 FLA_Cntl_sub_scal( cntl ) );
31
32 FLA_Part_2x2( A, &ATL, &ATR,
33 &ABL, &ABR, 0, 0, FLA_TL );
34
35 FLA_Part_2x1( B, &BT,
36 &BB, 0, FLA_TOP );
37
38 FLA_Part_2x1( C, &CT,
39 &CB, 0, FLA_TOP );
40
41 while ( FLA_Obj_length( ATL ) < FLA_Obj_length( A ) ){
42
44
45 FLA_Repart_2x2_to_3x3( ATL, /**/ ATR, &A00, /**/ &A01, &A02,
46 /* ************* */ /* ******************** */
47 &A10, /**/ &A11, &A12,
48 ABL, /**/ ABR, &A20, /**/ &A21, &A22,
49 b, b, FLA_BR );
50
52 /* ** */ /* ** */
53 &B1,
54 BB, &B2, b, FLA_BOTTOM );
55
57 /* ** */ /* ** */
58 &C1,
59 CB, &C2, b, FLA_BOTTOM );
60
61 /*------------------------------------------------------------*/
62
63 /* C0 = C0 + A10' * B1 */
65 alpha, A10, B1, FLA_ONE, C0,
66 FLA_Cntl_sub_gemm1( cntl ) );
67
68 /* C1 = C1 + A11 * B1 */
70 alpha, A11, B1, FLA_ONE, C1,
71 FLA_Cntl_sub_hemm( cntl ) );
72
73 /* C2 = C2 + A21 * B1 */
75 alpha, A21, B1, FLA_ONE, C2,
76 FLA_Cntl_sub_gemm2( cntl ) );
77
78 /*------------------------------------------------------------*/
79
80 FLA_Cont_with_3x3_to_2x2( &ATL, /**/ &ATR, A00, A01, /**/ A02,
81 A10, A11, /**/ A12,
82 /* ************** */ /* ****************** */
83 &ABL, /**/ &ABR, A20, A21, /**/ A22,
84 FLA_TL );
85
87 B1,
88 /* ** */ /* ** */
89 &BB, B2, FLA_TOP );
90
92 C1,
93 /* ** */ /* ** */
94 &CB, C2, FLA_TOP );
95
96 }
97
98 return FLA_SUCCESS;
99}
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_Hemm_internal(FLA_Side side, FLA_Uplo uplo, FLA_Obj alpha, FLA_Obj A, FLA_Obj B, FLA_Obj beta, FLA_Obj C, fla_hemm_t *cntl)
Definition FLA_Hemm_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_3x3_to_2x2(FLA_Obj *ATL, FLA_Obj *ATR, FLA_Obj A00, FLA_Obj A01, FLA_Obj A02, FLA_Obj A10, FLA_Obj A11, FLA_Obj A12, FLA_Obj *ABL, FLA_Obj *ABR, FLA_Obj A20, FLA_Obj A21, FLA_Obj A22, FLA_Quadrant quadrant)
Definition FLA_View.c:304
FLA_Error FLA_Part_2x2(FLA_Obj A, FLA_Obj *A11, FLA_Obj *A12, FLA_Obj *A21, FLA_Obj *A22, dim_t mb, dim_t nb, FLA_Quadrant quadrant)
Definition FLA_View.c:17
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
FLA_Error FLA_Repart_2x2_to_3x3(FLA_Obj ATL, FLA_Obj ATR, FLA_Obj *A00, FLA_Obj *A01, FLA_Obj *A02, FLA_Obj *A10, FLA_Obj *A11, FLA_Obj *A12, FLA_Obj ABL, FLA_Obj ABR, FLA_Obj *A20, FLA_Obj *A21, FLA_Obj *A22, dim_t mb, dim_t nb, FLA_Quadrant quadrant)
Definition FLA_View.c:142
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
int i
Definition bl1_axmyv2.c:145
Definition FLA_type_defs.h:159

References FLA_Cont_with_3x1_to_2x1(), FLA_Cont_with_3x3_to_2x2(), FLA_Determine_blocksize(), FLA_Gemm_internal(), FLA_Hemm_internal(), FLA_Obj_length(), FLA_ONE, FLA_Part_2x1(), FLA_Part_2x2(), FLA_Repart_2x1_to_3x1(), FLA_Repart_2x2_to_3x3(), FLA_Scal_internal(), and i.

Referenced by FLA_Hemm_ll().