libflame revision_anchor
Functions
FLA_Sylv_nn_blk_var16.c File Reference

(r)

Functions

FLA_Error FLA_Sylv_nn_blk_var16 (FLA_Obj isgn, FLA_Obj A, FLA_Obj B, FLA_Obj C, FLA_Obj scale, fla_sylv_t *cntl)
 

Function Documentation

◆ FLA_Sylv_nn_blk_var16()

FLA_Error FLA_Sylv_nn_blk_var16 ( FLA_Obj  isgn,
FLA_Obj  A,
FLA_Obj  B,
FLA_Obj  C,
FLA_Obj  scale,
fla_sylv_t cntl 
)
16{
17 FLA_Obj ATL, ATR, A00, A01, A02,
18 ABL, ABR, A10, A11, A12,
19 A20, A21, A22;
20
21 FLA_Obj CT, C0,
22 CB, C1,
23 C2;
24
25 dim_t b;
26
27 FLA_Part_2x2( A, &ATL, &ATR,
28 &ABL, &ABR, 0, 0, FLA_BR );
29
30 FLA_Part_2x1( C, &CT,
31 &CB, 0, FLA_BOTTOM );
32
33 while ( FLA_Obj_length( ABR ) < FLA_Obj_length( A ) ){
34
36
37 FLA_Repart_2x2_to_3x3( ATL, /**/ ATR, &A00, &A01, /**/ &A02,
38 &A10, &A11, /**/ &A12,
39 /* ************* */ /* ******************** */
40 ABL, /**/ ABR, &A20, &A21, /**/ &A22,
41 b, b, FLA_TL );
42
44 &C1,
45 /* ** */ /* ** */
46 CB, &C2, b, FLA_TOP );
47
48 // Loop Invariant:
49 // CT = CT - ATR * sylv( ABR, B, CB )
50 // CB = sylv( ABR, B, CB )
51
52 /*------------------------------------------------------------*/
53
54 // C1 = sylv( A11, B, C1 );
56 isgn, A11, B, C1, scale,
57 FLA_Cntl_sub_sylv1( cntl ) );
58
59 // C0 = C0 - A01 * C1;
62 FLA_Cntl_sub_gemm1( cntl ) );
63
64 /*------------------------------------------------------------*/
65
66 FLA_Cont_with_3x3_to_2x2( &ATL, /**/ &ATR, A00, /**/ A01, A02,
67 /* ************** */ /* ****************** */
68 A10, /**/ A11, A12,
69 &ABL, /**/ &ABR, A20, /**/ A21, A22,
70 FLA_BR );
71
73 /* ** */ /* ** */
74 C1,
75 &CB, C2, FLA_BOTTOM );
76
77 }
78
79 return FLA_SUCCESS;
80}
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_Sylv_internal(FLA_Trans transa, FLA_Trans transb, FLA_Obj isgn, FLA_Obj A, FLA_Obj B, FLA_Obj C, FLA_Obj scale, fla_sylv_t *cntl)
Definition FLA_Sylv_internal.c:16
FLA_Obj FLA_MINUS_ONE
Definition FLA_Init.c:22
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_MINUS_ONE, FLA_Obj_length(), FLA_ONE, FLA_Part_2x1(), FLA_Part_2x2(), FLA_Repart_2x1_to_3x1(), FLA_Repart_2x2_to_3x3(), FLA_Sylv_internal(), and i.

Referenced by FLA_Sylv_nn().