libflame  revision_anchor
Functions
FLA_Sylv_hn_blk_var18.c File Reference

(r)

Functions

FLA_Error FLA_Sylv_hn_blk_var18 (FLA_Obj isgn, FLA_Obj A, FLA_Obj B, FLA_Obj C, FLA_Obj scale, fla_sylv_t *cntl)
 

Function Documentation

◆ FLA_Sylv_hn_blk_var18()

FLA_Error FLA_Sylv_hn_blk_var18 ( FLA_Obj  isgn,
FLA_Obj  A,
FLA_Obj  B,
FLA_Obj  C,
FLA_Obj  scale,
fla_sylv_t cntl 
)
14 {
15  FLA_Obj BTL, BTR, B00, B01, B02,
16  BBL, BBR, B10, B11, B12,
17  B20, B21, B22;
18 
19  FLA_Obj CL, CR, C0, C1, C2;
20 
21  dim_t b;
22 
23  FLA_Part_2x2( B, &BTL, &BTR,
24  &BBL, &BBR, 0, 0, FLA_TL );
25 
26  FLA_Part_1x2( C, &CL, &CR, 0, FLA_LEFT );
27 
28  while ( FLA_Obj_length( BTL ) < FLA_Obj_length( B ) ){
29 
30  b = FLA_Determine_blocksize( CR, FLA_RIGHT, FLA_Cntl_blocksize( cntl ) );
31 
32  FLA_Repart_2x2_to_3x3( BTL, /**/ BTR, &B00, /**/ &B01, &B02,
33  /* ************* */ /* ******************** */
34  &B10, /**/ &B11, &B12,
35  BBL, /**/ BBR, &B20, /**/ &B21, &B22,
36  b, b, FLA_BR );
37 
38  FLA_Repart_1x2_to_1x3( CL, /**/ CR, &C0, /**/ &C1, &C2,
39  b, FLA_RIGHT );
40 
41  // Loop Invariant:
42  // CL =
43  // CR =
44 
45  /*------------------------------------------------------------*/
46 
47  // C1 = sylv( A', B11, C1 );
48  FLA_Sylv_internal( FLA_CONJ_TRANSPOSE, FLA_NO_TRANSPOSE,
49  isgn, A, B11, C1, scale,
50  FLA_Cntl_sub_sylv1( cntl ) );
51 
52  // C2 = C2 -/+ C1 * B12;
53  FLA_Gemm_internal( FLA_NO_TRANSPOSE, FLA_NO_TRANSPOSE,
54  FLA_NEGATE( isgn ), C1, B12, FLA_ONE, C2,
55  FLA_Cntl_sub_gemm1( cntl ) );
56 
57  /*------------------------------------------------------------*/
58 
59  FLA_Cont_with_3x3_to_2x2( &BTL, /**/ &BTR, B00, B01, /**/ B02,
60  B10, B11, /**/ B12,
61  /* ************** */ /* ****************** */
62  &BBL, /**/ &BBR, B20, B21, /**/ B22,
63  FLA_TL );
64 
65  FLA_Cont_with_1x3_to_1x2( &CL, /**/ &CR, C0, C1, /**/ C2,
66  FLA_LEFT );
67  }
68 
69  return FLA_SUCCESS;
70 }
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_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_1x3_to_1x2(FLA_Obj *AL, FLA_Obj *AR, FLA_Obj A0, FLA_Obj A1, FLA_Obj A2, FLA_Side side)
Definition: FLA_View.c:475
FLA_Error FLA_Part_1x2(FLA_Obj A, FLA_Obj *A1, FLA_Obj *A2, dim_t nb, FLA_Side side)
Definition: FLA_View.c:110
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
FLA_Error FLA_Repart_1x2_to_1x3(FLA_Obj AL, FLA_Obj AR, FLA_Obj *A0, FLA_Obj *A1, FLA_Obj *A2, dim_t nb, FLA_Side side)
Definition: FLA_View.c:267
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_1x3_to_1x2(), FLA_Cont_with_3x3_to_2x2(), FLA_Determine_blocksize(), FLA_Gemm_internal(), FLA_Obj_length(), FLA_ONE, FLA_Part_1x2(), FLA_Part_2x2(), FLA_Repart_1x2_to_1x3(), FLA_Repart_2x2_to_3x3(), and FLA_Sylv_internal().

Referenced by FLA_Sylv_hn().