1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
|
DOUBLE PRECISION FUNCTION DOPBL2( SUBNAM, M, N, KKL, KKU )
*
* -- LAPACK timing routine (version 3.0) --
* Univ. of Tennessee, Univ. of California Berkeley, NAG Ltd.,
* Courant Institute, Argonne National Lab, and Rice University
* June 30, 1999
*
* .. Scalar Arguments ..
CHARACTER*6 SUBNAM
INTEGER KKL, KKU, M, N
* ..
*
* Purpose
* =======
*
* DOPBL2 computes an approximation of the number of floating point
* operations used by a subroutine SUBNAM with the given values
* of the parameters M, N, KL, and KU.
*
* This version counts operations for the Level 2 BLAS.
*
* Arguments
* =========
*
* SUBNAM (input) CHARACTER*6
* The name of the subroutine.
*
* M (input) INTEGER
* The number of rows of the coefficient matrix. M >= 0.
*
* N (input) INTEGER
* The number of columns of the coefficient matrix.
* If the matrix is square (such as in a solve routine) then
* N is the number of right hand sides. N >= 0.
*
* KKL (input) INTEGER
* The lower band width of the coefficient matrix.
* KL is set to max( 0, min( M-1, KKL ) ).
*
* KKU (input) INTEGER
* The upper band width of the coefficient matrix.
* KU is set to max( 0, min( N-1, KKU ) ).
*
* =====================================================================
*
* .. Local Scalars ..
CHARACTER C1
CHARACTER*2 C2
CHARACTER*3 C3
DOUBLE PRECISION ADDS, EK, EM, EN, KL, KU, MULTS
* ..
* .. External Functions ..
LOGICAL LSAME, LSAMEN
EXTERNAL LSAME, LSAMEN
* ..
* .. Intrinsic Functions ..
INTRINSIC MAX, MIN
* ..
* .. Executable Statements ..
*
* Quick return if possible
*
IF( M.LE.0 .OR. .NOT.( LSAME( SUBNAM, 'S' ) .OR. LSAME( SUBNAM,
$ 'D' ) .OR. LSAME( SUBNAM, 'C' ) .OR. LSAME( SUBNAM, 'Z' ) ) )
$ THEN
DOPBL2 = 0
RETURN
END IF
*
C1 = SUBNAM( 1: 1 )
C2 = SUBNAM( 2: 3 )
C3 = SUBNAM( 4: 6 )
MULTS = 0
ADDS = 0
KL = MAX( 0, MIN( M-1, KKL ) )
KU = MAX( 0, MIN( N-1, KKU ) )
EM = M
EN = N
EK = KL
*
* -------------------------------
* Matrix-vector multiply routines
* -------------------------------
*
IF( LSAMEN( 3, C3, 'MV ' ) ) THEN
*
IF( LSAMEN( 2, C2, 'GE' ) ) THEN
*
MULTS = EM*( EN+1.D0 )
ADDS = EM*EN
*
* Assume M <= N + KL and KL < M
* N <= M + KU and KU < N
* so that the zero sections are triangles.
*
ELSE IF( LSAMEN( 2, C2, 'GB' ) ) THEN
*
MULTS = EM*( EN+1.D0 ) - ( EM-1.D0-KL )*( EM-KL ) / 2.D0 -
$ ( EN-1.D0-KU )*( EN-KU ) / 2.D0
ADDS = EM*( EN+1.D0 ) - ( EM-1.D0-KL )*( EM-KL ) / 2.D0 -
$ ( EN-1.D0-KU )*( EN-KU ) / 2.D0
*
ELSE IF( LSAMEN( 2, C2, 'SY' ) .OR. LSAMEN( 2, C2, 'SP' ) .OR.
$ LSAMEN( 3, SUBNAM, 'CHE' ) .OR.
$ LSAMEN( 3, SUBNAM, 'ZHE' ) .OR.
$ LSAMEN( 3, SUBNAM, 'CHP' ) .OR.
$ LSAMEN( 3, SUBNAM, 'ZHP' ) ) THEN
*
MULTS = EM*( EM+1.D0 )
ADDS = EM*EM
*
ELSE IF( LSAMEN( 2, C2, 'SB' ) .OR.
$ LSAMEN( 3, SUBNAM, 'CHB' ) .OR.
$ LSAMEN( 3, SUBNAM, 'ZHB' ) ) THEN
*
MULTS = EM*( EM+1.D0 ) - ( EM-1.D0-EK )*( EM-EK )
ADDS = EM*EM - ( EM-1.D0-EK )*( EM-EK )
*
ELSE IF( LSAMEN( 2, C2, 'TR' ) .OR. LSAMEN( 2, C2, 'TP' ) )
$ THEN
*
MULTS = EM*( EM+1.D0 ) / 2.D0
ADDS = ( EM-1.D0 )*EM / 2.D0
*
ELSE IF( LSAMEN( 2, C2, 'TB' ) ) THEN
*
MULTS = EM*( EM+1.D0 ) / 2.D0 -
$ ( EM-EK-1.D0 )*( EM-EK ) / 2.D0
ADDS = ( EM-1.D0 )*EM / 2.D0 -
$ ( EM-EK-1.D0 )*( EM-EK ) / 2.D0
*
END IF
*
* ---------------------
* Matrix solve routines
* ---------------------
*
ELSE IF( LSAMEN( 3, C3, 'SV ' ) ) THEN
*
IF( LSAMEN( 2, C2, 'TR' ) .OR. LSAMEN( 2, C2, 'TP' ) ) THEN
*
MULTS = EM*( EM+1.D0 ) / 2.D0
ADDS = ( EM-1.D0 )*EM / 2.D0
*
ELSE IF( LSAMEN( 2, C2, 'TB' ) ) THEN
*
MULTS = EM*( EM+1.D0 ) / 2.D0 -
$ ( EM-EK-1.D0 )*( EM-EK ) / 2.D0
ADDS = ( EM-1.D0 )*EM / 2.D0 -
$ ( EM-EK-1.D0 )*( EM-EK ) / 2.D0
*
END IF
*
* ----------------
* Rank-one updates
* ----------------
*
ELSE IF( LSAMEN( 3, C3, 'R ' ) ) THEN
*
IF( LSAMEN( 3, SUBNAM, 'SGE' ) .OR.
$ LSAMEN( 3, SUBNAM, 'DGE' ) ) THEN
*
MULTS = EM*EN + MIN( EM, EN )
ADDS = EM*EN
*
ELSE IF( LSAMEN( 2, C2, 'SY' ) .OR. LSAMEN( 2, C2, 'SP' ) .OR.
$ LSAMEN( 3, SUBNAM, 'CHE' ) .OR.
$ LSAMEN( 3, SUBNAM, 'CHP' ) .OR.
$ LSAMEN( 3, SUBNAM, 'ZHE' ) .OR.
$ LSAMEN( 3, SUBNAM, 'ZHP' ) ) THEN
*
MULTS = EM*( EM+1.D0 ) / 2.D0 + EM
ADDS = EM*( EM+1.D0 ) / 2.D0
*
END IF
*
ELSE IF( LSAMEN( 3, C3, 'RC ' ) .OR. LSAMEN( 3, C3, 'RU ' ) ) THEN
*
IF( LSAMEN( 3, SUBNAM, 'CGE' ) .OR.
$ LSAMEN( 3, SUBNAM, 'ZGE' ) ) THEN
*
MULTS = EM*EN + MIN( EM, EN )
ADDS = EM*EN
*
END IF
*
* ----------------
* Rank-two updates
* ----------------
*
ELSE IF( LSAMEN( 3, C3, 'R2 ' ) ) THEN
IF( LSAMEN( 2, C2, 'SY' ) .OR. LSAMEN( 2, C2, 'SP' ) .OR.
$ LSAMEN( 3, SUBNAM, 'CHE' ) .OR.
$ LSAMEN( 3, SUBNAM, 'CHP' ) .OR.
$ LSAMEN( 3, SUBNAM, 'ZHE' ) .OR.
$ LSAMEN( 3, SUBNAM, 'ZHP' ) ) THEN
*
MULTS = EM*( EM+1.D0 ) + 2.D0*EM
ADDS = EM*( EM+1.D0 )
*
END IF
END IF
*
* ------------------------------------------------
* Compute the total number of operations.
* For real and double precision routines, count
* 1 for each multiply and 1 for each add.
* For complex and complex*16 routines, count
* 6 for each multiply and 2 for each add.
* ------------------------------------------------
*
IF( LSAME( C1, 'S' ) .OR. LSAME( C1, 'D' ) ) THEN
*
DOPBL2 = MULTS + ADDS
*
ELSE
*
DOPBL2 = 6*MULTS + 2*ADDS
*
END IF
*
RETURN
*
* End of DOPBL2
*
END
|
