1 /*********************************************************************/
2 /* Copyright 2009, 2010 The University of Texas at Austin. */
3 /* All rights reserved. */
5 /* Redistribution and use in source and binary forms, with or */
6 /* without modification, are permitted provided that the following */
7 /* conditions are met: */
9 /* 1. Redistributions of source code must retain the above */
10 /* copyright notice, this list of conditions and the following */
13 /* 2. Redistributions in binary form must reproduce the above */
14 /* copyright notice, this list of conditions and the following */
15 /* disclaimer in the documentation and/or other materials */
16 /* provided with the distribution. */
18 /* THIS SOFTWARE IS PROVIDED BY THE UNIVERSITY OF TEXAS AT */
19 /* AUSTIN ``AS IS'' AND ANY EXPRESS OR IMPLIED WARRANTIES, */
20 /* INCLUDING, BUT NOT LIMITED TO, THE IMPLIED WARRANTIES OF */
21 /* MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE */
22 /* DISCLAIMED. IN NO EVENT SHALL THE UNIVERSITY OF TEXAS AT */
23 /* AUSTIN OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, */
24 /* INCIDENTAL, SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES */
25 /* (INCLUDING, BUT NOT LIMITED TO, PROCUREMENT OF SUBSTITUTE */
26 /* GOODS OR SERVICES; LOSS OF USE, DATA, OR PROFITS; OR */
27 /* BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY THEORY OF */
28 /* LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT */
29 /* (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT */
30 /* OF THE USE OF THIS SOFTWARE, EVEN IF ADVISED OF THE */
31 /* POSSIBILITY OF SUCH DAMAGE. */
33 /* The views and conclusions contained in the software and */
34 /* documentation are those of the authors and should not be */
35 /* interpreted as representing official policies, either expressed */
36 /* or implied, of The University of Texas at Austin. */
37 /*********************************************************************/
42 #ifdef FUNCTION_PROFILE
43 #include "functable.h"
47 #define ERROR_NAME "QTBMV "
49 #define ERROR_NAME "DTBMV "
51 #define ERROR_NAME "STBMV "
54 static int (*tbmv[])(BLASLONG, BLASLONG, FLOAT *, BLASLONG, FLOAT *, BLASLONG, void *) = {
56 qtbmv_NUU, qtbmv_NUN, qtbmv_NLU, qtbmv_NLN,
57 qtbmv_TUU, qtbmv_TUN, qtbmv_TLU, qtbmv_TLN,
59 dtbmv_NUU, dtbmv_NUN, dtbmv_NLU, dtbmv_NLN,
60 dtbmv_TUU, dtbmv_TUN, dtbmv_TLU, dtbmv_TLN,
62 stbmv_NUU, stbmv_NUN, stbmv_NLU, stbmv_NLN,
63 stbmv_TUU, stbmv_TUN, stbmv_TLU, stbmv_TLN,
68 static int (*tbmv_thread[])(BLASLONG, BLASLONG, FLOAT *, BLASLONG, FLOAT *, BLASLONG, FLOAT *, int) = {
70 qtbmv_thread_NUU, qtbmv_thread_NUN, qtbmv_thread_NLU, qtbmv_thread_NLN,
71 qtbmv_thread_TUU, qtbmv_thread_TUN, qtbmv_thread_TLU, qtbmv_thread_TLN,
73 dtbmv_thread_NUU, dtbmv_thread_NUN, dtbmv_thread_NLU, dtbmv_thread_NLN,
74 dtbmv_thread_TUU, dtbmv_thread_TUN, dtbmv_thread_TLU, dtbmv_thread_TLN,
76 stbmv_thread_NUU, stbmv_thread_NUN, stbmv_thread_NLU, stbmv_thread_NLN,
77 stbmv_thread_TUU, stbmv_thread_TUN, stbmv_thread_TLU, stbmv_thread_TLN,
84 void NAME(char *UPLO, char *TRANS, char *DIAG,
85 blasint *N, blasint *K,
86 FLOAT *a, blasint *LDA, FLOAT *x, blasint *INCX){
88 char uplo_arg = *UPLO;
89 char trans_arg = *TRANS;
90 char diag_arg = *DIAG;
116 if (trans_arg == 'N') trans = 0;
117 if (trans_arg == 'T') trans = 1;
118 if (trans_arg == 'R') trans = 0;
119 if (trans_arg == 'C') trans = 1;
121 if (diag_arg == 'U') unit = 0;
122 if (diag_arg == 'N') unit = 1;
124 if (uplo_arg == 'U') uplo = 0;
125 if (uplo_arg == 'L') uplo = 1;
129 if (incx == 0) info = 9;
130 if (lda < k + 1) info = 7;
133 if (unit < 0) info = 3;
134 if (trans < 0) info = 2;
135 if (uplo < 0) info = 1;
138 BLASFUNC(xerbla)(ERROR_NAME, &info, sizeof(ERROR_NAME));
144 void CNAME(enum CBLAS_ORDER order, enum CBLAS_UPLO Uplo,
145 enum CBLAS_TRANSPOSE TransA, enum CBLAS_DIAG Diag,
146 blasint n, blasint k, FLOAT *a, blasint lda, FLOAT *x, blasint incx) {
148 int trans, uplo, unit;
162 if (order == CblasColMajor) {
163 if (Uplo == CblasUpper) uplo = 0;
164 if (Uplo == CblasLower) uplo = 1;
166 if (TransA == CblasNoTrans) trans = 0;
167 if (TransA == CblasTrans) trans = 1;
168 if (TransA == CblasConjNoTrans) trans = 0;
169 if (TransA == CblasConjTrans) trans = 1;
171 if (Diag == CblasUnit) unit = 0;
172 if (Diag == CblasNonUnit) unit = 1;
176 if (incx == 0) info = 9;
177 if (lda < k + 1) info = 7;
180 if (unit < 0) info = 3;
181 if (trans < 0) info = 2;
182 if (uplo < 0) info = 1;
185 if (order == CblasRowMajor) {
186 if (Uplo == CblasUpper) uplo = 1;
187 if (Uplo == CblasLower) uplo = 0;
189 if (TransA == CblasNoTrans) trans = 1;
190 if (TransA == CblasTrans) trans = 0;
191 if (TransA == CblasConjNoTrans) trans = 1;
192 if (TransA == CblasConjTrans) trans = 0;
194 if (Diag == CblasUnit) unit = 0;
195 if (Diag == CblasNonUnit) unit = 1;
199 if (incx == 0) info = 9;
200 if (lda < k + 1) info = 7;
203 if (unit < 0) info = 3;
204 if (trans < 0) info = 2;
205 if (uplo < 0) info = 1;
209 BLASFUNC(xerbla)(ERROR_NAME, &info, sizeof(ERROR_NAME));
219 FUNCTION_PROFILE_START();
221 if (incx < 0 ) x -= (n - 1) * incx;
223 buffer = (FLOAT *)blas_memory_alloc(1);
226 nthreads = num_cpu_avail(2);
231 (tbmv[(trans<<2) | (uplo<<1) | unit])(n, k, a, lda, x, incx, buffer);
236 (tbmv_thread[(trans<<2) | (uplo<<1) | unit])(n, k, a, lda, x, incx, buffer, nthreads);
241 blas_memory_free(buffer);
243 FUNCTION_PROFILE_END(1, n * k / 2 + n, n * k);