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 "QTRMV "
49 #define ERROR_NAME "DTRMV "
51 #define ERROR_NAME "STRMV "
54 static int (*trmv[])(BLASLONG, FLOAT *, BLASLONG, FLOAT *, BLASLONG, FLOAT *) = {
56 qtrmv_NUU, qtrmv_NUN, qtrmv_NLU, qtrmv_NLN,
57 qtrmv_TUU, qtrmv_TUN, qtrmv_TLU, qtrmv_TLN,
59 dtrmv_NUU, dtrmv_NUN, dtrmv_NLU, dtrmv_NLN,
60 dtrmv_TUU, dtrmv_TUN, dtrmv_TLU, dtrmv_TLN,
62 strmv_NUU, strmv_NUN, strmv_NLU, strmv_NLN,
63 strmv_TUU, strmv_TUN, strmv_TLU, strmv_TLN,
68 static int (*trmv_thread[])(BLASLONG, FLOAT *, BLASLONG, FLOAT *, BLASLONG, FLOAT *, int) = {
70 qtrmv_thread_NUU, qtrmv_thread_NUN, qtrmv_thread_NLU, qtrmv_thread_NLN,
71 qtrmv_thread_TUU, qtrmv_thread_TUN, qtrmv_thread_TLU, qtrmv_thread_TLN,
73 dtrmv_thread_NUU, dtrmv_thread_NUN, dtrmv_thread_NLU, dtrmv_thread_NLN,
74 dtrmv_thread_TUU, dtrmv_thread_TUN, dtrmv_thread_TLU, dtrmv_thread_TLN,
76 strmv_thread_NUU, strmv_thread_NUN, strmv_thread_NLU, strmv_thread_NLN,
77 strmv_thread_TUU, strmv_thread_TUN, strmv_thread_TLU, strmv_thread_TLN,
84 void NAME(char *UPLO, char *TRANS, char *DIAG,
85 blasint *N, FLOAT *a, blasint *LDA, FLOAT *x, blasint *INCX){
87 char uplo_arg = *UPLO;
88 char trans_arg = *TRANS;
89 char diag_arg = *DIAG;
114 if (trans_arg == 'N') trans = 0;
115 if (trans_arg == 'T') trans = 1;
116 if (trans_arg == 'R') trans = 0;
117 if (trans_arg == 'C') trans = 1;
119 if (diag_arg == 'U') unit = 0;
120 if (diag_arg == 'N') unit = 1;
122 if (uplo_arg == 'U') uplo = 0;
123 if (uplo_arg == 'L') uplo = 1;
127 if (incx == 0) info = 8;
128 if (lda < MAX(1, n)) info = 6;
130 if (unit < 0) info = 3;
131 if (trans < 0) info = 2;
132 if (uplo < 0) info = 1;
135 BLASFUNC(xerbla)(ERROR_NAME, &info, sizeof(ERROR_NAME));
141 void CNAME(enum CBLAS_ORDER order, enum CBLAS_UPLO Uplo,
142 enum CBLAS_TRANSPOSE TransA, enum CBLAS_DIAG Diag,
143 blasint n, FLOAT *a, blasint lda, FLOAT *x, blasint incx) {
145 int trans, uplo, unit;
159 if (order == CblasColMajor) {
160 if (Uplo == CblasUpper) uplo = 0;
161 if (Uplo == CblasLower) uplo = 1;
163 if (TransA == CblasNoTrans) trans = 0;
164 if (TransA == CblasTrans) trans = 1;
165 if (TransA == CblasConjNoTrans) trans = 0;
166 if (TransA == CblasConjTrans) trans = 1;
168 if (Diag == CblasUnit) unit = 0;
169 if (Diag == CblasNonUnit) unit = 1;
173 if (incx == 0) info = 8;
174 if (lda < MAX(1, n)) info = 6;
176 if (unit < 0) info = 3;
177 if (trans < 0) info = 2;
178 if (uplo < 0) info = 1;
181 if (order == CblasRowMajor) {
182 if (Uplo == CblasUpper) uplo = 1;
183 if (Uplo == CblasLower) uplo = 0;
185 if (TransA == CblasNoTrans) trans = 1;
186 if (TransA == CblasTrans) trans = 0;
187 if (TransA == CblasConjNoTrans) trans = 1;
188 if (TransA == CblasConjTrans) trans = 0;
190 if (Diag == CblasUnit) unit = 0;
191 if (Diag == CblasNonUnit) unit = 1;
195 if (incx == 0) info = 8;
196 if (lda < MAX(1, n)) info = 6;
198 if (unit < 0) info = 3;
199 if (trans < 0) info = 2;
200 if (uplo < 0) info = 1;
204 BLASFUNC(xerbla)(ERROR_NAME, &info, sizeof(ERROR_NAME));
214 FUNCTION_PROFILE_START();
216 if (incx < 0 ) x -= (n - 1) * incx;
218 buffer = (FLOAT *)blas_memory_alloc(1);
221 /* nthreads = num_cpu_avail(2);
223 FIXME trmv_thread was found to be broken, see issue 1332 */
229 (trmv[(trans<<2) | (uplo<<1) | unit])(n, a, lda, x, incx, buffer);
234 (trmv_thread[(trans<<2) | (uplo<<1) | unit])(n, a, lda, x, incx, buffer, nthreads);
239 blas_memory_free(buffer);
241 FUNCTION_PROFILE_END(1, n * n / 2 + n, n * n);