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 "XHER2 "
49 #define ERROR_NAME "ZHER2 "
51 #define ERROR_NAME "CHER2 "
54 static int (*her2[])(BLASLONG, FLOAT, FLOAT, FLOAT *, BLASLONG, FLOAT *, BLASLONG, FLOAT *, BLASLONG, FLOAT *) = {
56 xher2_U, xher2_L, xher2_V, xher2_M,
58 zher2_U, zher2_L, zher2_V, zher2_M,
60 cher2_U, cher2_L, cher2_V, cher2_M,
65 static int (*her2_thread[])(BLASLONG, FLOAT *, FLOAT *, BLASLONG, FLOAT *, BLASLONG, FLOAT *, BLASLONG, FLOAT *, int) = {
67 xher2_thread_U, xher2_thread_L, xher2_thread_V, xher2_thread_M,
69 zher2_thread_U, zher2_thread_L, zher2_thread_V, zher2_thread_M,
71 cher2_thread_U, cher2_thread_L, cher2_thread_V, cher2_thread_M,
78 void NAME(char *UPLO, blasint *N, FLOAT *ALPHA,
79 FLOAT *x, blasint *INCX, FLOAT *y, blasint *INCY, FLOAT *a, blasint *LDA){
81 char uplo_arg = *UPLO;
83 FLOAT alpha_r = ALPHA[0];
84 FLOAT alpha_i = ALPHA[1];
101 if (uplo_arg == 'U') uplo = 0;
102 if (uplo_arg == 'L') uplo = 1;
106 if (lda < MAX(1, n)) info = 9;
107 if (incy == 0) info = 7;
108 if (incx == 0) info = 5;
110 if (uplo < 0) info = 1;
113 BLASFUNC(xerbla)(ERROR_NAME, &info, sizeof(ERROR_NAME));
119 void CNAME(enum CBLAS_ORDER order, enum CBLAS_UPLO Uplo, blasint n, void *VALPHA, void *vx, blasint incx, void *vy, blasint incy, void *va, blasint lda) {
121 FLOAT* ALPHA = (FLOAT*) VALPHA;
122 FLOAT* a = (FLOAT*) va;
123 FLOAT* x = (FLOAT*) vx;
124 FLOAT* y = (FLOAT*) vy;
126 FLOAT alpha_r = ALPHA[0];
127 FLOAT alpha_i = ALPHA[1];
140 if (order == CblasColMajor) {
142 if (Uplo == CblasUpper) uplo = 0;
143 if (Uplo == CblasLower) uplo = 1;
147 if (lda < MAX(1, n)) info = 9;
148 if (incy == 0) info = 7;
149 if (incx == 0) info = 5;
151 if (uplo < 0) info = 1;
154 if (order == CblasRowMajor) {
156 if (Uplo == CblasUpper) uplo = 3;
157 if (Uplo == CblasLower) uplo = 2;
161 if (lda < MAX(1, n)) info = 9;
162 if (incx == 0) info = 7;
163 if (incy == 0) info = 5;
165 if (uplo < 0) info = 1;
169 BLASFUNC(xerbla)(ERROR_NAME, &info, sizeof(ERROR_NAME));
177 if ((alpha_r == ZERO) && (alpha_i == ZERO)) return;
181 FUNCTION_PROFILE_START();
183 if (incx < 0 ) x -= (n - 1) * incx * 2;
184 if (incy < 0 ) y -= (n - 1) * incy * 2;
186 buffer = (FLOAT *)blas_memory_alloc(1);
189 nthreads = num_cpu_avail(2);
194 (her2[uplo])(n, alpha_r, alpha_i, x, incx, y, incy, a, lda, buffer);
199 (her2_thread[uplo])(n, ALPHA, x, incx, y, incy, a, lda, buffer, nthreads);
204 blas_memory_free(buffer);
206 FUNCTION_PROFILE_END(4, n * n / 2 + 2 * n, 2 * n * n);