mirror of
https://github.com/lucaspalomodevelop/meshlab.git
synced 2026-03-20 11:26:11 +00:00
177 lines
4.0 KiB
C
177 lines
4.0 KiB
C
/*********************************************************************/
|
|
/* */
|
|
/* Optimized BLAS libraries */
|
|
/* By Kazushige Goto <kgoto@tacc.utexas.edu> */
|
|
/* */
|
|
/* Copyright (c) The University of Texas, 2009. All rights reserved. */
|
|
/* UNIVERSITY EXPRESSLY DISCLAIMS ANY AND ALL WARRANTIES CONCERNING */
|
|
/* THIS SOFTWARE AND DOCUMENTATION, INCLUDING ANY WARRANTIES OF */
|
|
/* MERCHANTABILITY, FITNESS FOR ANY PARTICULAR PURPOSE, */
|
|
/* NON-INFRINGEMENT AND WARRANTIES OF PERFORMANCE, AND ANY WARRANTY */
|
|
/* THAT MIGHT OTHERWISE ARISE FROM COURSE OF DEALING OR USAGE OF */
|
|
/* TRADE. NO WARRANTY IS EITHER EXPRESS OR IMPLIED WITH RESPECT TO */
|
|
/* THE USE OF THE SOFTWARE OR DOCUMENTATION. */
|
|
/* Under no circumstances shall University be liable for incidental, */
|
|
/* special, indirect, direct or consequential damages or loss of */
|
|
/* profits, interruption of business, or related expenses which may */
|
|
/* arise from use of Software or Documentation, including but not */
|
|
/* limited to those resulting from defects in Software and/or */
|
|
/* Documentation, or loss or inaccuracy of data of any kind. */
|
|
/*********************************************************************/
|
|
|
|
#include <stdio.h>
|
|
#include "common.h"
|
|
#ifdef FUNCTION_PROFILE
|
|
#include "functable.h"
|
|
#endif
|
|
|
|
#ifdef XDOUBLE
|
|
#define ERROR_NAME "QGER "
|
|
#elif defined DOUBLE
|
|
#define ERROR_NAME "DGER "
|
|
#else
|
|
#define ERROR_NAME "SGER "
|
|
#endif
|
|
|
|
#define GER GERU_K
|
|
|
|
#if defined XDOUBLE
|
|
#define GER_THREAD qger_thread
|
|
#elif defined DOUBLE
|
|
#define GER_THREAD dger_thread
|
|
#else
|
|
#define GER_THREAD sger_thread
|
|
#endif
|
|
|
|
|
|
#ifndef CBLAS
|
|
|
|
void NAME(blasint *M, blasint *N, FLOAT *Alpha,
|
|
FLOAT *x, blasint *INCX,
|
|
FLOAT *y, blasint *INCY,
|
|
FLOAT *a, blasint *LDA){
|
|
|
|
blasint m = *M;
|
|
blasint n = *N;
|
|
FLOAT alpha = *Alpha;
|
|
blasint incx = *INCX;
|
|
blasint incy = *INCY;
|
|
blasint lda = *LDA;
|
|
FLOAT *buffer;
|
|
#ifdef SMP
|
|
int nthreads;
|
|
#endif
|
|
|
|
blasint info;
|
|
|
|
PRINT_DEBUG_NAME;
|
|
|
|
info = 0;
|
|
|
|
if (lda < MAX(1,m)) info = 9;
|
|
if (incy == 0) info = 7;
|
|
if (incx == 0) info = 5;
|
|
if (n < 0) info = 2;
|
|
if (m < 0) info = 1;
|
|
|
|
if (info){
|
|
BLASFUNC(xerbla)(ERROR_NAME, &info, sizeof(ERROR_NAME));
|
|
return;
|
|
}
|
|
|
|
#else
|
|
|
|
void CNAME(enum CBLAS_ORDER order,
|
|
blasint m, blasint n,
|
|
FLOAT alpha,
|
|
FLOAT *x, blasint incx,
|
|
FLOAT *y, blasint incy,
|
|
FLOAT *a, blasint lda) {
|
|
|
|
FLOAT *buffer;
|
|
blasint info, t;
|
|
#ifdef SMP
|
|
int nthreads;
|
|
#endif
|
|
|
|
PRINT_DEBUG_CNAME;
|
|
|
|
info = 0;
|
|
|
|
if (order == CblasColMajor) {
|
|
info = -1;
|
|
|
|
if (lda < MAX(1,m)) info = 9;
|
|
if (incy == 0) info = 7;
|
|
if (incx == 0) info = 5;
|
|
if (n < 0) info = 2;
|
|
if (m < 0) info = 1;
|
|
}
|
|
|
|
if (order == CblasRowMajor) {
|
|
info = -1;
|
|
|
|
t = n;
|
|
n = m;
|
|
m = t;
|
|
|
|
t = incx;
|
|
incx = incy;
|
|
incy = t;
|
|
|
|
buffer = x;
|
|
x = y;
|
|
y = buffer;
|
|
|
|
if (lda < MAX(1,m)) info = 9;
|
|
if (incy == 0) info = 7;
|
|
if (incx == 0) info = 5;
|
|
if (n < 0) info = 2;
|
|
if (m < 0) info = 1;
|
|
}
|
|
|
|
if (info >= 0) {
|
|
BLASFUNC(xerbla)(ERROR_NAME, &info, sizeof(ERROR_NAME));
|
|
return;
|
|
}
|
|
|
|
#endif
|
|
|
|
/* Quick return if possible. */
|
|
if (m == 0 || n == 0) return;
|
|
if (alpha == 0.) return;
|
|
|
|
IDEBUG_START;
|
|
|
|
FUNCTION_PROFILE_START();
|
|
|
|
if (incy < 0) y -= (n - 1) * incy;
|
|
if (incx < 0) x -= (m - 1) * incx;
|
|
|
|
buffer = (FLOAT *)blas_memory_alloc(1);
|
|
|
|
#ifdef SMP
|
|
nthreads = num_cpu_avail(2);
|
|
|
|
if (nthreads == 1) {
|
|
#endif
|
|
|
|
GER(m, n, 0, alpha, x, incx, y, incy, a, lda, buffer);
|
|
|
|
#ifdef SMP
|
|
} else {
|
|
|
|
GER_THREAD(m, n, alpha, x, incx, y, incy, a, lda, buffer, nthreads);
|
|
|
|
}
|
|
#endif
|
|
|
|
blas_memory_free(buffer);
|
|
|
|
FUNCTION_PROFILE_END(1, m * n + m + n, 2 * m * n);
|
|
|
|
IDEBUG_END;
|
|
|
|
return;
|
|
}
|