void THBlas_swap(long size, real *x, long xStride, real *y, long yStride)
{
if(size == 1)
{
xStride = 1;
yStride = 1;
}
#if USE_CBLAS
if( (size < INT_MAX) && (xStride < INT_MAX) && (yStride < INT_MAX) )
{
#ifdef USE_DOUBLE
cblas_dswap(size, x, xStride, y, yStride);
#else
cblas_sswap(size, x, xStride, y, yStride);
#endif
return;
}
#endif
{
long i;
for(i = 0; i < size; i++)
{
real z = x[i*xStride];
x[i*xStride] = y[i*yStride];
y[i*yStride] = z;
}
}
}
void THBlas_(swap)(long n, real *x, long incx, real *y, long incy)
{
if(n == 1)
{
incx = 1;
incy = 1;
}
#if defined(USE_BLAS) && (defined(TH_REAL_IS_DOUBLE) || defined(TH_REAL_IS_FLOAT))
if( (n <= INT_MAX) && (incx <= INT_MAX) && (incy <= INT_MAX) )
{
int i_n = (int)n;
int i_incx = (int)incx;
int i_incy = (int)incy;
#if defined(TH_REAL_IS_DOUBLE)
cblas_dswap(i_n, x, i_incx, y, i_incy);
#else
cblas_sswap(i_n, x, i_incx, y, i_incy);
#endif
return;
}
#endif
{
long i;
for(i = 0; i < n; i++)
{
real z = x[i*incx];
x[i*incx] = y[i*incy];
y[i*incy] = z;
}
}
}
JNIEXPORT void JNICALL Java_uncomplicate_neanderthal_CBLAS_sswap
(JNIEnv *env, jclass clazz, jint N,
jobject X, jint offsetX, jint incX,
jobject Y, jint offsetY, jint incY) {
float *cX = (float *) (*env)->GetDirectBufferAddress(env, X);
float *cY = (float *) (*env)->GetDirectBufferAddress(env, Y);
cblas_sswap(N, cX + offsetX, incX, cY + offsetY, incY);
};
void bli_sswap( int n, float* x, int incx, float* y, int incy )
{
#ifdef BLIS_ENABLE_CBLAS_INTERFACES
cblas_sswap( n,
x, incx,
y, incy );
#else
F77_sswap( &n,
x, &incx,
y, &incy );
#endif
}
void
test_swap (void) {
const double flteps = 1e-4, dbleps = 1e-6;
{
int N = 1;
float X[] = { 0.539f };
int incX = 1;
float Y[] = { -0.262f };
int incY = -1;
float expected1[] = { -0.262f };
float expected2[] = { 0.539f };
cblas_sswap(N, X, incX, Y, incY);
{
int i;
for (i = 0; i < 1; i++) {
gsl_test_rel(X[i], expected1[i], flteps, "sswap(case 88)");
}
};
{
int i;
for (i = 0; i < 1; i++) {
gsl_test_rel(Y[i], expected2[i], flteps, "sswap(case 89)");
}
};
};
{
int N = 1;
double X[] = { 0.906 };
int incX = 1;
double Y[] = { 0.373 };
int incY = -1;
double expected1[] = { 0.373 };
double expected2[] = { 0.906 };
cblas_dswap(N, X, incX, Y, incY);
{
int i;
for (i = 0; i < 1; i++) {
gsl_test_rel(X[i], expected1[i], dbleps, "dswap(case 90)");
}
};
{
int i;
for (i = 0; i < 1; i++) {
gsl_test_rel(Y[i], expected2[i], dbleps, "dswap(case 91)");
}
};
};
{
int N = 1;
float X[] = { -0.316f, -0.529f };
int incX = 1;
float Y[] = { -0.313f, 0.363f };
int incY = -1;
float expected1[] = { -0.313f, 0.363f };
float expected2[] = { -0.316f, -0.529f };
cblas_cswap(N, X, incX, Y, incY);
{
int i;
for (i = 0; i < 1; i++) {
gsl_test_rel(X[2*i], expected1[2*i], flteps, "cswap(case 92) real");
gsl_test_rel(X[2*i+1], expected1[2*i+1], flteps, "cswap(case 92) imag");
};
};
{
int i;
for (i = 0; i < 1; i++) {
gsl_test_rel(Y[2*i], expected2[2*i], flteps, "cswap(case 93) real");
gsl_test_rel(Y[2*i+1], expected2[2*i+1], flteps, "cswap(case 93) imag");
};
};
};
{
int N = 1;
double X[] = { 0.512, -0.89 };
int incX = 1;
double Y[] = { -0.225, -0.511 };
int incY = -1;
double expected1[] = { -0.225, -0.511 };
double expected2[] = { 0.512, -0.89 };
cblas_zswap(N, X, incX, Y, incY);
{
int i;
for (i = 0; i < 1; i++) {
gsl_test_rel(X[2*i], expected1[2*i], dbleps, "zswap(case 94) real");
gsl_test_rel(X[2*i+1], expected1[2*i+1], dbleps, "zswap(case 94) imag");
};
};
{
int i;
for (i = 0; i < 1; i++) {
gsl_test_rel(Y[2*i], expected2[2*i], dbleps, "zswap(case 95) real");
gsl_test_rel(Y[2*i+1], expected2[2*i+1], dbleps, "zswap(case 95) imag");
};
};
};
{
int N = 1;
float X[] = { 0.336f };
int incX = -1;
float Y[] = { -0.431f };
int incY = 1;
float expected1[] = { -0.431f };
float expected2[] = { 0.336f };
cblas_sswap(N, X, incX, Y, incY);
{
int i;
for (i = 0; i < 1; i++) {
gsl_test_rel(X[i], expected1[i], flteps, "sswap(case 96)");
}
};
{
int i;
for (i = 0; i < 1; i++) {
gsl_test_rel(Y[i], expected2[i], flteps, "sswap(case 97)");
}
};
};
{
int N = 1;
double X[] = { 0.764 };
int incX = -1;
double Y[] = { -0.293 };
int incY = 1;
double expected1[] = { -0.293 };
double expected2[] = { 0.764 };
cblas_dswap(N, X, incX, Y, incY);
{
int i;
for (i = 0; i < 1; i++) {
gsl_test_rel(X[i], expected1[i], dbleps, "dswap(case 98)");
}
};
{
int i;
for (i = 0; i < 1; i++) {
gsl_test_rel(Y[i], expected2[i], dbleps, "dswap(case 99)");
}
};
};
{
int N = 1;
float X[] = { -0.239f, 0.361f };
int incX = -1;
float Y[] = { 0.149f, 0.347f };
int incY = 1;
float expected1[] = { 0.149f, 0.347f };
float expected2[] = { -0.239f, 0.361f };
cblas_cswap(N, X, incX, Y, incY);
{
int i;
for (i = 0; i < 1; i++) {
gsl_test_rel(X[2*i], expected1[2*i], flteps, "cswap(case 100) real");
gsl_test_rel(X[2*i+1], expected1[2*i+1], flteps, "cswap(case 100) imag");
};
};
{
int i;
for (i = 0; i < 1; i++) {
gsl_test_rel(Y[2*i], expected2[2*i], flteps, "cswap(case 101) real");
gsl_test_rel(Y[2*i+1], expected2[2*i+1], flteps, "cswap(case 101) imag");
};
};
};
{
int N = 1;
double X[] = { -0.171, -0.936 };
int incX = -1;
double Y[] = { 0.495, -0.835 };
int incY = 1;
double expected1[] = { 0.495, -0.835 };
double expected2[] = { -0.171, -0.936 };
cblas_zswap(N, X, incX, Y, incY);
{
int i;
for (i = 0; i < 1; i++) {
gsl_test_rel(X[2*i], expected1[2*i], dbleps, "zswap(case 102) real");
gsl_test_rel(X[2*i+1], expected1[2*i+1], dbleps, "zswap(case 102) imag");
};
};
{
int i;
for (i = 0; i < 1; i++) {
gsl_test_rel(Y[2*i], expected2[2*i], dbleps, "zswap(case 103) real");
gsl_test_rel(Y[2*i+1], expected2[2*i+1], dbleps, "zswap(case 103) imag");
};
};
};
{
int N = 1;
float X[] = { -0.405f };
int incX = -1;
float Y[] = { -0.213f };
int incY = -1;
float expected1[] = { -0.213f };
float expected2[] = { -0.405f };
cblas_sswap(N, X, incX, Y, incY);
{
int i;
for (i = 0; i < 1; i++) {
gsl_test_rel(X[i], expected1[i], flteps, "sswap(case 104)");
}
};
{
int i;
for (i = 0; i < 1; i++) {
gsl_test_rel(Y[i], expected2[i], flteps, "sswap(case 105)");
}
};
};
{
int N = 1;
double X[] = { -0.761 };
int incX = -1;
double Y[] = { -0.585 };
int incY = -1;
double expected1[] = { -0.585 };
double expected2[] = { -0.761 };
cblas_dswap(N, X, incX, Y, incY);
{
int i;
for (i = 0; i < 1; i++) {
gsl_test_rel(X[i], expected1[i], dbleps, "dswap(case 106)");
}
};
{
int i;
for (i = 0; i < 1; i++) {
gsl_test_rel(Y[i], expected2[i], dbleps, "dswap(case 107)");
}
};
};
{
int N = 1;
float X[] = { 0.853f, 0.146f };
int incX = -1;
float Y[] = { 0.009f, -0.178f };
int incY = -1;
float expected1[] = { 0.009f, -0.178f };
float expected2[] = { 0.853f, 0.146f };
cblas_cswap(N, X, incX, Y, incY);
{
int i;
for (i = 0; i < 1; i++) {
gsl_test_rel(X[2*i], expected1[2*i], flteps, "cswap(case 108) real");
gsl_test_rel(X[2*i+1], expected1[2*i+1], flteps, "cswap(case 108) imag");
};
};
{
int i;
for (i = 0; i < 1; i++) {
gsl_test_rel(Y[2*i], expected2[2*i], flteps, "cswap(case 109) real");
gsl_test_rel(Y[2*i+1], expected2[2*i+1], flteps, "cswap(case 109) imag");
};
};
};
{
int N = 1;
double X[] = { -0.228, 0.386 };
int incX = -1;
double Y[] = { 0.988, -0.084 };
int incY = -1;
double expected1[] = { 0.988, -0.084 };
double expected2[] = { -0.228, 0.386 };
cblas_zswap(N, X, incX, Y, incY);
{
int i;
for (i = 0; i < 1; i++) {
gsl_test_rel(X[2*i], expected1[2*i], dbleps, "zswap(case 110) real");
gsl_test_rel(X[2*i+1], expected1[2*i+1], dbleps, "zswap(case 110) imag");
};
};
{
int i;
for (i = 0; i < 1; i++) {
gsl_test_rel(Y[2*i], expected2[2*i], dbleps, "zswap(case 111) real");
gsl_test_rel(Y[2*i+1], expected2[2*i+1], dbleps, "zswap(case 111) imag");
};
};
};
}
void sswap_(int *n, float *X, int *incx, float *Y, int *incy)
{
Blasx_Debug_Output("Calling sswap_\n ");
cblas_sswap(*n,X,*incx,Y,*incy);
}
//
// Overloaded function for dispatching to
// * CBLAS backend, and
// * float value-type.
//
inline void swap( const int n, float* x, const int incx, float* y,
const int incy ) {
cblas_sswap( n, x, incx, y, incy );
}
int CORE_ststrf(int M, int N, int IB, int NB,
float *U, int LDU,
float *A, int LDA,
float *L, int LDL,
int *IPIV,
float *WORK, int LDWORK,
int *INFO)
{
static float zzero = 0.0;
static float mzone =-1.0;
float alpha;
int i, j, ii, sb;
int im, ip;
/* Check input arguments */
*INFO = 0;
if (M < 0) {
coreblas_error(1, "Illegal value of M");
return -1;
}
if (N < 0) {
coreblas_error(2, "Illegal value of N");
return -2;
}
if (IB < 0) {
coreblas_error(3, "Illegal value of IB");
return -3;
}
if ((LDU < max(1,NB)) && (NB > 0)) {
coreblas_error(6, "Illegal value of LDU");
return -6;
}
if ((LDA < max(1,M)) && (M > 0)) {
coreblas_error(8, "Illegal value of LDA");
return -8;
}
if ((LDL < max(1,IB)) && (IB > 0)) {
coreblas_error(10, "Illegal value of LDL");
return -10;
}
/* Quick return */
if ((M == 0) || (N == 0) || (IB == 0))
return PLASMA_SUCCESS;
/* Set L to 0 */
memset(L, 0, LDL*N*sizeof(float));
ip = 0;
for (ii = 0; ii < N; ii += IB) {
sb = min(N-ii, IB);
for (i = 0; i < sb; i++) {
im = cblas_isamax(M, &A[LDA*(ii+i)], 1);
IPIV[ip] = ii+i+1;
if (fabsf(A[LDA*(ii+i)+im]) > fabsf(U[LDU*(ii+i)+ii+i])) {
/*
* Swap behind.
*/
cblas_sswap(i, &L[LDL*ii+i], LDL, &WORK[im], LDWORK );
/*
* Swap ahead.
*/
cblas_sswap(sb-i, &U[LDU*(ii+i)+ii+i], LDU, &A[LDA*(ii+i)+im], LDA );
/*
* Set IPIV.
*/
IPIV[ip] = NB + im + 1;
for (j = 0; j < i; j++) {
A[LDA*(ii+j)+im] = zzero;
}
}
if ((*INFO == 0) && (fabsf(A[LDA*(ii+i)+im]) == zzero)
&& (fabsf(U[LDU*(ii+i)+ii+i]) == zzero)) {
*INFO = ii+i+1;
}
alpha = ((float)1. / U[LDU*(ii+i)+ii+i]);
cblas_sscal(M, (alpha), &A[LDA*(ii+i)], 1);
cblas_scopy(M, &A[LDA*(ii+i)], 1, &WORK[LDWORK*i], 1);
cblas_sger(
CblasColMajor, M, sb-i-1,
(mzone), &A[LDA*(ii+i)], 1,
&U[LDU*(ii+i+1)+ii+i], LDU,
&A[LDA*(ii+i+1)], LDA );
ip = ip+1;
}
/*
* Apply the subpanel to the rest of the panel.
*/
if(ii+i < N) {
for(j = ii; j < ii+sb; j++) {
if (IPIV[j] <= NB) {
IPIV[j] = IPIV[j] - ii;
}
}
CORE_sssssm(
NB, N-(ii+sb), M, N-(ii+sb), sb, sb,
&U[LDU*(ii+sb)+ii], LDU,
&A[LDA*(ii+sb)], LDA,
&L[LDL*ii], LDL,
WORK, LDWORK, &IPIV[ii]);
for(j = ii; j < ii+sb; j++) {
if (IPIV[j] <= NB) {
IPIV[j] = IPIV[j] + ii;
}
}
}
}
return PLASMA_SUCCESS;
}
void STARPU_SSWAP(const int n, float *x, const int incx, float *y, const int incy)
{
cblas_sswap(n, x, incx, y, incy);
}