void matrix_set_many_on_column(matrix_type * matrix , int row_offset , int elements , const double * data , int column) {
if ((row_offset + elements) <= matrix->rows) {
if (matrix->row_stride == 1) /* Memory is continous ... */
memcpy( &matrix->data[ GET_INDEX( matrix , row_offset , column) ] , data , elements * sizeof * data);
else {
int i;
for (i = 0; i < elements; i++)
matrix->data[ row_offset + GET_INDEX( matrix , i , column) ] = data[i];
}
} else
util_abort("%s: range violation \n" , __func__);
}
double matrix_det2( const matrix_type * A) {
if ((A->rows == 2) && (A->columns == 2)) {
double a00 = A->data[GET_INDEX(A,0,0)];
double a01 = A->data[GET_INDEX(A,0,1)];
double a10 = A->data[GET_INDEX(A,1,0)];
double a11 = A->data[GET_INDEX(A,1,1)];
return a00 * a11 - a10 * a01;
} else {
util_abort("%s: hardcoded for 2x2 matrices A is: %d x %d \n",__func__, A->rows , A->columns);
return 0;
}
}
double matrix_row_column_dot_product(const matrix_type * m1 , int row1 , const matrix_type * m2 , int col2) {
if (m1->columns != m2->rows)
util_abort("%s: size mismatch: m1:[%d,%d] m2:[%d,%d] \n",__func__ , matrix_get_rows( m1 ) , matrix_get_columns( m1 ) , matrix_get_rows( m2 ) , matrix_get_columns( m2 ));
{
int k;
double sum = 0;
for( k = 0; k < m1->columns; k++)
sum += m1->data[ GET_INDEX(m1 , row1 , k) ] * m2->data[ GET_INDEX(m2, k , col2) ];
return sum;
}
}
bool matrix_columns_equal( const matrix_type * m1 , int col1 , const matrix_type * m2 , int col2) {
if (m1->rows != m2->rows)
return false;
{
int row;
for (row=0; row < m1->rows; row++) {
if (memcmp( &m1->data[ GET_INDEX(m1 , row , col1)] , &m2->data[ GET_INDEX(m2 , row , col2)] , sizeof * m1->data) != 0)
return false;
}
}
return true;
}
void matrix_transpose(const matrix_type * A , matrix_type * T) {
if ((A->columns == T->rows) && (A->rows == T->columns)) {
int i,j;
for (i=0; i < A->rows; i++) {
for (j=0; j < A->columns; j++) {
size_t src_index = GET_INDEX(A , i , j );
size_t target_index = GET_INDEX(T , j , i );
T->data[ target_index ] = A->data[ src_index ];
}
}
} else
util_abort("%s: size mismatch\n",__func__);
}
int reset_pnodes(int curr, int pnode)
{
struct msm_bus_inode_info *info;
struct msm_bus_fabric_device *fabdev;
int index, next_pnode;
fabdev = msm_bus_get_fabric_device(GET_FABID(curr));
index = GET_INDEX(pnode);
info = fabdev->algo->find_node(fabdev, curr);
if (!info) {
MSM_BUS_ERR("Cannot find node info!\n");
return -ENXIO;
}
MSM_BUS_DBG("Starting the loop--remove\n");
do {
struct msm_bus_inode_info *hop;
fabdev = msm_bus_get_fabric_device(GET_FABID(curr));
if (!fabdev) {
MSM_BUS_ERR("Fabric not found\n");
return -ENXIO;
}
next_pnode = info->pnode[index].next;
info->pnode[index].next = -2;
curr = GET_NODE(next_pnode);
index = GET_INDEX(next_pnode);
if (IS_NODE(curr))
hop = fabdev->algo->find_node(fabdev, curr);
else
hop = fabdev->algo->find_gw_node(fabdev, curr);
if (!hop) {
MSM_BUS_ERR("Null Info found for hop\n");
return -ENXIO;
}
MSM_BUS_DBG("%d[%d] = %d\n", info->node_info->priv_id, index,
info->pnode[index].next);
MSM_BUS_DBG("num_pnodes: %d: %d\n", info->node_info->priv_id,
info->num_pnodes);
info = hop;
} while (GET_NODE(info->pnode[index].next) != info->node_info->priv_id);
info->pnode[index].next = -2;
MSM_BUS_DBG("%d[%d] = %d\n", info->node_info->priv_id, index,
info->pnode[index].next);
MSM_BUS_DBG("num_pnodes: %d: %d\n", info->node_info->priv_id,
info->num_pnodes);
return 0;
}
// Comment
void matrix_inplace_diag_sqrt(matrix_type *Cd)
{
int nrows = Cd->rows;
if (Cd->rows != Cd->columns) {
util_abort("%s: size mismatch \n",__func__);
}
else{
int i;
for ( i=0; i<nrows; i++)
{
Cd->data[GET_INDEX(Cd , i , i)] = sqrt(Cd->data[GET_INDEX(Cd , i , i)]);
}
}
}
double matrix_column_column_dot_product(const matrix_type * m1 , int col1 , const matrix_type * m2 , int col2) {
if (m1->rows != m2->rows)
util_abort("%s: size mismatch \n",__func__);
if (col1 >= m1->columns || col2 >= m2->columns)
util_abort("%s: size mismatch \n",__func__);
{
int row;
double sum = 0;
for( row = 0; row < m1->rows; row++)
sum += m1->data[ GET_INDEX(m1 , row , col1) ] * m2->data[ GET_INDEX(m2, row , col2) ];
return sum;
}
}
/* Updates matrix A by subtracting matrix B - elementwise. */
void matrix_inplace_sub(matrix_type * A , const matrix_type * B) {
if ((A->rows == B->rows) && (A->columns == B->columns)) {
int i,j;
for (j = 0; j < A->columns; j++)
for (i=0; i < A->rows; i++)
A->data[ GET_INDEX(A,i,j) ] -= B->data[ GET_INDEX(B,i,j) ];
} else
util_abort("%s: size mismatch A:[%d,%d] B:[%d,%d]\n",__func__ ,
matrix_get_rows(A),
matrix_get_columns(A),
matrix_get_rows(B),
matrix_get_columns(B));
}
/*****************************************
* Read nappe characterization in a file *
*****************************************/
GEO *
file_geo_nappe (BYTE Type, FILE *File)
{
GEO_NAPPE *Geo;
PNT *Pnt, *PntA, *PntB, *PntC, *PntD;
FCT *Fct;
VECTOR U, V;
REAL Real;
INDEX Index;
INIT_MEM (Geo, 1, GEO_NAPPE);
Geo->Type = Type;
GET_INDEX (Geo->NbrPnt);
INIT_MEM (Geo->TabPnt, Geo->NbrPnt, PNT);
GET_INDEX (Geo->NbrFct);
INIT_MEM (Geo->TabFct, Geo->NbrFct, FCT);
Geo->Min.x = Geo->Min.y = Geo->Min.z = INFINITY;
Geo->Max.x = Geo->Max.y = Geo->Max.z = -INFINITY;
for (Index = 0, Pnt = Geo->TabPnt; Index < Geo->NbrPnt; Index++, Pnt++) {
GET_VECTOR (Pnt->Point);
VEC_MIN (Geo->Min, Pnt->Point);
VEC_MAX (Geo->Max, Pnt->Point);
}
for (Index = 0, Fct = Geo->TabFct; Index < Geo->NbrFct; Index++, Fct++) {
if (fscanf (File, " ( %d %d %d %d )", &Fct->i, &Fct->j, &Fct->k, &Fct->l) < 4)
return (FALSE);
Fct->NumFct = Index;
PntA = Geo->TabPnt + Fct->i;
PntB = Geo->TabPnt + Fct->j;
PntC = Geo->TabPnt + Fct->k;
PntD = Geo->TabPnt + Fct->l;
VEC_SUB (U, PntC->Point, PntA->Point);
VEC_SUB (V, PntD->Point, PntB->Point);
VEC_CROSS (Fct->Normal, U, V);
VEC_UNIT (Fct->Normal, Real);
VEC_INC (PntA->Normal, Fct->Normal);
VEC_INC (PntB->Normal, Fct->Normal);
VEC_INC (PntC->Normal, Fct->Normal);
VEC_INC (PntD->Normal, Fct->Normal);
}
for (Index = 0, Pnt = Geo->TabPnt; Index < Geo->NbrPnt; Index++, Pnt++)
VEC_UNIT (Pnt->Normal, Real);
return ((GEO *) Geo);
}
U_CAPI UBiDiDirection U_EXPORT2
ubidi_getVisualRun(UBiDi *pBiDi, int32_t runIndex,
int32_t *pLogicalStart, int32_t *pLength)
{
int32_t start;
UErrorCode errorCode = U_ZERO_ERROR;
RETURN_IF_NOT_VALID_PARA_OR_LINE(pBiDi, errorCode, UBIDI_LTR);
ubidi_getRuns(pBiDi, &errorCode);
if(U_FAILURE(errorCode)) {
return UBIDI_LTR;
}
RETURN_IF_BAD_RANGE(runIndex, 0, pBiDi->runCount, errorCode, UBIDI_LTR);
start=pBiDi->runs[runIndex].logicalStart;
if(pLogicalStart!=NULL) {
*pLogicalStart=GET_INDEX(start);
}
if(pLength!=NULL) {
if(runIndex>0) {
*pLength=pBiDi->runs[runIndex].visualLimit-
pBiDi->runs[runIndex-1].visualLimit;
} else {
*pLength=pBiDi->runs[0].visualLimit;
}
}
return (UBiDiDirection)GET_ODD_BIT(start);
}
static void output_line(FILE *out_file, DATA *data, DPOINT *where,
double *est, int n_outfl) {
int i;
assert(out_file != NULL);
assert(out_file != stdout);
if (data->mode & X_BIT_SET) {
fprintf(out_file, " ");
fprintf(out_file, gl_format, where->x);
}
if (data->mode & Y_BIT_SET) {
fprintf(out_file, " ");
fprintf(out_file, gl_format, where->y);
}
if (data->mode & Z_BIT_SET) {
fprintf(out_file, " ");
fprintf(out_file, gl_format, where->z);
}
if (data->mode & S_BIT_SET)
fprintf(out_file, " %d", where->u.stratum + strata_min);
if (data->colnvalue > 0) {
fprintf(out_file, " ");
fprintf(out_file, gl_format, where->attr);
}
for (i = 0; i < n_outfl; i++)
fprintf(out_file, " %s", my_dtoa(gl_format, &(est[i])));
if (data->point_ids)
fprintf(out_file, " %s",data->point_ids[GET_INDEX(where)]);
fprintf(out_file, "\n");
return;
}
double matrix_get_column_abssum(const matrix_type * matrix , int column) {
double sum = 0;
int i;
for (i=0; i < matrix->rows; i++)
sum += fabs( matrix->data[ GET_INDEX( matrix , i , column ) ] );
return sum;
}
double matrix_get_row_sum(const matrix_type * matrix , int row) {
double sum = 0;
int j;
for (j=0; j < matrix->columns; j++)
sum += matrix->data[ GET_INDEX( matrix , row , j ) ];
return sum;
}
double run_experiment_ji ( mtype_t * matrix , mtype_t scalar , int buffer_size ) {
int i, j, iter;
hptimer_t start , end;
double final_time ;
start = get_time ();
for ( iter = 0; iter < MAX_ITERS ; ++ iter ) {
for (j = 0; j < buffer_size ; ++j ) {
for (i = 0; i < buffer_size ; ++i ) {
matrix [ GET_INDEX (i, j, buffer_size ) ] = scalar * matrix [ GET_INDEX (i, j, buffer_size ) ];
}
}
}
end = get_time ();
return final_time = diff_timers (start , end) / MAX_ITERS ;
}
static
int
choice( void )
{
int i, n;
rb->yield();
for( n = i = 0 ; i < 81 ; ++i )
if( IS_EMPTY( i ) )
{
possible[ n ] = SET_INDEX( i ) | SET_DIGIT( numset( board[ i ] ) );
/* Inconsistency if square unknown, but nothing possible */
if( 9 == GET_DIGIT( possible[ n ] ) )
return -2;
++n;
}
if( 0 == n )
return -1; /* All squares known */
rb->qsort( possible, n, sizeof( possible[ 0 ] ), cmp );
return GET_INDEX( possible[ 0 ] );
}
/***************************************************************
* Ensures the velocity field is non divergent i.e. incompressible:
* Solves a poisson equation to compute a gradient field and then
* subtracts this from the current velocity field to obtain an
* incompressible field. When solving for pressure it sets the
* cells pressure field.
***************************************************************/
void FGS_Fluid_Solver2DS :: project (FLUID_DATA ux, FLUID_DATA uy, FLUID_DATA pressure, FLUID_DATA divergence){
double h = 1.0 / N * -0.5;
double half_width = width * 0.5;
double half_height = height * 0.5;
int cell;
double north, south, east, west;
for (int i=1 ; i <= width; i++ )
for (int j=1 ; j <= height; j++ ){
cell = GET_INDEX(i,j);
north = grid[GET_INDEX(i,j-1)].data[uy];
south = grid[GET_INDEX(i,j+1)].data[uy];
east = grid[GET_INDEX(i+1,j)].data[ux];
west = grid[GET_INDEX(i-1,j)].data[ux];
grid[cell].data[divergence] = h * (east-west+south-north);
grid[cell].data[pressure] = 0;
}
set_boundary (SET_FOR_NON_VELOCITY_COMPONENT, divergence);
set_boundary (SET_FOR_NON_VELOCITY_COMPONENT, pressure);
computing_pressure = true;
linear_solve (SET_FOR_NON_VELOCITY_COMPONENT, pressure, divergence, 1, 0.25);
computing_pressure = false;
for (int i = 1; i <= width; i++)
for (int j = 1; j <= height; j++){
cell = GET_INDEX(i,j);
north = grid[GET_INDEX(i,j-1)].data[pressure];
south = grid[GET_INDEX(i,j+1)].data[pressure];
east = grid[GET_INDEX(i+1,j)].data[pressure];
west = grid[GET_INDEX(i-1,j)].data[pressure];
double u_in_cell = grid[cell].data[ux],
v_in_cell = grid[cell].data[uy];
grid[cell].data[ux] = u_in_cell - (half_width * (east - west));
grid[cell].data[uy] = v_in_cell - (half_height * (south - north));
}
set_boundary(SET_FOR_HORIZONTAL_COMPONENT, ux);
set_boundary(SET_FOR_VERTICAL_COMPONENT, uy);
}
/*
* Prevents PCI Express ASPM (Active State Power Management) being enabled.
*
* Save the register offset, where the ASPM control bits are located,
* for each PCI Express device that is in the device list of
* the root port in an array for fast indexing. Replace the bus ops
* with the modified one.
*/
static void pcie_rootport_aspm_quirk(struct pci_dev *pdev)
{
int i;
struct pci_bus *pbus;
struct pci_dev *dev;
if ((pbus = pdev->subordinate) == NULL)
return;
/*
* Check if the DID of pdev matches one of the six root ports. This
* check is needed in the case this function is called directly by the
* hot-plug driver.
*/
if ((pdev->device < PCI_DEVICE_ID_INTEL_MCH_PA) ||
(pdev->device > PCI_DEVICE_ID_INTEL_MCH_PC1))
return;
if (list_empty(&pbus->devices)) {
/*
* If no device is attached to the root port at power-up or
* after hot-remove, the pbus->devices is empty and this code
* will set the offsets to zero and the bus ops to parent's bus
* ops, which is unmodified.
*/
for (i = GET_INDEX(pdev->device, 0); i <= GET_INDEX(pdev->device, 7); ++i)
quirk_aspm_offset[i] = 0;
pci_bus_set_ops(pbus, pbus->parent->ops);
} else {
/*
* If devices are attached to the root port at power-up or
* after hot-add, the code loops through the device list of
* each root port to save the register offsets and replace the
* bus ops.
*/
list_for_each_entry(dev, &pbus->devices, bus_list)
/* There are 0 to 8 devices attached to this bus */
quirk_aspm_offset[GET_INDEX(pdev->device, dev->devfn)] =
dev->pcie_cap + PCI_EXP_LNKCTL;
pci_bus_set_ops(pbus, &quirk_pcie_aspm_ops);
dev_info(&pbus->dev, "writes to ASPM control bits will be ignored\n");
}
}
/***************************************************************
* Computes a Gauss seidel relaxation to solve AX=B for the
* unknowns X using forward substitution (the higher the number
* of iterations the closer the resultant vector X will be
* to convergence).
***************************************************************/
void FGS_Fluid_Solver2DS :: linear_solve (BOUNDARY_CONDITION b, FLUID_DATA to, FLUID_DATA from, double a, double coef){
for (int k = 0 ; k < iterations; k++){
for (int i = 1 ; i <= width; i++){
for (int j = 1 ; j <= height; j++){
int cell_ij_index = GET_INDEX(i, j);
double w = grid[GET_INDEX(i-1, j)].data[to], e = grid[GET_INDEX(i+1, j)].data[to],
n = grid[GET_INDEX(i, j-1)].data[to], s = grid[GET_INDEX(i, j+1)].data[to],
c0 = grid[cell_ij_index].data[from];
grid[cell_ij_index].data[to] = coef * (c0 + a * (w+e+s+n));
if (computing_pressure) grid[cell_ij_index].data[PRESSURE] = grid[cell_ij_index].data[to];
}
}
set_boundary(b, to);
}
}
void matrix_copy_row(matrix_type * target_matrix, const matrix_type * src_matrix , int target_row, int src_row) {
matrix_assert_equal_columns( target_matrix , src_matrix );
{
int col;
for(col = 0; col < target_matrix->columns; col++)
target_matrix->data[ GET_INDEX( target_matrix , target_row , col)] = src_matrix->data[ GET_INDEX( src_matrix, src_row, col)];
}
}
void matrix_copy_column(matrix_type * target_matrix, const matrix_type * src_matrix , int target_column, int src_column) {
matrix_assert_equal_rows( target_matrix , src_matrix );
{
int row;
for(row = 0; row < target_matrix->rows; row++)
target_matrix->data[ GET_INDEX( target_matrix, row , target_column)] = src_matrix->data[ GET_INDEX( src_matrix, row, src_column)];
}
}
/**
* add_path_node: Adds the path information to the current node
* @info: Internal node info structure
* @next: Combination of the id and index of the next node
* Function returns: Number of pnodes (path_nodes) on success,
* error on failure.
*
* Every node maintains the list of path nodes. A path node is
* reached by finding the node-id and index stored at the current
* node. This makes updating the paths with requested bw and clock
* values efficient, as it avoids lookup for each update-path request.
*/
static int add_path_node(struct msm_bus_inode_info *info, int next)
{
struct path_node *pnode;
int i;
if (ZERO_OR_NULL_PTR(info)) {
MSM_BUS_ERR("Cannot find node info!: id :%d\n",
info->node_info->priv_id);
return -ENXIO;
}
for (i = 0; i <= info->num_pnodes; i++) {
if (info->pnode[i].next == -2) {
MSM_BUS_DBG("Reusing pnode for info: %d at index: %d\n",
info->node_info->priv_id, i);
info->pnode[i].clk[DUAL_CTX] = 0;
info->pnode[i].clk[ACTIVE_CTX] = 0;
info->pnode[i].bw[DUAL_CTX] = 0;
info->pnode[i].bw[ACTIVE_CTX] = 0;
info->pnode[i].next = next;
MSM_BUS_DBG("%d[%d] : (%d, %d)\n",
info->node_info->priv_id, i, GET_NODE(next),
GET_INDEX(next));
return i;
}
}
info->num_pnodes++;
pnode = krealloc(info->pnode,
((info->num_pnodes + 1) * sizeof(struct path_node))
, GFP_KERNEL);
if (ZERO_OR_NULL_PTR(pnode)) {
MSM_BUS_ERR("Error creating path node!\n");
info->num_pnodes--;
return -ENOMEM;
}
info->pnode = pnode;
info->pnode[info->num_pnodes].clk[DUAL_CTX] = 0;
info->pnode[info->num_pnodes].clk[ACTIVE_CTX] = 0;
info->pnode[info->num_pnodes].bw[DUAL_CTX] = 0;
info->pnode[info->num_pnodes].bw[ACTIVE_CTX] = 0;
info->pnode[info->num_pnodes].next = next;
MSM_BUS_DBG("%d[%d] : (%d, %d)\n", info->node_info->priv_id,
info->num_pnodes, GET_NODE(next), GET_INDEX(next));
return info->num_pnodes;
}
double matrix_get_row_abssum(const matrix_type * matrix , int row) {
double sum_abs = 0;
int j;
for ( j=0; j < matrix->columns; j++) {
double m = matrix->data[ GET_INDEX( matrix , row , j ) ];
sum_abs += fabs( m );
}
return sum_abs;
}
double matrix_get_row_sum2(const matrix_type * matrix , int row) {
double sum2 = 0;
int j;
for ( j=0; j < matrix->columns; j++) {
double m = matrix->data[ GET_INDEX( matrix , row , j ) ];
sum2 += m*m;
}
return sum2;
}
void matrix_diag_set_scalar(matrix_type * matrix , double value) {
if (matrix->rows == matrix->columns) {
int i;
matrix_set(matrix , 0);
for ( i=0; i < matrix->rows; i++)
matrix->data[ GET_INDEX(matrix , i , i) ] = value;
} else
util_abort("%s: size mismatch \n",__func__);
}
/*************************************************************************************************
* Inlines
*/
INLINE struct rs_resource* rs_resource_get(reshandle_t hdl) {
uint idx = GET_INDEX(hdl);
ASSERT(idx < (uint)g_rs.ress.item_cnt);
struct rs_resource* r = &((struct rs_resource*)g_rs.ress.buffer)[idx];
ASSERT(r->hdl != INVALID_HANDLE);
ASSERT(GET_ID(r->hdl) == GET_ID(hdl));
return r;
}
/**
Return the sum of the squares on column.
*/
double matrix_get_column_sum2(const matrix_type * matrix , int column) {
double sum2 = 0;
int i;
for ( i=0; i < matrix->rows; i++) {
double m = matrix->data[ GET_INDEX( matrix , i , column ) ];
sum2 += m*m;
}
return sum2;
}
void matrix_diag_set(matrix_type * matrix , const double * diag) {
if (matrix->rows == matrix->columns) {
int i;
matrix_set(matrix , 0);
for ( i=0; i < matrix->rows; i++)
matrix->data[ GET_INDEX(matrix , i , i) ] = diag[i];
} else
util_abort("%s: size mismatch \n",__func__);
}
extern int
mem_blocks_pool_get_block_index(const MEMORY_BLOCKS_POOL *pool, int id) {
int real_id = GET_ID(id);
int index = GET_INDEX(id);
if (pool->ids[index] == real_id)
return index;
return -1;
}
/***************************************************************
* Computes for each cell a backwards particle trace: Linearly
* interpolates a virtual particle from the center of each cell
* backwards in time (dt) using the velocity in the cell (Uxy)
* and copies the value (density or velocity) from this
* 'previous' position into the cell. Also computes the average
* density if called from the density step.
***************************************************************/
void FGS_Fluid_Solver2DS :: advect (BOUNDARY_CONDITION b, FLUID_DATA to, FLUID_DATA from, FLUID_DATA ux, FLUID_DATA uy){
int i0, j0, i1, j1;
double x, y, s0, t0,
s1, t1, deltaT0,
nw, ne, se, sw;
deltaT0 = deltaT*N;
for (int i = 1; i <= width; i++) {
for (int j = 1; j <= height; j++) {
int cell_ij_index = GET_INDEX(i,j);
x = i-deltaT0 * grid[cell_ij_index].data[ux];
y = j-deltaT0 * grid[cell_ij_index].data[uy];
if (x < 0.5 ) x = 0.5;
else if (x > width + 0.5 ) x = width+0.5;
if (y < 0.5 ) y = 0.5;
else if (y > height + 0.5) y = height+0.5;
i0 = ((int) x); j0 = ((int) y);
i1 = i0 + 1; j1 = j0 + 1;
s1 = x-i0; s0 = 1-s1;
t1 = y-j0; t0 = 1-t1;
nw = grid[GET_INDEX(i0, j0)].data[from];
ne = grid[GET_INDEX(i1, j0)].data[from];
se = grid[GET_INDEX(i1, j1)].data[from];
sw = grid[GET_INDEX(i0, j1)].data[from];
grid[cell_ij_index].data[to] = s0 * (t0 * nw + t1 * sw) + s1 * (t0 * ne + t1 * se);
if(computing_density)
d_av += grid[cell_ij_index].data[to];
}
}
if(computing_density)
d_av *= inv_dims;
set_boundary (b, to);
}