// TODO: Can use dynamic scheduling here
int BuildNeigh( void* arg, int id ) {
double rd, cutoffSquare;
int i,j;
cutoffSquare = (cutoffRadius * TOLERANCE)*(cutoffRadius * TOLERANCE);
for (i = (numMoles / NTHREADS) * id; i < (numMoles / NTHREADS) * (id + 1); ++i) {
for (j = i+1; j<numMoles; j++ ) {
rd = Foo ( (double)x[IND(0,i)], (double)x[IND(1,i)], (double)x[IND(2,i)],
(double)x[IND(0,j)], (double)x[IND(1,j)], (double)x[IND(2,j)]);
if (rd > cutoffSquare) {
continue;
}
BEGIN_TRANSACTION();
//printf("APP: writing i: %x (%x), with value: %d\n", &inter[INDX(ninter,0)], &(inter[INDX(ninter,0)]._value), i);
inter[INDX(ninter,0)] = i;
//printf("APP: writing j: %x (%x), with value: %d\n", &inter[INDX(ninter,1)], &(inter[INDX(ninter,1)]._value), j);
inter[INDX(ninter,1)] = j;
++ninter;
COMMIT_TRANSACTION();
if ( ninter >= MAXINTERACT) {
perror("MAXINTERACT limit");
}
}
}
return 0;
}
void trialswap(int *m, int *nr, int *nc, int *thin)
{
int i, a, b, c, d, row[2], col[2];
GetRNGstate();
for (i=0; i < *thin; i++) {
i2rand(row, (*nr) - 1);
i2rand(col, (*nc) - 1);
a = INDX(row[0], col[0], *nr);
b = INDX(row[0], col[1], *nr);
c = INDX(row[1], col[0], *nr);
d = INDX(row[1], col[1], *nr);
if (m[a] == 1 && m[d] == 1 && m[b] == 0 && m[c] == 0) {
m[a] = 0;
m[d] = 0;
m[b] = 1;
m[c] = 1;
} else if (m[c] == 1 && m[b] == 1 && m[d] == 0 &&
m[a] == 0) {
m[a] = 1;
m[d] = 1;
m[b] = 0;
m[c] = 0;
}
}
PutRNGstate();
}
void PrintInteractionList(INPARAMS int ninter)
{
int i;
printf("%d\n", ninter);
for (i=0;i<ninter;i++) {
printf("%d %d\n", (int)inter[INDX(i,0)], (int)inter[INDX(i,1)]);
}
}
//we should use dynamic scheduling here
int BuildNeigh( void* arg, int id )
{
double rd, cutoffSquare;
int i,j;
//double Foo();
cutoffSquare = (cutoffRadius * TOLERANCE)*(cutoffRadius * TOLERANCE);
for (i = id; i < numMoles; i += NTHREADS) {
num_neighbours [i] = 0;
//resetting these two as well
counter_f [i] = 0;
counter_x [i] = 0;
}
//for ( i=0; i<numMoles; i++) //single-threaded version
for ( i=id; i<numMoles; i+=NTHREADS)
{
//BEGIN_TRANSACTION();
//if (id == 0) {
for ( j = i+1; j<numMoles; j++ )
{
//x is global var, would need exclusive access unless this method is surrounded
//by barriers, because while x is only read here (and not written), other methods
//do write to x.
rd = Foo ( (double)x[IND(0,i)], (double)x[IND(1,i)], (double)x[IND(2,i)],
(double)x[IND(0,j)], (double)x[IND(1,j)], (double)x[IND(2,j)]);
if ( rd <= cutoffSquare)
{
//inter[] and ninter are global vars, need exclusive access
//do this more efficiently? atomic read
pthread_mutex_lock (&mutex_ninter);
inter[INDX(ninter,0)] = i;
inter[INDX(ninter,1)] = j;
//make sure atomic_inc works properly, try pthread_mutex_lock to double-check
//atomic_inc (&ninter);
ninter ++;
num_neighbours [i] ++;
num_neighbours [j] ++;
pthread_mutex_unlock (&mutex_ninter);
//this is a useless statement
if ( ninter >= MAXINTERACT) perror("MAXINTERACT limit");
}
}
//COMMIT_TRANSACTION();
//}
}
return 0;
}
void rswapcount(double *m, int *nr, int *nc, int *mfill)
{
int row[2], col[2], i, k, ij[4], n, change, cfill,
pm[4] = {1, -1, -1, 1} ;
double sm[4], ev;
/* Get the current fill 'cfill' */
n = (*nr) * (*nc);
for (i = 0, cfill=0; i < n; i++) {
if (m[i] > 0)
cfill++;
}
GetRNGstate();
/* Loop while fills differ */
while (cfill != *mfill) {
/* Select a random 2x2 matrix*/
i2rand(row, *nr - 1);
i2rand(col, *nc - 1);
ij[0] = INDX(row[0], col[0], *nr);
ij[1] = INDX(row[1], col[0], *nr);
ij[2] = INDX(row[0], col[1], *nr);
ij[3] = INDX(row[1], col[1], *nr);
for (k = 0; k < 4; k ++)
sm[k] = m[ij[k]];
/* The largest value that can be swapped */
ev = isDiag(sm);
if (ev != 0) {
/* Check the change in fills */
for (k = 0, change=0; k < 4; k++) {
if(sm[k] > 0)
change--;
if (sm[k] + pm[k]*ev > 0)
change++;
}
/* Fill does not change, but swap to bail out from
* non-swappable configurations */
if (change == 0) {
for (k = 0; k < 4; k++)
m[ij[k]] += pm[k]*ev;
}
else if ((change < 0 && *mfill < cfill) ||
(change > 0 && *mfill > cfill)) {
for (k = 0; k < 4; k++)
m[ij[k]] += pm[k]*ev;
cfill += change;
}
}
}
PutRNGstate();
}
static void swap(int *m, int *nr, int *nc, int *thin)
{
int i, a, b, c, d, row[2], col[2], nr1 = (*nr) - 1, nc1 = (*nc) -1,
n1 = (*nr) * (*nc) - 1;
size_t intcheck;
/* Get and Put RNG in calling C function */
/* GetRNGstate(); */
for (i=0, intcheck=0; i < *thin; i++) {
for(;;) {
if (intcheck % 10000 == 9999)
R_CheckUserInterrupt();
intcheck++;
/* see trialswap & quasiswap for the logic */
a = IRAND(n1);
row[0] = a % (*nr);
col[0] = a / (*nr);
do {row[1] = IRAND(nr1);} while (row[1] == row[0]);
c = INDX(row[1], col[0], *nr);
/* bail out before next row if non-swappable */
if (m[a] == m[c])
continue;
do {col[1] = IRAND(nc1);} while (col[1] == col[0]);
b = INDX(row[0], col[1], *nr);
d = INDX(row[1], col[1], *nr);
/* if we are here m[a] != m[c], and only three elements
need be tested for the unique matrix -- start from
tests that fail most likely */
if (m[d] == 1 && m[a] == 1 && m[b] == 0) {
m[a] = 0;
m[d] = 0;
m[b] = 1;
m[c] = 1;
break;
}
if (m[b] == 1 && m[c] == 1 && m[d] == 0) {
m[a] = 1;
m[d] = 1;
m[b] = 0;
m[c] = 0;
break;
}
}
}
/* PutRNGstate(); */
}
void quasiswap(int *m, int *nr, int *nc)
{
int i, n, mtot, ss, row[2], col[2], nr1, nc1, a, b, c, d;
nr1 = (*nr) - 1;
nc1 = (*nc) - 1;
/* Get matrix total 'mtot' and sum-of-squares 'ss' */
n = (*nr) * (*nc);
for (i = 0, mtot = 0, ss = 0; i < n; i++) {
mtot += m[i];
ss += m[i] * m[i];
}
/* Get R RNG */
GetRNGstate();
/* Quasiswap while there are entries > 1 */
while (ss > mtot) {
i2rand(row, nr1);
i2rand(col, nc1);
/* a,b,c,d notation for a 2x2 table */
a = INDX(row[0], col[0], *nr);
b = INDX(row[0], col[1], *nr);
c = INDX(row[1], col[0], *nr);
d = INDX(row[1], col[1], *nr);
if (m[a] > 0 && m[d] > 0 && m[a] + m[d] - m[b] - m[c] >= 2) {
ss -= 2 * (m[a] + m[d] - m[b] - m[c] - 2);
m[a]--;
m[d]--;
m[b]++;
m[c]++;
} else if (m[b] > 0 && m[c] > 0 &&
m[b] + m[c] - m[a] - m[d] >= 2) {
ss -= 2 * (m[b] + m[c] - m[a] - m[d] - 2);
m[a]++;
m[d]++;
m[b]--;
m[c]--;
}
}
/* Set R RNG */
PutRNGstate();
}
static void trialswap(int *m, int *nr, int *nc, int *thin)
{
int i, a, b, c, d, row[2], col[2];
/* Get and Set RNG in calling C function */
/* GetRNGstate(); */
for (i=0; i < *thin; i++) {
/* get corner item m[a] and its row and column index */
a = IRAND((*nr) * (*nc) - 1);
row[0] = a % (*nr);
col[0] = a / (*nr);
/* get its side-by-side neighbour in a different row */
do {row[1] = IRAND((*nr) - 1);} while (row[1] == row[0]);
c = INDX(row[1], col[0], *nr);
/* not swappable if neighbours are identical: bail out at
probability (1-f)^2 + f^2 where f is the relative column
fill */
if (m[a] == m[c])
continue;
/* get second col and its items */
do {col[1] = IRAND((*nc) - 1);} while (col[1] == col[0]);
b = INDX(row[0], col[1], *nr);
d = INDX(row[1], col[1], *nr);
/* there are 16 possible matrices, but only two can be
* swapped. Find signature of each matrix with bitwise shift
* and OR. */
switch(m[a] | m[b] << 1 | m[c] << 2 | m[d] << 3) {
case 6: /* 0110 -> 1001 */
m[a] = 1;
m[b] = 0;
m[c] = 0;
m[d] = 1;
break;
case 9: /* 1001 -> 0110 */
m[a] = 0;
m[b] = 1;
m[c] = 1;
m[d] = 0;
break;
default:
break;
}
}
/* PutRNGstate(); */
}
static void curveball(int *m, int *nr, int *nc, int *thin, int *uniq)
{
int row[2], i, j, jind, ind, nsp1, nsp2, itmp, tmp;
/* Set RNG in calling C code */
/* GetRNGstate(); */
for (i = 0; i < *thin; i++) {
/* Random sites */
I2RAND(row, (*nr)-1);
/* uniq is a vector of unique species for a random pair of
rows, It need not be zeroed between thin loops because ind
keeps track of used elements. */
for (j = 0, ind = -1, nsp1 = 0, nsp2 = 0; j < (*nc); j++) {
jind = j * (*nr);
if (m[row[0] + jind] > 0 && m[row[1] + jind] == 0) {
uniq[++ind] = j;
nsp1++;
}
if (m[row[1] + jind] > 0 && m[row[0] + jind] == 0) {
uniq[++ind] = j;
nsp2++;
}
}
/* uniq contains indices of unique species: shuffle these and
* allocate nsp1 first to row[0] and the rest to row[1] */
if (nsp1 > 0 && nsp2 > 0) { /* something to swap? */
for (j = ind; j >= nsp1; j--) {
itmp = IRAND(j);
tmp = uniq[j];
uniq[j] = uniq[itmp];
uniq[itmp] = tmp;
}
for (j = 0; j < nsp1; j++) {
m[INDX(row[0], uniq[j], *nr)] = 1;
m[INDX(row[1], uniq[j], *nr)] = 0;
}
for (j = nsp1; j <= ind; j++) {
m[INDX(row[0], uniq[j], *nr)] = 0;
m[INDX(row[1], uniq[j], *nr)] = 1;
}
}
}
/* PutRNGstate(); */
}
T Selector::sub_select(const T &x, const Selector &rhs) const {
assert(rhs.nvars() <= this->nvars());
assert(this->covers(rhs));
Selector tmp(nvars(), false);
for (uint i = 0; i < rhs.nvars(); ++i) {
tmp.add(INDX(rhs.indx(i)));
}
return tmp.select(x);
}
void abuswap(double *m, int *nr, int *nc, int *thin, int *direct)
{
int row[2], col[2], k, ij[4], changed, ev ;
double sm[4];
GetRNGstate();
changed = 0;
while (changed < *thin) {
/* Select a random 2x2 matrix*/
i2rand(row, *nr - 1);
i2rand(col, *nc - 1);
ij[0] = INDX(row[0], col[0], *nr);
ij[1] = INDX(row[1], col[0], *nr);
ij[2] = INDX(row[0], col[1], *nr);
ij[3] = INDX(row[1], col[1], *nr);
for (k = 0; k < 4; k ++)
sm[k] = m[ij[k]];
ev = isDiagSimple(sm);
/* Swap */
if (ev == 1) {
/* fixed column sums */
if (*direct == 0) {
m[ij[0]] = sm[1];
m[ij[1]] = sm[0];
m[ij[2]] = sm[3];
m[ij[3]] = sm[2];
}
/* fixed row sums */
else {
m[ij[0]] = sm[2];
m[ij[1]] = sm[3];
m[ij[2]] = sm[0];
m[ij[3]] = sm[1];
}
changed++;
}
}
PutRNGstate();
}
void PrintConnectivity()
{
int ii, i;
int min, max;
float sum, sumsq, stdev, avg;
bzero((char *)connect, sizeof(int) * NUM_PARTICLES);
for (ii=0; ii<ninter; ii++)
{
assert(inter[INDX(ii,0)] < NUM_PARTICLES);
assert(inter[INDX(ii,1)] < NUM_PARTICLES);
connect[inter[INDX(ii,0)]]++;
connect[inter[INDX(ii,1)]]++;
}
sum = 0.0;
sumsq = 0.0;
sum = connect[0];
sumsq = SQR(connect[0]);
min = connect[0];
max = connect[0];
for (i=1; i<NUM_PARTICLES; i++)
{
sum += connect[i];
sumsq += SQR(connect[i]);
if (min > connect[i])
min = connect[i];
if (max < connect[i])
max = connect[i];
}
avg = sum / NUM_PARTICLES;
stdev = sqrt((sumsq / NUM_PARTICLES) - SQR(avg));
printf("avg = %4.1lf, dev = %4.1lf, min = %d, max = %d\n",
avg, stdev, min, max);
}
void swapcount(double *m, int *nr, int *nc, int *thin)
{
int row[2], col[2], k, ij[4], changed, oldn, newn,
pm[4] = {1, -1, -1, 1} ;
double sm[4], ev;
GetRNGstate();
changed = 0;
while (changed < *thin) {
/* Select a random 2x2 matrix*/
i2rand(row, *nr - 1);
i2rand(col, *nc - 1);
ij[0] = INDX(row[0], col[0], *nr);
ij[1] = INDX(row[1], col[0], *nr);
ij[2] = INDX(row[0], col[1], *nr);
ij[3] = INDX(row[1], col[1], *nr);
for (k = 0; k < 4; k ++)
sm[k] = m[ij[k]];
/* The largest value that can be swapped */
ev = isDiag(sm);
if (ev != 0) {
/* Check that the fill doesn't change*/
for (k = 0, oldn = 0, newn = 0; k < 4; k++) {
if(sm[k] > 0)
oldn++;
if (sm[k] + pm[k]*ev > 0)
newn++;
}
/* Swap */
if (oldn == newn) {
for (k = 0; k < 4; k++)
m[ij[k]] += pm[k]*ev;
changed++;
}
}
}
PutRNGstate();
}
//we should use dynamic scheduling here
int ComputeForces( void* arg, int id )
{
double cutoffSquare;
double xx, yy, zz, rd, rrd, rrd2, rrd3, rrd4, rrd5, rrd6, rrd7, r148;
double forcex, forcey, forcez;
int i,j,ii;
double vir_tmp = 0;
double epot_tmp = 0;
cutoffSquare = cutoffRadius*cutoffRadius ;
//ninter is global var, needs exclusive access unless this method is protected
//by barriers from methods that modify ninter.
//Similarly, x and inter are global vars, and need exclusive access; they are also
//a reason why this method needs to be protected by barriers from methods that
//modify x and inter; specifically, you do not want some threads executing some
//other method that modifies x or inter or ninter, and some threads executing this
//method, where these variables are read.
//keep in mind you need to protect the shared variables accessed in here not only
//from threads executing THIS function (parallel region), but also from threads
//executing OTHER parallel regions, which may modify these shared variables;
//for(ii=0; ii<ninter; ii++) { //single-threaded version
for(ii=id; ii<ninter; ii+=NTHREADS) {
//BEGIN_TRANSACTION();
//if (id == 0) {
i = inter[INDX(ii,0)];
j = inter[INDX(ii,1)];
//we're waiting on x[] to be updated for _both_ molecules (i and j)
//by UpdateCoordinates in the previous timestep.
//spin until coords for molecules i and j are updated (by UpdateCoordinates).
//while ((counter_x[i] != 1) && (counter_x[j] != 1));
//reset the counters for x for molecules i and j;
//must be atomic writes
//should this be at the end of the transaction?...
//set_mb (&counter_x[i], 0);
//set_mb (&counter_x[j], 0);
xx = x[IND(0,i)] - x[IND(0,j)];
yy = x[IND(1,i)] - x[IND(1,j)];
zz = x[IND(2,i)] - x[IND(2,j)];
if (xx < -sideHalf) xx += side;
if (yy < -sideHalf) yy += side;
if (zz < -sideHalf) zz += side;
if (xx > sideHalf) xx -= side;
if (yy > sideHalf) yy -= side;
if (zz > sideHalf) zz -= side;
rd = (xx*xx + yy*yy + zz*zz);
if ( rd < cutoffSquare ) {
rrd = 1.0/rd;
rrd2 = rrd*rrd ;
rrd3 = rrd2*rrd ;
rrd4 = rrd2*rrd2 ;
rrd6 = rrd2*rrd4;
rrd7 = rrd6*rrd ;
r148 = rrd7 - 0.5 * rrd4 ;
forcex = xx*r148;
forcey = yy*r148;
forcez = zz*r148;
f[IND(0,i)] += forcex ;
f[IND(1,i)] += forcey ;
f[IND(2,i)] += forcez ;
f[IND(0,j)] -= forcex ;
f[IND(1,j)] -= forcey ;
f[IND(2,j)] -= forcez ;
//global vars, need exclusive access
vir_tmp += rd*r148 ;
epot_tmp += (rrd6 - rrd3);
/*
pthread_mutex_lock (&global_mutex1);
vir -= rd*r148 ;
epot += (rrd6 - rrd3);
pthread_mutex_unlock (&global_mutex1);
*/
//atomically increment the counter_f for both molecules i and j;
//these _must_ be the last thing in this "transaction";
//assumes we'll only have 2 txns, one in ComputeForces, one in UpdateCoordinates;
//if we have more, then you may need to put these under the previous mutex as well.
//atomic_inc(&counter_f[i]);
//atomic_inc(&counter_f[j]);
}
//COMMIT_TRANSACTION();
//}
}
pthread_mutex_lock (&global_mutex1);
vir -= vir_tmp;
epot += epot_tmp;
pthread_mutex_unlock (&global_mutex1);
return 0;
}
static inline long long handle_lxrt_request (unsigned int lxsrq, long *arg, RT_TASK *task)
{
#define larg ((struct arg *)arg)
union {unsigned long name; RT_TASK *rt_task; SEM *sem; MBX *mbx; RWL *rwl; SPL *spl; int i; void *p; long long ll; } arg0;
int srq;
if (likely((srq = SRQ(lxsrq)) < MAX_LXRT_FUN)) {
unsigned long type;
struct rt_fun_entry *funcm;
/*
* The next two lines of code do a lot. It makes possible to extend the use of
* USP to any other real time module service in user space, both for soft and
* hard real time. Concept contributed and copyrighted by: Giuseppe Renoldi
* ([email protected]).
*/
if (unlikely(!(funcm = rt_fun_ext[INDX(lxsrq)]))) {
rt_printk("BAD: null rt_fun_ext, no module for extension %d?\n", INDX(lxsrq));
return -ENOSYS;
}
if (!(type = funcm[srq].type)) {
return ((RTAI_SYSCALL_MODE long long (*)(unsigned long, ...))funcm[srq].fun)(RTAI_FUN_ARGS);
}
if (unlikely(NEED_TO_RW(type))) {
lxrt_fun_call_wbuf(task, funcm[srq].fun, NARG(lxsrq), arg, type);
} else {
lxrt_fun_call(task, funcm[srq].fun, NARG(lxsrq), arg);
}
return task->retval;
}
arg0.name = arg[0];
switch (srq) {
case LXRT_GET_ADR: {
arg0.p = rt_get_adr(arg0.name);
return arg0.ll;
}
case LXRT_GET_NAME: {
arg0.name = rt_get_name(arg0.p);
return arg0.ll;
}
case LXRT_TASK_INIT: {
struct arg { unsigned long name; long prio, stack_size, max_msg_size, cpus_allowed; };
arg0.rt_task = __task_init(arg0.name, larg->prio, larg->stack_size, larg->max_msg_size, larg->cpus_allowed);
return arg0.ll;
}
case LXRT_TASK_DELETE: {
arg0.i = __task_delete(arg0.rt_task ? arg0.rt_task : task);
return arg0.ll;
}
case LXRT_SEM_INIT: {
if (rt_get_adr(arg0.name)) {
return 0;
}
if ((arg0.sem = rt_malloc(sizeof(SEM)))) {
struct arg { unsigned long name; long cnt; long typ; };
lxrt_typed_sem_init(arg0.sem, larg->cnt, larg->typ);
if (rt_register(larg->name, arg0.sem, IS_SEM, current)) {
return arg0.ll;
} else {
rt_free(arg0.sem);
}
}
return 0;
}
case LXRT_SEM_DELETE: {
if (lxrt_sem_delete(arg0.sem)) {
arg0.i = -EFAULT;
return arg0.ll;
}
rt_free(arg0.sem);
arg0.i = rt_drg_on_adr(arg0.sem);
return arg0.ll;
}
case LXRT_MBX_INIT: {
if (rt_get_adr(arg0.name)) {
return 0;
}
if ((arg0.mbx = rt_malloc(sizeof(MBX)))) {
struct arg { unsigned long name; long size; int qtype; };
if (lxrt_typed_mbx_init(arg0.mbx, larg->size, larg->qtype) < 0) {
rt_free(arg0.mbx);
return 0;
}
if (rt_register(larg->name, arg0.mbx, IS_MBX, current)) {
return arg0.ll;
} else {
rt_free(arg0.mbx);
}
}
return 0;
}
case LXRT_MBX_DELETE: {
if (lxrt_mbx_delete(arg0.mbx)) {
arg0.i = -EFAULT;
return arg0.ll;
}
rt_free(arg0.mbx);
arg0.i = rt_drg_on_adr(arg0.mbx);
return arg0.ll;
}
case LXRT_RWL_INIT: {
if (rt_get_adr(arg0.name)) {
return 0;
}
if ((arg0.rwl = rt_malloc(sizeof(RWL)))) {
struct arg { unsigned long name; long type; };
lxrt_typed_rwl_init(arg0.rwl, larg->type);
if (rt_register(larg->name, arg0.rwl, IS_SEM, current)) {
return arg0.ll;
} else {
rt_free(arg0.rwl);
}
}
return 0;
}
case LXRT_RWL_DELETE: {
if (lxrt_rwl_delete(arg0.rwl)) {
arg0.i = -EFAULT;
return arg0.ll;
}
rt_free(arg0.rwl);
arg0.i = rt_drg_on_adr(arg0.rwl);
return arg0.ll;
}
case LXRT_SPL_INIT: {
if (rt_get_adr(arg0.name)) {
return 0;
}
if ((arg0.spl = rt_malloc(sizeof(SPL)))) {
struct arg { unsigned long name; };
lxrt_spl_init(arg0.spl);
if (rt_register(larg->name, arg0.spl, IS_SEM, current)) {
return arg0.ll;
} else {
rt_free(arg0.spl);
}
}
return 0;
}
case LXRT_SPL_DELETE: {
if (lxrt_spl_delete(arg0.spl)) {
arg0.i = -EFAULT;
return arg0.ll;
}
rt_free(arg0.spl);
arg0.i = rt_drg_on_adr(arg0.spl);
return arg0.ll;
}
case MAKE_HARD_RT: {
rt_make_hard_real_time(task);
return 0;
if (!task || task->is_hard) {
return 0;
}
steal_from_linux(task);
return 0;
}
case MAKE_SOFT_RT: {
rt_make_soft_real_time(task);
return 0;
if (!task || !task->is_hard) {
return 0;
}
if (task->is_hard < 0) {
task->is_hard = 0;
} else {
give_back_to_linux(task, 0);
}
return 0;
}
case PRINT_TO_SCREEN: {
struct arg { char *display; long nch; };
arg0.i = rtai_print_to_screen("%s", larg->display);
return arg0.ll;
}
case PRINTK: {
struct arg { char *display; long nch; };
arg0.i = rt_printk("%s", larg->display);
return arg0.ll;
}
case NONROOT_HRT: {
#if LINUX_VERSION_CODE < KERNEL_VERSION(2,6,24)
current->cap_effective |= ((1 << CAP_IPC_LOCK) |
(1 << CAP_SYS_RAWIO) |
(1 << CAP_SYS_NICE));
#else
set_lxrt_perm(CAP_IPC_LOCK);
set_lxrt_perm(CAP_SYS_RAWIO);
set_lxrt_perm(CAP_SYS_NICE);
#endif
return 0;
}
case RT_BUDDY: {
arg0.rt_task = task && current->rtai_tskext(TSKEXT1) == current ? task : NULL;
return arg0.ll;
}
case HRT_USE_FPU: {
struct arg { RT_TASK *task; long use_fpu; };
if(!larg->use_fpu) {
clear_lnxtsk_uses_fpu((larg->task)->lnxtsk);
} else {
init_fpu((larg->task)->lnxtsk);
}
return 0;
}
case GET_USP_FLAGS: {
arg0.name = arg0.rt_task->usp_flags;
return arg0.ll;
}
case SET_USP_FLAGS: {
struct arg { RT_TASK *task; unsigned long flags; };
arg0.rt_task->usp_flags = larg->flags;
arg0.rt_task->force_soft = (arg0.rt_task->is_hard > 0) && (larg->flags & arg0.rt_task->usp_flags_mask & FORCE_SOFT);
return 0;
}
case GET_USP_FLG_MSK: {
arg0.name = arg0.rt_task->usp_flags_mask;
return arg0.ll;
}
case SET_USP_FLG_MSK: {
task->usp_flags_mask = arg0.name;
task->force_soft = (task->is_hard > 0) && (task->usp_flags & arg0.name & FORCE_SOFT);
return 0;
}
case FORCE_TASK_SOFT: {
extern void rt_do_force_soft(RT_TASK *rt_task);
struct task_struct *ltsk;
if ((ltsk = find_task_by_pid(arg0.name))) {
if ((arg0.rt_task = ltsk->rtai_tskext(TSKEXT0))) {
if ((arg0.rt_task->force_soft = (arg0.rt_task->is_hard != 0) && FORCE_SOFT)) {
rt_do_force_soft(arg0.rt_task);
}
return arg0.ll;
}
}
return 0;
}
case IS_HARD: {
arg0.i = arg0.rt_task || (arg0.rt_task = current->rtai_tskext(TSKEXT0)) ? arg0.rt_task->is_hard : 0;
return arg0.ll;
}
case GET_EXECTIME: {
struct arg { RT_TASK *task; RTIME *exectime; };
if ((larg->task)->exectime[0] && (larg->task)->exectime[1]) {
larg->exectime[0] = (larg->task)->exectime[0];
larg->exectime[1] = (larg->task)->exectime[1];
larg->exectime[2] = rtai_rdtsc();
}
return 0;
}
case GET_TIMEORIG: {
struct arg { RTIME *time_orig; };
if (larg->time_orig) {
RTIME time_orig[2];
rt_gettimeorig(time_orig);
rt_copy_to_user(larg->time_orig, time_orig, sizeof(time_orig));
} else {
rt_gettimeorig(NULL);
}
return 0;
}
case LINUX_SERVER: {
struct arg { struct linux_syscalls_list syscalls; };
if (larg->syscalls.nr) {
if (larg->syscalls.task->linux_syscall_server) {
RT_TASK *serv;
rt_get_user(serv, &larg->syscalls.serv);
rt_task_masked_unblock(serv, ~RT_SCHED_READY);
}
larg->syscalls.task->linux_syscall_server = larg->syscalls.serv;
rtai_set_linux_task_priority(current, (larg->syscalls.task)->lnxtsk->policy, (larg->syscalls.task)->lnxtsk->rt_priority);
arg0.rt_task = __task_init((unsigned long)larg->syscalls.task, larg->syscalls.task->base_priority >= BASE_SOFT_PRIORITY ? larg->syscalls.task->base_priority - BASE_SOFT_PRIORITY : larg->syscalls.task->base_priority, 0, 0, 1 << larg->syscalls.task->runnable_on_cpus);
larg->syscalls.task->linux_syscall_server = arg0.rt_task;
arg0.rt_task->linux_syscall_server = larg->syscalls.serv;
return arg0.ll;
} else {
if (!larg->syscalls.task) {
larg->syscalls.task = RT_CURRENT;
}
if ((arg0.rt_task = larg->syscalls.task->linux_syscall_server)) {
larg->syscalls.task->linux_syscall_server = NULL;
arg0.rt_task->suspdepth = -RTE_HIGERR;
rt_task_masked_unblock(arg0.rt_task, ~RT_SCHED_READY);
}
}
return 0;
}
default: {
rt_printk("RTAI/LXRT: Unknown srq #%d\n", srq);
arg0.i = -ENOSYS;
return arg0.ll;
}
}
return 0;
}
int EvalNormal::Evaluate() {
const Side side = board_->SideToMove();
MoveArray move_array;
movegen_->GenerateMoves(&move_array);
if (move_array.size() == 0) {
const U64 attack_map = ComputeAttackMap(*board_, OppositeSide(side));
if (attack_map & board_->BitBoard(PieceOfSide(KING, side))) {
return -WIN;
} else {
return DRAW; // stalemate
}
}
int score = MATERIAL_FACTOR * PieceValDifference();
for (size_t i = 0; i < move_array.size(); ++i) {
const Move& move = move_array.get(i);
board_->MakeMove(move);
U64 attack_map = ComputeAttackMap(*board_, side);
if (attack_map & board_->BitBoard(PieceOfSide(KING, OppositeSide(side)))) {
MoveArray opp_move_array;
movegen_->GenerateMoves(&opp_move_array);
if (opp_move_array.size() == 0) {
score = WIN;
}
}
board_->UnmakeLastMove();
}
if (score == WIN) {
return score;
}
U64 self_pawns = board_->BitBoard(PieceOfSide(PAWN, side));
int self_pawns_strength = 0;
while (self_pawns) {
const int lsb_index = Lsb1(self_pawns);
self_pawns_strength += sq_strength[lsb_index];
self_pawns ^= (1ULL << lsb_index);
}
U64 opp_pawns = board_->BitBoard(PieceOfSide(PAWN, OppositeSide(side)));
int opp_pawns_strength = 0;
while (opp_pawns) {
const int lsb_index = Lsb1(opp_pawns);
opp_pawns_strength += sq_strength[lsb_index];
opp_pawns ^= (1ULL << lsb_index);
}
const int pawns_strength =
(self_pawns_strength - opp_pawns_strength) *
((board_->Ply() <= 10) ? OPENING_PAWNS_STRENGTH_FACTOR
: MIDDLE_PAWNS_STRENGTH_FACTOR);
if (board_->Ply() > 8) {
const int main_row = ((side == Side::WHITE) ? 0 : 7);
// Better to move the bishop.
if ((board_->BitBoard(PieceOfSide(BISHOP, side)) &
((1ULL << INDX(main_row, 2)) | (1ULL << INDX(main_row, 5))))) {
score -= 5;
}
// Better to move the knight.
if ((board_->BitBoard(PieceOfSide(KNIGHT, side)) &
((1ULL << INDX(main_row, 1)) | (1ULL << INDX(main_row, 6))))) {
score -= 5;
}
}
return score + pawns_strength + (move_array.size() / 15);
}
int ComputeForces( void* arg, int id ) {
double cutoffSquare;
double xx, yy, zz, rd, rrd, rrd2, rrd3, rrd4, rrd6, rrd7, r148; // rrd5 never used, removed
double forcex, forcey, forcez;
int i, k;
double vir_tmp, epot_tmp;
cutoffSquare = cutoffRadius*cutoffRadius ;
vir_tmp = 0.0;
epot_tmp = 0.0;
int finish_outer;
if (id == NTHREADS - 1) {
finish_outer = ninter;
} else {
finish_outer = (ninter / NTHREADS) * (id + 1);
}
for(int ii = (ninter / NTHREADS) * id; ii < finish_outer; ++ii) {
i = inter[INDX(ii,0)];
k = inter[INDX(ii,1)];
xx = x[IND(0,i)] - x[IND(0,k)];
yy = x[IND(1,i)] - x[IND(1,k)];
zz = x[IND(2,i)] - x[IND(2,k)];
if (xx < -sideHalf) xx += side;
if (yy < -sideHalf) yy += side;
if (zz < -sideHalf) zz += side;
if (xx > sideHalf) xx -= side;
if (yy > sideHalf) yy -= side;
if (zz > sideHalf) zz -= side;
rd = (xx*xx + yy*yy + zz*zz);
if ( rd < cutoffSquare ) {
rrd = 1.0/rd;
rrd2 = rrd*rrd ;
rrd3 = rrd2*rrd ;
rrd4 = rrd2*rrd2 ;
rrd6 = rrd2*rrd4;
rrd7 = rrd6*rrd ;
r148 = rrd7 - 0.5 * rrd4 ;
forcex = xx*r148;
forcey = yy*r148;
forcez = zz*r148;
BEGIN_TRANSACTION();
f[IND(0,i)] += forcex ;
f[IND(1,i)] += forcey ;
f[IND(2,i)] += forcez ;
f[IND(0,k)] -= forcex ;
f[IND(1,k)] -= forcey ;
f[IND(2,k)] -= forcez ;
COMMIT_TRANSACTION();
vir_tmp += rd*r148 ;
epot_tmp += (rrd6 - rrd3);
}
}
BEGIN_TRANSACTION();
vir -= vir_tmp ;
epot += epot_tmp;
COMMIT_TRANSACTION();
return 0;
}
static void quasiswap(int *m, int *nr, int *nc, int *thin)
{
int i, n, mtot, ss, row[2], col[2], n1, nr1, nc1, a, b, c, d;
size_t intcheck;
nr1 = (*nr) - 1;
nc1 = (*nc) - 1;
/* Get matrix total 'mtot' and sum-of-squares 'ss' */
n = (*nr) * (*nc);
for (i = 0, mtot = 0, ss = 0; i < n; i++) {
mtot += m[i];
ss += m[i] * m[i];
}
n1 = n - 1;
/* Get R RNG in the calling C function */
/* GetRNGstate(); */
/* Quasiswap while there are entries > 1 */
intcheck = 0; /* check interrupts */
while (ss > mtot) {
for (i = 0; i < *thin; i++) {
/* first item and its row & column indices */
a = IRAND(n1);
row[0] = a % (*nr);
col[0] = a / (*nr);
/* neighbour from the same col but different row */
do {row[1] = IRAND(nr1);} while (row[1] == row[0]);
c = INDX(row[1], col[0], *nr);
/* if neighbours are both zero, cannot be quasiswapped
(probability (1-f)^2 with relative col fill f) */
if (m[c] == 0 && m[a] == 0)
continue;
/* second col */
do {col[1] = IRAND(nc1);} while (col[1] == col[0]);
b = INDX(row[0], col[1], *nr);
d = INDX(row[1], col[1], *nr);
/* m[a] has 50% chance of being > 0, m[d] less, so have it first */
if (m[d] > 0 && m[a] > 0 && m[a] + m[d] - m[b] - m[c] >= 2) {
ss -= 2 * (m[a] + m[d] - m[b] - m[c] - 2);
m[a]--;
m[d]--;
m[b]++;
m[c]++;
} else if (m[c] > 0 && m[b] > 0 &&
m[b] + m[c] - m[a] - m[d] >= 2) {
ss -= 2 * (m[b] + m[c] - m[a] - m[d] - 2);
m[a]++;
m[d]++;
m[b]--;
m[c]--;
}
}
/* interrupt? */
if (intcheck % 10000 == 9999)
R_CheckUserInterrupt();
intcheck++;
}
/* Set R RNG in the calling function */
/* PutRNGstate(); */
}
static void greedyqswap(int *m, int nr, int nc, int thin, int *big)
{
int i, j, n, biglen, pick, row[2], col[2], nr1, nc1, a, b, c, d;
size_t intcheck;
nr1 = nr - 1;
nc1 = nc - 1;
/* big contains indices of cells > 1 */
n = nr * nc;
for (i = 0, biglen = -1; i < n; i++) {
if (m[i] > 1)
big[++biglen] = i;
}
intcheck = 0; /* check interrupts */
while (biglen > -1) {
for (i = 0; i < thin; i++) {
/* pick one item to the m[a] corner */
if (i == 0) { /* greedy! */
pick = IRAND(biglen);
a = big[pick];
} else { /* thin! */
a = IRAND(n-1);
}
row[0] = a % nr;
col[0] = a / nr;
/* get the second item in the first column */
do {row[1] = IRAND(nr1);} while (row[1] == row[0]);
c = INDX(row[1], col[0], nr);
/* unswappable if the first row is all zeros */
if (m[a] == 0 && m[c] == 0)
continue;
/* second column, third and fourth items */
do {col[1] = IRAND(nc1);} while (col[1] == col[0]);
b = INDX(row[0], col[1], nr);
d = INDX(row[1], col[1], nr);
if (m[d] > 0 && m[a] > 0 && m[a] + m[d] - m[b] - m[c] >= 2) {
m[a]--;
m[d]--;
m[b]++;
m[c]++;
/* Update big & biglen. a & d were decremented, and if
* they now are 1, they are removed from big. We know
* pick for a, but the location of d must be searched
* in big. b & c were incremented and if they now are
* 2, they must be added to big. */
if (m[a] == 1) {
if (i == 0) { /* not thinning: know the pick */
big[pick] = big[biglen--];
} else { /* thinning: must search in big */
for (j = 0; j <= biglen; j++) {
if (a == big[j]) {
big[j] = big[biglen--];
break;
}
}
}
}
if (m[d] == 1) {
for (j = 0; j <= biglen; j++) {
if (d == big[j]) {
big[j] = big[biglen--];
break;
}
}
}
if (m[b] == 2)
big[++biglen] = b;
if (m[c] == 2)
big[++biglen] = c;
} else if (m[c] > 0 && m[b] > 0 &&
m[b] + m[c] - m[a] - m[d] >= 2) {
m[a]++;
m[d]++;
m[b]--;
m[c]--;
/* update is mirror operation of the one above */
if (m[b] == 1) {
for (j = 0; j <= biglen; j++) {
if (b == big[j]) {
big[j] = big[biglen--];
break;
}
}
}
if (m[c] == 1) {
for (j = 0; j <= biglen; j++) {
if (c == big[j]) {
big[j] = big[biglen--];
break;
}
}
}
if (m[a] == 2)
big[++biglen] = a;
if (m[d] == 2)
big[++biglen] = d;
}
}
/* interrupt? */
if (intcheck % 10000 == 9999)
R_CheckUserInterrupt();
intcheck++;
}
}
