void *
my_worker2(void *v)
{
int i;
unsigned int start, dur;
DET_LOCK(&mutexsum);
sum++;
DET_UNLOCK(&mutexsum);
start = get_usecs();
for ( i = 0; i < iteration; i++ ) {
if ( i % 100 == 0 ) {
DET_LOCK(&mutexsum);
sum++;
DET_UNLOCK(&mutexsum);
}
fib(10); // average time = x
}
dur = get_usecs() - start;
printf("%s:dur = %d, avg = %f\n", __FUNCTION__, dur, (float)dur / iteration );
DET_LOCK(&mutexsum);
sum++;
DET_UNLOCK(&mutexsum);
return NULL;
}
void * my_worker1(void *v)
{
int i;
unsigned start, dur;
char buf[1024];
g_fd = open("file.dat", O_RDONLY);
if ( g_fd < 0 ) {
perror("open failed");
exit(1);
}
start = get_usecs();
for ( i = 0; i < iteration; i++ ) {
int cnt;
if ( i % 100 == 0) {
DET_LOCK(&mutexsum);
sum++;
DET_UNLOCK(&mutexsum);
}
cnt = read(g_fd, buf, size); // average time = x
if ( cnt <= 0 ) {
perror("finish");
break;
}
}
dur = get_usecs() - start;
printf("%s:dur = %d, avg = %f\n", __FUNCTION__, dur, (float)dur / iteration );
return NULL;
}
int main(int argc,char **argv) {
hupcpp::init(&argc, &argv);
#if 0
if(argc!=5) {
printf("(ERROR) USAGE: ./a.out <total bodies> <tileSize> <in data file> <out data file>\n");
hupcpp::finalize();
}
assert(atoi(argv[1]) == NUMBODIES);
assert(atoi(argv[2]) == TILESIZE);
//Init(argv[3]);
#endif
long start = get_usecs();
for(int time_steps=0;time_steps<MAX_STEPS;time_steps++) {
hupcpp::finish_spmd([=]() {
if(upcxx::global_myrank() ==0) {
calculate_forces();
}
});
}
long end = get_usecs();
double dur = ((double)(end-start))/1000000;
#ifdef VERIFY_SHORT
counter_t sum = 0;
for(int i=0; i<MAX_HCPP_WORKERS; i++) {
sum+=TOTAL_COUNTS[i];
}
counter_t total_sum;
upcxx::reduce<counter_t>(&sum, &total_sum, 1, 0, UPCXX_SUM, UPCXX_ULONG_LONG);
if(upcxx::global_myrank() == 0) {
const counter_t expected = NUMBODIES * MAX_STEPS;
const char* res = expected == total_sum ? "PASSED" : "FAILED";
printf("Test %s, Time = %0.3f\n",res,dur);
}
#endif
#if 0
// gather result from all other processes using upcxx_reduce
float accx_all[NUMBODIES], accy_all[NUMBODIES], accz_all[NUMBODIES];
upcxx::upcxx_reduce<float>(accx, accx_all, NUMBODIES, 0, UPCXX_SUM, UPCXX_FLOAT);
upcxx::upcxx_reduce<float>(accy, accy_all, NUMBODIES, 0, UPCXX_SUM, UPCXX_FLOAT);
upcxx::upcxx_reduce<float>(accz, accz_all, NUMBODIES, 0, UPCXX_SUM, UPCXX_FLOAT);
if(upcxx::global_myrank() ==0) {
printf("0: Computation done\n");
printf("Test Passed=%d\n",verify_result(argv[4],accx_all));
}
#endif
hupcpp::barrier();
hupcpp::finalize();
return 0;
}
/* Adjust for the fact that psched_ticks aren't always usecs
(based on kernel PSCHED_CLOCK configuration */
static int get_ticks(__u32 *ticks, const char *str)
{
unsigned t;
if(get_usecs(&t, str))
return -1;
*ticks = tc_core_usec2tick(t);
return 0;
}
void quit(int param)
{
float dur_in_sec;
float bw;
float dur = get_usecs() - g_start;
dur_in_sec = (float)dur / 1000000;
printf("g_nread(bytes read) = %lld\n", (long long)g_nread);
printf("elapsed = %.2f sec ( %.0f usec )\n", dur_in_sec, dur);
bw = (float)g_nread / dur_in_sec / 1024 / 1024;
printf("CPU%d: B/W = %.2f MB/s | ",cpuid, bw);
printf("CPU%d: average = %.2f ns\n", cpuid, (dur*1000)/(g_nread/CACHE_LINE_SIZE));
exit(0);
}
static int
hfsc_get_sc1(int *argcp, char ***argvp, struct tc_service_curve *sc)
{
char **argv = *argvp;
int argc = *argcp;
unsigned int m1 = 0, d = 0, m2 = 0;
if (matches(*argv, "m1") == 0) {
NEXT_ARG();
if (get_rate(&m1, *argv) < 0) {
explain1("m1");
return -1;
}
NEXT_ARG();
}
if (matches(*argv, "d") == 0) {
NEXT_ARG();
if (get_usecs(&d, *argv) < 0) {
explain1("d");
return -1;
}
NEXT_ARG();
}
if (matches(*argv, "m2") == 0) {
NEXT_ARG();
if (get_rate(&m2, *argv) < 0) {
explain1("m2");
return -1;
}
} else
return -1;
sc->m1 = m1;
sc->d = d;
sc->m2 = m2;
*argvp = argv;
*argcp = argc;
return 0;
}
static int tbf_parse_opt(struct qdisc_util *qu, int argc, char **argv, struct nlmsghdr *n)
{
int ok=0;
struct tc_tbf_qopt opt;
__u32 rtab[256];
__u32 ptab[256];
unsigned buffer=0, mtu=0, mpu=0, latency=0;
int Rcell_log=-1, Pcell_log = -1;
struct rtattr *tail;
memset(&opt, 0, sizeof(opt));
while (argc > 0) {
if (matches(*argv, "limit") == 0) {
NEXT_ARG();
if (opt.limit || latency) {
fprintf(stderr, "Double \"limit/latency\" spec\n");
return -1;
}
if (get_size(&opt.limit, *argv)) {
explain1("limit");
return -1;
}
ok++;
} else if (matches(*argv, "latency") == 0) {
NEXT_ARG();
if (opt.limit || latency) {
fprintf(stderr, "Double \"limit/latency\" spec\n");
return -1;
}
if (get_usecs(&latency, *argv)) {
explain1("latency");
return -1;
}
ok++;
} else if (matches(*argv, "burst") == 0 ||
strcmp(*argv, "buffer") == 0 ||
strcmp(*argv, "maxburst") == 0) {
NEXT_ARG();
if (buffer) {
fprintf(stderr, "Double \"buffer/burst\" spec\n");
return -1;
}
if (get_size_and_cell(&buffer, &Rcell_log, *argv) < 0) {
explain1("buffer");
return -1;
}
ok++;
} else if (strcmp(*argv, "mtu") == 0 ||
strcmp(*argv, "minburst") == 0) {
NEXT_ARG();
if (mtu) {
fprintf(stderr, "Double \"mtu/minburst\" spec\n");
return -1;
}
if (get_size_and_cell(&mtu, &Pcell_log, *argv) < 0) {
explain1("mtu");
return -1;
}
ok++;
} else if (strcmp(*argv, "mpu") == 0) {
NEXT_ARG();
if (mpu) {
fprintf(stderr, "Double \"mpu\" spec\n");
return -1;
}
if (get_size(&mpu, *argv)) {
explain1("mpu");
return -1;
}
ok++;
} else if (strcmp(*argv, "rate") == 0) {
NEXT_ARG();
if (opt.rate.rate) {
fprintf(stderr, "Double \"rate\" spec\n");
return -1;
}
if (get_rate(&opt.rate.rate, *argv)) {
explain1("rate");
return -1;
}
ok++;
} else if (matches(*argv, "peakrate") == 0) {
NEXT_ARG();
if (opt.peakrate.rate) {
fprintf(stderr, "Double \"peakrate\" spec\n");
return -1;
}
if (get_rate(&opt.peakrate.rate, *argv)) {
explain1("peakrate");
return -1;
}
ok++;
} else if (strcmp(*argv, "help") == 0) {
explain();
return -1;
} else {
fprintf(stderr, "What is \"%s\"?\n", *argv);
explain();
return -1;
}
argc--; argv++;
}
if (!ok)
return 0;
if (opt.rate.rate == 0 || !buffer) {
fprintf(stderr, "Both \"rate\" and \"burst\" are required.\n");
return -1;
}
if (opt.peakrate.rate) {
if (!mtu) {
fprintf(stderr, "\"mtu\" is required, if \"peakrate\" is requested.\n");
return -1;
}
}
if (opt.limit == 0 && latency == 0) {
fprintf(stderr, "Either \"limit\" or \"latency\" are required.\n");
return -1;
}
if (opt.limit == 0) {
double lim = opt.rate.rate*(double)latency/1000000 + buffer;
if (opt.peakrate.rate) {
double lim2 = opt.peakrate.rate*(double)latency/1000000 + mtu;
if (lim2 < lim)
lim = lim2;
}
opt.limit = lim;
}
if ((Rcell_log = tc_calc_rtable(opt.rate.rate, rtab, Rcell_log, mtu, mpu)) < 0) {
fprintf(stderr, "TBF: failed to calculate rate table.\n");
return -1;
}
opt.buffer = tc_calc_xmittime(opt.rate.rate, buffer);
opt.rate.cell_log = Rcell_log;
opt.rate.mpu = mpu;
if (opt.peakrate.rate) {
if ((Pcell_log = tc_calc_rtable(opt.peakrate.rate, ptab, Pcell_log, mtu, mpu)) < 0) {
fprintf(stderr, "TBF: failed to calculate peak rate table.\n");
return -1;
}
opt.mtu = tc_calc_xmittime(opt.peakrate.rate, mtu);
opt.peakrate.cell_log = Pcell_log;
opt.peakrate.mpu = mpu;
}
tail = (struct rtattr*)(((void*)n)+NLMSG_ALIGN(n->nlmsg_len));
addattr_l(n, 1024, TCA_OPTIONS, NULL, 0);
addattr_l(n, 2024, TCA_TBF_PARMS, &opt, sizeof(opt));
addattr_l(n, 3024, TCA_TBF_RTAB, rtab, 1024);
if (opt.peakrate.rate)
addattr_l(n, 4096, TCA_TBF_PTAB, ptab, 1024);
tail->rta_len = (((void*)n)+NLMSG_ALIGN(n->nlmsg_len)) - (void*)tail;
return 0;
}
int main(int argc, char *argv[])
{
int64_t sum = 0;
unsigned finish = 5;
int prio = 0;
int num_processors;
int acc_type = READ;
int opt;
cpu_set_t cmask;
int iterations = 0;
int i;
struct sched_param param;
/*
* get command line options
*/
while ((opt = getopt(argc, argv, "m:a:n:t:c:i:p:r:f:l:xh")) != -1) {
switch (opt) {
case 'm': /* set memory size */
g_mem_size = 1024 * strtol(optarg, NULL, 0);
break;
case 'a': /* set access type */
if (!strcmp(optarg, "read"))
acc_type = READ;
else if (!strcmp(optarg, "write"))
acc_type = WRITE;
else
exit(1);
break;
case 't': /* set time in secs to run */
finish = strtol(optarg, NULL, 0);
break;
case 'c': /* set CPU affinity */
cpuid = strtol(optarg, NULL, 0);
num_processors = sysconf(_SC_NPROCESSORS_CONF);
CPU_ZERO(&cmask);
CPU_SET(cpuid % num_processors, &cmask);
if (sched_setaffinity(0, num_processors, &cmask) < 0)
perror("error");
else
fprintf(stderr, "assigned to cpu %d\n", cpuid);
break;
case 'r':
prio = strtol(optarg, NULL, 0);
param.sched_priority = prio; /* 1(low)- 99(high) for SCHED_FIFO or SCHED_RR
0 for SCHED_OTHER or SCHED_BATCH */
if(sched_setscheduler(0, SCHED_FIFO, ¶m) == -1) {
perror("sched_setscheduler failed");
}
break;
case 'p': /* set priority */
prio = strtol(optarg, NULL, 0);
if (setpriority(PRIO_PROCESS, 0, prio) < 0)
perror("error");
else
fprintf(stderr, "assigned priority %d\n", prio);
break;
case 'i': /* iterations */
iterations = strtol(optarg, NULL, 0);
break;
case 'h':
usage(argc, argv);
break;
}
}
/*
* allocate contiguous region of memory
*/
g_mem_ptr = (int *)malloc(g_mem_size);
memset((char *)g_mem_ptr, 1, g_mem_size);
for (i = 0; i < g_mem_size / sizeof(int); i++)
g_mem_ptr[i] = i;
/* print experiment info before starting */
printf("memsize=%d KB, type=%s, cpuid=%d\n",
g_mem_size/1024,
((acc_type==READ) ?"read": "write"),
cpuid);
printf("stop at %d\n", finish);
/* set signals to terminate once time has been reached */
signal(SIGINT, &quit);
if (finish > 0) {
signal(SIGALRM, &quit);
alarm(finish);
}
/*
* actual memory access
*/
g_start = get_usecs();
for (i=0;; i++) {
switch (acc_type) {
case READ:
sum += bench_read();
break;
case WRITE:
sum += bench_write();
break;
}
if (iterations > 0 && i >= iterations)
break;
}
printf("total sum = %ld\n", (long)sum);
quit(0);
return 0;
}
static int
hfsc_get_sc2(int *argcp, char ***argvp, struct tc_service_curve *sc)
{
char **argv = *argvp;
int argc = *argcp;
unsigned int umax = 0, dmax = 0, rate = 0;
if (matches(*argv, "umax") == 0) {
NEXT_ARG();
if (get_size(&umax, *argv) < 0) {
explain1("umax");
return -1;
}
NEXT_ARG();
}
if (matches(*argv, "dmax") == 0) {
NEXT_ARG();
if (get_usecs(&dmax, *argv) < 0) {
explain1("dmax");
return -1;
}
NEXT_ARG();
}
if (matches(*argv, "rate") == 0) {
NEXT_ARG();
if (get_rate(&rate, *argv) < 0) {
explain1("rate");
return -1;
}
} else
return -1;
if (umax != 0 && dmax == 0) {
fprintf(stderr, "HFSC: umax given but dmax is zero.\n");
return -1;
}
if (dmax != 0 && ceil(umax * 1000000.0 / dmax) > rate) {
/*
* concave curve, slope of first segment is umax/dmax,
* intersection is at dmax
*/
sc->m1 = ceil(umax * 1000000.0 / dmax); /* in bps */
sc->d = dmax;
sc->m2 = rate;
} else {
/*
* convex curve, slope of first segment is 0, intersection
* is at dmax - umax / rate
*/
sc->m1 = 0;
sc->d = ceil(dmax - umax * 1000000.0 / rate); /* in usec */
sc->m2 = rate;
}
*argvp = argv;
*argcp = argc;
return 0;
}
int main(int argc, char* argv[]) {
if(argc < 2) {
usage(argv[0]);
}
int num_threads = 2;
if(argc > 2) {
for(int i = 0; i < argc; i++) {
if(strcmp(argv[i], "-t") == 0 && argc > i+1) {
num_threads = atoi(argv[i+1]);
}
}
}
omp_set_num_threads(num_threads);
// Open files
FILE* input_file = fopen(argv[1], "r");
if(input_file == NULL) {
usage(argv[0]);
}
// Read the matrix
int dim = 0;
fscanf(input_file, "%u\n", &dim);
int mat[dim][dim];
int element = 0;
for(int i=0; i<dim; i++) {
for(int j=0; j<dim; j++) {
if (j != (dim-1))
fscanf(input_file, "%d\t", &element);
else
fscanf(input_file, "%d\n",&element);
mat[i][j] = element;
}
}
#ifdef _PRINT_INFO
// Print the matrix
printf("Input matrix [%d]\n", dim);
for(int i=0; i<dim; i++) {
for(int j=0; j<dim; j++) {
printf("%d\t", mat[i][j]);
}
printf("\n");
}
#endif
// Algorithm based on information obtained here:
// http://stackoverflow.com/questions/2643908/getting-the-submatrix-with-maximum-sum
long alg_start = get_usecs();
// Compute vertical prefix sum
int ps[dim][dim];
for (int j=0; j<dim; j++) {
ps[0][j] = mat[0][j];
for (int i=1; i<dim; i++) {
ps[i][j] = ps[i-1][j] + mat[i][j];
}
}
#ifdef _PRINT_INFO
// Print the matrix
printf("Vertical prefix sum matrix [%d]\n", dim);
for(int i=0; i<dim; i++) {
for(int j=0; j<dim; j++) {
printf("%d\t", ps[i][j]);
}
printf("\n");
}
#endif
int max_sum = mat[0][0];
int top = 0, left = 0, bottom = 0, right = 0;
//Auxilliary variables
int sum[dim];
int pos[dim];
int local_max;
#pragma omp parallel for private(sum, pos, local_max) schedule(static, 10)
for (int i=0; i<dim; i++) {
for (int k=i; k<dim; k++) {
// Kandane over all columns with the i..k rows
clear(sum, dim);
clear(pos, dim);
local_max = 0;
// We keep track of the position of the max value over each Kandane's execution
// Notice that we do not keep track of the max value, but only its position
sum[0] = ps[k][0] - (i==0 ? 0 : ps[i-1][0]);
for (int j=1; j<dim; j++) {
if (sum[j-1] > 0) {
sum[j] = sum[j-1] + ps[k][j] - (i==0 ? 0 : ps[i-1][j]);
pos[j] = pos[j-1];
}
else {
sum[j] = ps[k][j] - (i==0 ? 0 : ps[i-1][j]);
pos[j] = j;
}
if (sum[j] > sum[local_max]) {
local_max = j;
}
} //Kandane ends here
#pragma omp critical
if (sum[local_max] > max_sum) {
// sum[local_max] is the new max value
// the corresponding submatrix goes from rows i..k.
// and from columns pos[local_max]..local_max
max_sum = sum[local_max];
top = i;
left = pos[local_max];
bottom = k;
right = local_max;
}
}
}
// Compose the output matrix
int outmat_row_dim = bottom - top + 1;
int outmat_col_dim = right - left + 1;
int outmat[outmat_row_dim][outmat_col_dim];
for(int i=top, k=0; i<=bottom; i++, k++) {
for(int j=left, l=0; j<=right ; j++, l++) {
outmat[k][l] = mat[i][j];
}
}
long alg_end = get_usecs();
// Print output matrix
printf("Sub-matrix [%dX%d] with max sum = %d, top = %d, bottom = %d, left = %d, right = %d\n", outmat_row_dim, outmat_col_dim, max_sum, top, bottom, left, right);
#ifdef _PRINT_INFO
for(int i=0; i<outmat_row_dim; i++) {
for(int j=0; j<outmat_col_dim; j++) {
printf("%d\t", outmat[i][j]);
}
printf("\n");
}
#endif
printf("%s,arg(%s),%s,%f sec, threads: %d\n", argv[0], argv[1], "CHECK_NOT_PERFORMED", ((double)(alg_end-alg_start))/1000000, num_threads);
// Release resources
fclose(input_file);
return 0;
}
void
max_sub_arr(
int **mat,
int **ps,
int **outmat,
int dim)
{
int max_sum = mat[0][0];
int top = 0, left = 0, bottom = 0, right = 0;
long alg_start, alg_end;
/* auxilliary variables */
int sum[dim];
int pos[dim];
int local_max;
alg_start = get_usecs();
/* precompute vertical prefix sum */
precomp_matrix(mat, ps, dim);
#ifdef _PRINT_INFO
/* Print the matrix */
printf("[INFO] Vertical prefix sum matrix [%d]\n", dim);
print_matrix(ps, dim, dim);
#endif
for (int i=0; i<dim; i++) {
for (int k=i; k<dim; k++) {
/* Kandane over all columns with the i..k rows */
clear(sum, dim);
clear(pos, dim);
local_max = 0;
/* we keep track of the position of the max value over each
* Kandane's execution
* -> Notice that we do not keep track of the max value,
* but only its position */
sum[0] = ps[k][0] - (i==0 ? 0 : ps[i-1][0]);
for (int j=1; j<dim; j++) {
if (sum[j-1] > 0){
sum[j] = sum[j-1] + ps[k][j] - (i==0 ? 0 : ps[i-1][j]);
pos[j] = pos[j-1];
}
else {
sum[j] = ps[k][j] - (i==0 ? 0 : ps[i-1][j]);
pos[j] = j;
}
if (sum[j] > sum[local_max]){
local_max = j;
}
} /* Kandane ends here */
if (sum[local_max] > max_sum) {
/* sum[local_max] is the new max value
* the corresponding submatrix goes from rows i..k.
* and from columns pos[local_max]..local_max */
max_sum = sum[local_max];
top = i;
left = pos[local_max];
bottom = k;
right = local_max;
}
}
}
/* FIXME - Question: Do we need to compute the output matrix? */
/* Compose the output matrix */
int outmat_row_dim = bottom - top + 1;
int outmat_col_dim = right - left + 1;
outmat = alloc_matrix(outmat_row_dim, outmat_col_dim);
for(int i=top, k=0; i<=bottom; i++, k++) {
#ifdef OPTIMIZE
int j=left;
memcpy((void*) outmat[k], (void*)(&mat[i][j]),
sizeof(**mat)*(right-left+1));
#else
for(int j=left, l=0; j<=right; j++, l++) {
outmat[k][l] = mat[i][j];
}
#endif
}
alg_end = get_usecs();
runtime = (double)(alg_end-alg_start)/1000000;
#ifdef PRINT_RESULT
/* prints the usual result */
printf("[RESULT] Sub-matrix [%dX%d] in %f sec\n",
outmat_row_dim, outmat_col_dim, runtime);
printf(" -> max sum=%d, left=%d, top=%d right=%d, bottom=%d\n",
max_sum, left, top, right, bottom);
#endif
#ifdef _PRINT_INFO
/* print output matrix */
printf("[INFO] output matrix: \n");
print_matrix(outmat, outmat_row_dim, outmat_col_dim);
#endif
free_matrix(outmat, outmat_row_dim);
}
int main(int argc, char *argv[])
{
long sum = 0;
unsigned finish = 5;
int prio = 0;
int num_processors;
int acc_type = READ;
int opt;
cpu_set_t cmask;
int use_mmap = 0;
int iterations = 0;
int i;
/*
* get command line options
*/
while ((opt = getopt(argc, argv, "m:a:n:t:c:i:p:f:l:xh")) != -1) {
switch (opt) {
case 'm': /* set memory size */
g_mem_size = 1024 * strtol(optarg, NULL, 0);
break;
case 'a': /* set access type */
if (!strcmp(optarg, "read"))
acc_type = READ;
else if (!strcmp(optarg, "write"))
acc_type = WRITE;
else if (!strcmp(optarg, "rdwr"))
acc_type = RDWR;
else
exit(1);
break;
case 'n': /* set access pattern */
/* sequential */
if( strcmp(optarg,"Seq") == 0 ) {
g_indx = 0;
g_next = (CACHE_LINE_SIZE/4);
}
/* same bank */
#if P4080_MCTRL_INTRV_NONE
else if( strcmp(optarg,"Row") == 0 ) {
g_indx = 0;
g_next = (CACHE_LINE_SIZE/4) * 1024;
}
/* diff bank */
else if( strcmp(optarg,"Bank") == 0 ) {
g_indx = 128*(CACHE_LINE_SIZE/4);
g_next = (CACHE_LINE_SIZE/4) * 1024;
}
#elif P4080_MCTRL_INTRV_CLCS
else if( strcmp(optarg,"Row") == 0 ) {
g_indx = 0;
g_next = (CACHE_LINE_SIZE/4) * 1024 * 8;// 2^19
}
/* diff bank */
else if( strcmp(optarg,"Bank") == 0 ) {
g_indx = 256*(CACHE_LINE_SIZE/4); // 2^16
g_next = (CACHE_LINE_SIZE/4) * 1024 * 8;// 2^19
}
#endif
else
exit(1);
break;
case 't': /* set time in secs to run */
finish = strtol(optarg, NULL, 0);
break;
case 'c': /* set CPU affinity */
cpuid = strtol(optarg, NULL, 0);
num_processors = sysconf(_SC_NPROCESSORS_CONF);
CPU_ZERO(&cmask);
CPU_SET(cpuid % num_processors, &cmask);
if (sched_setaffinity(0, num_processors, &cmask) < 0)
perror("error");
else
fprintf(stderr, "assigned to cpu %d\n", cpuid);
break;
case 'p': /* set priority */
prio = strtol(optarg, NULL, 0);
if (setpriority(PRIO_PROCESS, 0, prio) < 0)
perror("error");
else
fprintf(stderr, "assigned priority %d\n", prio);
break;
case 'i': /* iterations */
iterations = strtol(optarg, NULL, 0);
break;
case 'l': /* set label */
g_label = strdup(optarg);
break;
case 'f': /* set file descriptor */
g_fd = fopen(optarg, "a+");
if (g_fd == NULL)
perror("error");
break;
case 'x': /* mapping to /dev/mem. !! DANGEROUS !! */
use_mmap = 1;
break;
case 'h':
usage(argc, argv);
break;
}
}
g_indx *= cpuid;
/*
* allocate contiguous region of memory
*/
if (use_mmap) {
/* open /dev/mem for accessing memory in physical addr. */
int fd = -1;
unsigned long offset;
fprintf(stderr, "Use mmap| g_indx: 0x%x g_next: 0x%x\n", g_indx, g_next);
fd = open("/dev/mem", O_RDWR | O_SYNC);
if(fd == -1) {
fprintf(stderr, "ERROR Opening /dev/mem\n");
exit(1);
}
/* offset variable is used to allocate each cpu to a different offset
from each other */
offset = ADDR_2ND_RANK; /* + cpuid*g_mem_size;*/
fprintf(stderr, "offset: %p\n", (void *)offset);
/* use mmap to allocate each cpu to the specific address in memory */
g_mem_ptr = (int *)mmap(NULL, g_mem_size, PROT_READ|PROT_WRITE,
MAP_SHARED, fd, offset);
if(g_mem_ptr == NULL) {
fprintf(stderr, "could not allocate memarea");
exit(1);
}
fprintf(stderr, "mmap was successful: addr=%p\n", g_mem_ptr);
} else {
printf("Use standard malloc\n");
g_mem_ptr = (int *)malloc(g_mem_size);
}
for (i = 0; i < g_mem_size / sizeof(int); i++)
g_mem_ptr[i] = i;
memset((char *)g_mem_ptr, 1, g_mem_size);
fprintf(stderr, "VADDR: %p-%p\n", (char *)g_mem_ptr, (char *)g_mem_ptr + g_mem_size);
/* print experiment info before starting */
printf("memsize=%d KB, type=%s, cpuid=%d\n",
g_mem_size/1024,
((acc_type==READ) ?"read":
(acc_type==WRITE)? "write" :
(acc_type==RDWR) ? "rdwr" : "worst"),
cpuid);
printf("stop at %d\n", finish);
/* set signals to terminate once time has been reached */
signal(SIGINT, &quit);
if (finish > 0) {
signal(SIGALRM, &quit);
alarm(finish);
}
/*
* actual memory access
*/
g_start = get_usecs();
for (i=0;; i++) {
switch (acc_type) {
case READ:
sum += bench_read();
break;
case WRITE:
sum += bench_write();
break;
case RDWR:
sum += bench_rdwr();
break;
}
if (iterations > 0 && i >= iterations)
break;
}
printf("total sum = %ld\n", sum);
quit(0);
}
