int main(void) {
//initialize
int i;
unsigned row, col, delay, num;
unsigned *ivt;
e_irq_global_mask(E_FALSE);
e_irq_attach(E_WAND_INT, wand_isr);
e_irq_mask(E_WAND_INT, E_FALSE);
row = e_group_config.core_row;
col = e_group_config.core_col;
num = row * e_group_config.group_cols + col;
pause = (volatile uint32_t *) (0x7000);
result = (volatile unsigned *) (0x8f000000 + 0x4*num);
delay = 0x2000 * num + 0x2000;
if (num == 0)
*pause = 0;
*result = 0xdeadbeef;
for(i=0; i<0x10000; i++)
{
*result = i;
e_wait(E_CTIMER_0, delay);
if (num == 0)
while ((*pause));
__asm__ __volatile__("wand");
__asm__ __volatile__("idle");
// clear wand bit
state = e_reg_read(E_REG_STATUS);
state = state & (~0x8);
e_reg_write(E_REG_FSTATUS, state);
}
return EXIT_SUCCESS;
}
int main () {
e_coreid_t coreid;
unsigned int row, col, core, trow, tcol, *ec;
volatile msg_block_t *msg = (msg_block_t *) BUF_ADDRESS;
e_ctimer_set(E_CTIMER_0, 0xffffffff);
e_ctimer_start(E_CTIMER_0, E_CTIMER_CLK);
ec = (unsigned int *) 0x4000;
*ec = 0;
coreid = e_get_coreid();
e_coords_from_coreid(coreid, &row, &col);
core = row*e_group_config.group_cols + col;
srand(msg->shared_msg[core].seed);
msg->shared_msg[core].seed = msg->shared_msg[core].seed + 10;
trow = ((core + 1) % E_GROUP_CORES) / e_group_config.group_rows;
tcol = (core + 1) % e_group_config.group_cols;
ec = e_get_global_address(trow, tcol, ec);
e_write(&e_group_config, &coreid, 0, 0, ec, sizeof(e_coreid_t));
/**ec = coreid;*/
msg->shared_msg[core].msg = e_coreid_from_coords(trow, tcol);
msg->shared_msg[core].external = *(unsigned int *) 0x4000;
/* Sync */
e_wait(E_CTIMER_1, 2000);
msg->shared_msg[core].timer = 0xffffffff - e_ctimer_get(E_CTIMER_0);
msg->shared_msg[core].coreid = coreid;
e_ctimer_stop(E_CTIMER_0);
return 0;
}
int main(void) {
//initialize
int *p0, *p1, *p2;
int core_num, value, fault;
unsigned conf;
int i,j,k,co;
// co = 0;
fault = 0;
for(i=0;i<4;i++)
{
for(j=0;j<4;j++)
{
core_num = i * 4 + j;
//initialize the target mem
p2 = e_get_global_address(i,j,(void *)target);
*p2 = 0x0;
p1 = e_get_global_address(i,j,(void *)E_REG_MULTICAST);
/*
switch(core_num)
{
case 0: *p1 = _Z; break;
case 1: *p1 = _X; break;
case 2: *p1 = _Y; break;
case 3: *p1 = _Z; break;
case 4: *p1 = _W; break;
case 5: *p1 = _Y; break;
case 6: *p1 = _Z; break;
case 7: *p1 = _W; break;
case 8: *p1 = _X; break;
case 9: *p1 = _Z; break;
case 10: *p1 = _W; break;
case 11: *p1 = _X; break;
case 12: *p1 = _Y; break;
case 13: *p1 = _W; break;
case 14: *p1 = _X; break;
case 15: *p1 = _Y; break;
default: break;
}
*/
switch(core_num)
{
case 0: *p1 = _X; break;
case 1: *p1 = _X; break;
case 2: *p1 = _Y; break;
case 3: *p1 = _Y; break;
case 4: *p1 = _X; break;
case 5: *p1 = _X; break;
case 6: *p1 = _Y; break;
case 7: *p1 = _Y; break;
case 8: *p1 = _Z; break;
case 9: *p1 = _Z; break;
case 10: *p1 = _W; break;
case 11: *p1 = _W; break;
case 12: *p1 = _Z; break;
case 13: *p1 = _Z; break;
case 14: *p1 = _W; break;
case 15: *p1 = _W; break;
default: break;
}
}
}
e_wait(E_CTIMER_0, 0x30000);
//set the transaction mod to multicast
conf = e_reg_read(E_REG_CONFIG);
conf = conf & 0xffff0fff;
conf = conf | 0x00003000;
e_reg_write(E_REG_CONFIG, conf);
//do the multicast
p0 = (int *) _Xmail;
*p0 = 0x111;
//wait for some time
// for(k=0;k<50;k++) {co++;}
p0 = (int *) _Ymail;
*p0 = 0x222;
// for(k=0;k<50;k++) {co++;}
p0 = (int *) _Zmail;
*p0 = 0x333;
// for(k=0;k<50;k++) {co++;}
p0 = (int *) _Wmail;
*p0 = 0x444;
//set the transaction mod to regular transaction
conf = e_reg_read(E_REG_CONFIG);
conf = conf & 0xffff0fff;
e_reg_write(E_REG_CONFIG, conf);
e_wait(E_CTIMER_0, 0x30000);
//check the results
for(i=0;i<4;i++)
{
for(j=0;j<4;j++)
{
core_num = i * 4 + j;
//get the target mem address
p2 = e_get_global_address(i,j,(void *)target);
/*
switch(core_num)
{
case 0: value = _Zvalue; break;
case 1: value = _Xvalue; break;
case 2: value = _Yvalue; break;
case 3: value = _Zvalue; break;
case 4: value = _Wvalue; break;
case 5: value = _Yvalue; break;
case 6: value = _Zvalue; break;
case 7: value = _Wvalue; break;
case 8: value = _Xvalue; break;
case 9: value = _Zvalue; break;
case 10: value = _Wvalue; break;
case 11: value = _Xvalue; break;
case 12: value = _Yvalue; break;
case 13: value = _Wvalue; break;
case 14: value = _Xvalue; break;
case 15: value = _Yvalue; break;
default: break;
}
*/
switch(core_num)
{
case 0: value = _Xvalue; break;
case 1: value = _Xvalue; break;
case 2: value = _Yvalue; break;
case 3: value = _Yvalue; break;
case 4: value = _Xvalue; break;
case 5: value = _Xvalue; break;
case 6: value = _Yvalue; break;
case 7: value = _Yvalue; break;
case 8: value = _Zvalue; break;
case 9: value = _Zvalue; break;
case 10: value = _Wvalue; break;
case 11: value = _Wvalue; break;
case 12: value = _Zvalue; break;
case 13: value = _Zvalue; break;
case 14: value = _Wvalue; break;
case 15: value = _Wvalue; break;
default: break;
}
if(*p2 != value)
fault++;
}
}
//check the final result
*mailbox = fault;
return EXIT_SUCCESS;
}
int main()
{
msg_t *m = (msg_t *) 0x4000;
uint64_t d64;
uint32_t d32;
uint16_t d16;
uint8_t d8;
unsigned int dt;
void *da, *dua, *dc;
for (unsigned int row = 0; row < E_ROWS; row++) {
for (unsigned int col = 0; col < E_COLS; col++) {
if (row == 0 && col == 0) continue;
unsigned int tcore = row * E_COLS + col;
da = e_get_global_address(row, col, (void *) MEM_ALINEADA);
dua = e_get_global_address(row, col, (void *) MEM_NO_ALINEADA);
dc = e_get_global_address(row, col, (void *) MEM_A_CABALLO);
e_dma_wait(E_DMA_1);
e_ctimer_set(E_CTIMER_0, E_CTIMER_MAX);
e_ctimer_start(E_CTIMER_0, E_CTIMER_CLK);
for (int i = 0; i < VECES; i++)
e_dma_copy(&d64, da, sizeof(d64));
e_dma_wait(E_DMA_1);
e_wait(E_CTIMER_1, DUMMY_WAIT);
e_ctimer_stop(E_CTIMER_0);
dt = E_CTIMER_MAX - e_ctimer_get(E_CTIMER_0);
m->ticks[tcore].t64 = (dt - DUMMY_WAIT) / (double) VECES;
e_dma_wait(E_DMA_1);
e_ctimer_set(E_CTIMER_0, E_CTIMER_MAX);
e_ctimer_start(E_CTIMER_0, E_CTIMER_CLK);
for (int i = 0; i < VECES; i++)
e_dma_copy(&d32, da, sizeof(d32));
e_dma_wait(E_DMA_1);
e_wait(E_CTIMER_1, DUMMY_WAIT);
e_ctimer_stop(E_CTIMER_0);
dt = E_CTIMER_MAX - e_ctimer_get(E_CTIMER_0);
m->ticks[tcore].t32 = (dt - DUMMY_WAIT) / (double) VECES;
e_dma_wait(E_DMA_1);
e_ctimer_set(E_CTIMER_0, E_CTIMER_MAX);
e_ctimer_start(E_CTIMER_0, E_CTIMER_CLK);
for (int i = 0; i < VECES; i++)
e_dma_copy(&d16, da, sizeof(d16));
e_dma_wait(E_DMA_1);
e_wait(E_CTIMER_1, DUMMY_WAIT);
e_ctimer_stop(E_CTIMER_0);
dt = E_CTIMER_MAX - e_ctimer_get(E_CTIMER_0);
m->ticks[tcore].t16 = (dt - DUMMY_WAIT) / (double) VECES;
e_dma_wait(E_DMA_1);
e_ctimer_set(E_CTIMER_0, E_CTIMER_MAX);
e_ctimer_start(E_CTIMER_0, E_CTIMER_CLK);
for (int i = 0; i < VECES; i++)
e_dma_copy(&d8, da, sizeof(d8));
e_dma_wait(E_DMA_1);
e_wait(E_CTIMER_1, DUMMY_WAIT);
e_ctimer_stop(E_CTIMER_0);
dt = E_CTIMER_MAX - e_ctimer_get(E_CTIMER_0);
m->ticks[tcore].t8 = (dt - DUMMY_WAIT) / (double) VECES;
e_dma_wait(E_DMA_1);
e_ctimer_set(E_CTIMER_0, E_CTIMER_MAX);
e_ctimer_start(E_CTIMER_0, E_CTIMER_CLK);
for (int i = 0; i < VECES; i++)
e_dma_copy(&d64, dua, sizeof(d64));
e_dma_wait(E_DMA_1);
e_wait(E_CTIMER_1, DUMMY_WAIT);
e_ctimer_stop(E_CTIMER_0);
dt = E_CTIMER_MAX - e_ctimer_get(E_CTIMER_0);
m->ticks[tcore].ua64 = (dt - DUMMY_WAIT) / (double) VECES;
e_dma_wait(E_DMA_1);
e_ctimer_set(E_CTIMER_0, E_CTIMER_MAX);
e_ctimer_start(E_CTIMER_0, E_CTIMER_CLK);
for (int i = 0; i < VECES; i++)
e_dma_copy(&d32, dua, sizeof(d32));
e_dma_wait(E_DMA_1);
e_wait(E_CTIMER_1, DUMMY_WAIT);
e_ctimer_stop(E_CTIMER_0);
dt = E_CTIMER_MAX - e_ctimer_get(E_CTIMER_0);
m->ticks[tcore].ua32 = (dt - DUMMY_WAIT) / (double) VECES;
e_dma_wait(E_DMA_1);
e_ctimer_set(E_CTIMER_0, E_CTIMER_MAX);
e_ctimer_start(E_CTIMER_0, E_CTIMER_CLK);
for (int i = 0; i < VECES; i++)
e_dma_copy(&d16, dua, sizeof(d16));
e_dma_wait(E_DMA_1);
e_wait(E_CTIMER_1, DUMMY_WAIT);
e_ctimer_stop(E_CTIMER_0);
dt = E_CTIMER_MAX - e_ctimer_get(E_CTIMER_0);
m->ticks[tcore].ua16 = (dt - DUMMY_WAIT) / (double) VECES;
e_dma_wait(E_DMA_1);
e_ctimer_set(E_CTIMER_0, E_CTIMER_MAX);
e_ctimer_start(E_CTIMER_0, E_CTIMER_CLK);
for (int i = 0; i < VECES; i++)
e_dma_copy(&d8, dua, sizeof(d8));
e_dma_wait(E_DMA_1);
e_wait(E_CTIMER_1, DUMMY_WAIT);
e_ctimer_stop(E_CTIMER_0);
dt = E_CTIMER_MAX - e_ctimer_get(E_CTIMER_0);
m->ticks[tcore].ua8 = (dt - DUMMY_WAIT) / (double) VECES;
e_dma_wait(E_DMA_1);
e_ctimer_set(E_CTIMER_0, E_CTIMER_MAX);
e_ctimer_start(E_CTIMER_0, E_CTIMER_CLK);
for (int i = 0; i < VECES; i++)
e_dma_copy(&d64, dua, sizeof(d64));
e_dma_wait(E_DMA_1);
e_wait(E_CTIMER_1, DUMMY_WAIT);
e_ctimer_stop(E_CTIMER_0);
dt = E_CTIMER_MAX - e_ctimer_get(E_CTIMER_0);
m->ticks[tcore].c64 = (dt - DUMMY_WAIT) / (double) VECES;
e_dma_wait(E_DMA_1);
e_ctimer_set(E_CTIMER_0, E_CTIMER_MAX);
e_ctimer_start(E_CTIMER_0, E_CTIMER_CLK);
for (int i = 0; i < VECES; i++)
e_dma_copy(&d32, dc, sizeof(d32));
e_dma_wait(E_DMA_1);
e_wait(E_CTIMER_1, DUMMY_WAIT);
e_ctimer_stop(E_CTIMER_0);
dt = E_CTIMER_MAX - e_ctimer_get(E_CTIMER_0);
m->ticks[tcore].c32 = (dt - DUMMY_WAIT) / (double) VECES;
e_dma_wait(E_DMA_1);
e_ctimer_set(E_CTIMER_0, E_CTIMER_MAX);
e_ctimer_start(E_CTIMER_0, E_CTIMER_CLK);
for (int i = 0; i < VECES; i++)
e_dma_copy(&d16, dc, sizeof(d16));
e_dma_wait(E_DMA_1);
e_wait(E_CTIMER_1, DUMMY_WAIT);
e_ctimer_stop(E_CTIMER_0);
dt = E_CTIMER_MAX - e_ctimer_get(E_CTIMER_0);
m->ticks[tcore].c16 = (dt - DUMMY_WAIT) / (double) VECES;
e_dma_wait(E_DMA_1);
e_ctimer_set(E_CTIMER_0, E_CTIMER_MAX);
e_ctimer_start(E_CTIMER_0, E_CTIMER_CLK);
for (int i = 0; i < VECES; i++)
e_dma_copy(&d8, dc, sizeof(d8));
e_dma_wait(E_DMA_1);
e_wait(E_CTIMER_1, DUMMY_WAIT);
e_ctimer_stop(E_CTIMER_0);
dt = E_CTIMER_MAX - e_ctimer_get(E_CTIMER_0);
m->ticks[tcore].c8 = (dt - DUMMY_WAIT) / (double) VECES;
}
}
m->finalizado = E_TRUE;
return 0;
}
