OpenFPGA/vpr7_x2p/vpr/SRC/base/stats.c

534 lines
17 KiB
C

#include <assert.h>
#include <stdio.h>
#include <math.h>
#include <string.h>
#include "util.h"
#include "vpr_types.h"
#include "globals.h"
#include "rr_graph_area.h"
#include "segment_stats.h"
#include "stats.h"
#include "net_delay.h"
#include "path_delay.h"
#include "read_xml_arch_file.h"
#include "ReadOptions.h"
/* mrFPGA: Xifan TANG*/
#include "mrfpga_globals.h"
#include "buffer_insertion.h"
#include "cal_capacitance.h"
/* end */
/* Xifan TANG: pb_pin_eq_auto_detect */
void print_net_opin_occupancy();
/* end */
/********************** Subroutines local to this module *********************/
static void load_channel_occupancies(int **chanx_occ, int **chany_occ);
static void get_length_and_bends_stats(void);
static void get_channel_occupancy_stats(void);
/************************* Subroutine definitions ****************************/
void routing_stats(boolean full_stats, enum e_route_type route_type,
int num_switch, t_segment_inf * segment_inf, int num_segment,
float R_minW_nmos, float R_minW_pmos,
enum e_directionality directionality, boolean timing_analysis_enabled,
float **net_delay, t_slack * slacks, float sram_area) {
/* Prints out various statistics about the current routing. Both a routing *
* and an rr_graph must exist when you call this routine. */
float area, used_area;
int i, j;
/* mrFPGA: Xifan TANG */
static float buffer_size;
/* end */
get_length_and_bends_stats();
get_channel_occupancy_stats();
vpr_printf(TIO_MESSAGE_INFO, "Logic area (in minimum width transistor areas, excludes I/Os and empty grid tiles)...\n");
area = 0;
for (i = 1; i <= nx; i++) {
for (j = 1; j <= ny; j++) {
if (grid[i][j].offset == 0) {
if (grid[i][j].type->area == UNDEFINED) {
area += grid_logic_tile_area * grid[i][j].type->height;
} else {
area += grid[i][j].type->area;
}
}
}
}
/* Todo: need to add pitch of routing to blocks with height > 3 */
vpr_printf(TIO_MESSAGE_INFO, "\tTotal logic block area (Warning, need to add pitch of routing to blocks with height > 3): %g\n", area);
used_area = 0;
for (i = 0; i < num_blocks; i++) {
if (block[i].type != IO_TYPE) {
if (block[i].type->area == UNDEFINED) {
used_area += grid_logic_tile_area * block[i].type->height;
} else {
used_area += block[i].type->area;
}
}
}
vpr_printf(TIO_MESSAGE_INFO, "\tTotal used logic block area: %g\n", used_area);
if (route_type == DETAILED) {
/* mrFPGA: Xifan TANG */
if (is_mrFPGA) {
if (is_wire_buffer) {
buffer_size = trans_per_buf(wire_buffer_inf.R - memristor_inf.R, R_minW_nmos, R_minW_pmos);
count_routing_memristor_buffer(print_stat_memristor_buffer("mrFPGA_buffer.echo", buffer_size),
buffer_size);
} else {
vpr_printf(TIO_MESSAGE_INFO, "Tile area: %#g\n", grid_logic_tile_area);
}
} else {
if (!is_mrFPGA && is_stack) {
for (i = 0; i < num_normal_switch; i++) {
switch_inf[i].R -= Rseg_global;
}
}
if (rram_pass_tran_value > 0.01) {
for (i = 0; i < num_normal_switch; i++) {
switch_inf[i].R = (switch_inf[i].R - rram_pass_tran_value)*2;
}
}
/* END */
count_routing_transistors(directionality, num_switch, segment_inf,
R_minW_nmos, R_minW_pmos, sram_area);
/* mrFPGA : Xifan TANG */
if (rram_pass_tran_value > 0.01) {
for (i = 0; i < num_normal_switch; i++) {
switch_inf[i].R = (switch_inf[i].R / 2. - rram_pass_tran_value);
}
}
if (!is_mrFPGA && is_stack) {
for (i = 0; i < num_normal_switch; i++) {
switch_inf[i].R += Rseg_global;
}
}
}
/* END */
get_segment_usage_stats(num_segment, segment_inf);
if (timing_analysis_enabled) {
load_net_delay_from_routing(net_delay, clb_net, num_nets);
/* mrFPGA: Xifan TANG */
cal_capacitance_from_routing();
/* END */
load_timing_graph_net_delays(net_delay);
#ifdef HACK_LUT_PIN_SWAPPING
do_timing_analysis(slacks, FALSE, TRUE, TRUE);
#else
do_timing_analysis(slacks, FALSE, FALSE, TRUE);
#endif
if (getEchoEnabled()) {
if(isEchoFileEnabled(E_ECHO_TIMING_GRAPH))
print_timing_graph(getEchoFileName(E_ECHO_TIMING_GRAPH));
if (isEchoFileEnabled(E_ECHO_NET_DELAY))
print_net_delay(net_delay, getEchoFileName(E_ECHO_NET_DELAY));
if(isEchoFileEnabled(E_ECHO_LUT_REMAPPING))
print_lut_remapping(getEchoFileName(E_ECHO_LUT_REMAPPING));
}
print_slack(slacks->slack, TRUE, getOutputFileName(E_SLACK_FILE));
print_criticality(slacks, TRUE, getOutputFileName(E_CRITICALITY_FILE));
print_critical_path(getOutputFileName(E_CRIT_PATH_FILE));
print_timing_stats();
}
}
if (full_stats == TRUE)
print_wirelen_prob_dist();
/* Xifan TANG: Statisitcs for each net which has occupied more than 1 OPIN */
print_net_opin_occupancy();
}
void get_length_and_bends_stats(void) {
/* Figures out maximum, minimum and average number of bends and net length *
* in the routing. */
int inet, bends, total_bends, max_bends;
int length, total_length, max_length;
int segments, total_segments, max_segments;
float av_bends, av_length, av_segments;
int num_global_nets, num_clb_opins_reserved;
max_bends = 0;
total_bends = 0;
max_length = 0;
total_length = 0;
max_segments = 0;
total_segments = 0;
num_global_nets = 0;
num_clb_opins_reserved = 0;
for (inet = 0; inet < num_nets; inet++) {
if (clb_net[inet].is_global == FALSE && clb_net[inet].num_sinks != 0) { /* Globals don't count. */
get_num_bends_and_length(inet, &bends, &length, &segments);
total_bends += bends;
max_bends = std::max(bends, max_bends);
total_length += length;
max_length = std::max(length, max_length);
total_segments += segments;
max_segments = std::max(segments, max_segments);
} else if (clb_net[inet].is_global) {
num_global_nets++;
} else {
num_clb_opins_reserved++;
}
}
av_bends = (float) total_bends / (float) (num_nets - num_global_nets);
vpr_printf(TIO_MESSAGE_INFO, "\n");
vpr_printf(TIO_MESSAGE_INFO, "Average number of bends per net: %#g Maximum # of bends: %d\n", av_bends, max_bends);
vpr_printf(TIO_MESSAGE_INFO, "\n");
av_length = (float) total_length / (float) (num_nets - num_global_nets);
vpr_printf(TIO_MESSAGE_INFO, "Number of routed nets (nonglobal): %d\n", num_nets - num_global_nets);
vpr_printf(TIO_MESSAGE_INFO, "Wirelength results (in units of 1 clb segments)...\n");
vpr_printf(TIO_MESSAGE_INFO, "\tTotal wirelength: %d, average net length: %#g\n", total_length, av_length);
vpr_printf(TIO_MESSAGE_INFO, "\tMaximum net length: %d\n", max_length);
vpr_printf(TIO_MESSAGE_INFO, "\n");
av_segments = (float) total_segments / (float) (num_nets - num_global_nets);
vpr_printf(TIO_MESSAGE_INFO, "Wirelength results in terms of physical segments...\n");
vpr_printf(TIO_MESSAGE_INFO, "\tTotal wiring segments used: %d, average wire segments per net: %#g\n", total_segments, av_segments);
vpr_printf(TIO_MESSAGE_INFO, "\tMaximum segments used by a net: %d\n", max_segments);
vpr_printf(TIO_MESSAGE_INFO, "\tTotal local nets with reserved CLB opins: %d\n", num_clb_opins_reserved);
}
static void get_channel_occupancy_stats(void) {
/* Determines how many tracks are used in each channel. */
int i, j, max_occ, total_x, total_y;
float av_occ;
int **chanx_occ; /* [1..nx][0..ny] */
int **chany_occ; /* [0..nx][1..ny] */
chanx_occ = (int **) alloc_matrix(1, nx, 0, ny, sizeof(int));
chany_occ = (int **) alloc_matrix(0, nx, 1, ny, sizeof(int));
load_channel_occupancies(chanx_occ, chany_occ);
vpr_printf(TIO_MESSAGE_INFO, "\n");
vpr_printf(TIO_MESSAGE_INFO, "X - Directed channels: j\tmax occ\tav_occ\t\tcapacity\n");
total_x = 0;
for (j = 0; j <= ny; j++) {
total_x += chan_width_x[j];
av_occ = 0.;
max_occ = -1;
for (i = 1; i <= nx; i++) {
max_occ = std::max(chanx_occ[i][j], max_occ);
av_occ += chanx_occ[i][j];
}
av_occ /= nx;
vpr_printf(TIO_MESSAGE_INFO, "%d\t%d\t%-#9g\t%d\n", j, max_occ, av_occ, chan_width_x[j]);
}
vpr_printf(TIO_MESSAGE_INFO, "\n");
vpr_printf(TIO_MESSAGE_INFO, "Y - Directed channels: i\tmax occ\tav_occ\t\tcapacity\n");
total_y = 0;
for (i = 0; i <= nx; i++) {
total_y += chan_width_y[i];
av_occ = 0.;
max_occ = -1;
for (j = 1; j <= ny; j++) {
max_occ = std::max(chany_occ[i][j], max_occ);
av_occ += chany_occ[i][j];
}
av_occ /= ny;
vpr_printf(TIO_MESSAGE_INFO, "%d\t%d\t%-#9g\t%d\n", i, max_occ, av_occ, chan_width_y[i]);
}
vpr_printf(TIO_MESSAGE_INFO, "\n");
vpr_printf(TIO_MESSAGE_INFO, "Total tracks in x-direction: %d, in y-direction: %d\n", total_x, total_y);
vpr_printf(TIO_MESSAGE_INFO, "\n");
free_matrix(chanx_occ, 1, nx, 0, sizeof(int));
free_matrix(chany_occ, 0, nx, 1, sizeof(int));
}
static void load_channel_occupancies(int **chanx_occ, int **chany_occ) {
/* Loads the two arrays passed in with the total occupancy at each of the *
* channel segments in the FPGA. */
int i, j, inode, inet;
struct s_trace *tptr;
t_rr_type rr_type;
/* First set the occupancy of everything to zero. */
for (i = 1; i <= nx; i++)
for (j = 0; j <= ny; j++)
chanx_occ[i][j] = 0;
for (i = 0; i <= nx; i++)
for (j = 1; j <= ny; j++)
chany_occ[i][j] = 0;
/* Now go through each net and count the tracks and pins used everywhere */
for (inet = 0; inet < num_nets; inet++) {
if (clb_net[inet].is_global && clb_net[inet].num_sinks != 0) /* Skip global and empty nets. */
continue;
tptr = trace_head[inet];
while (tptr != NULL) {
inode = tptr->index;
rr_type = rr_node[inode].type;
if (rr_type == SINK) {
tptr = tptr->next; /* Skip next segment. */
if (tptr == NULL)
break;
}
else if (rr_type == CHANX) {
j = rr_node[inode].ylow;
for (i = rr_node[inode].xlow; i <= rr_node[inode].xhigh; i++)
chanx_occ[i][j]++;
}
else if (rr_type == CHANY) {
i = rr_node[inode].xlow;
for (j = rr_node[inode].ylow; j <= rr_node[inode].yhigh; j++)
chany_occ[i][j]++;
}
tptr = tptr->next;
}
}
}
void get_num_bends_and_length(int inet, int *bends_ptr, int *len_ptr,
int *segments_ptr) {
/* Counts and returns the number of bends, wirelength, and number of routing *
* resource segments in net inet's routing. */
struct s_trace *tptr, *prevptr;
int inode;
t_rr_type curr_type, prev_type;
int bends, length, segments;
bends = 0;
length = 0;
segments = 0;
prevptr = trace_head[inet]; /* Should always be SOURCE. */
if (prevptr == NULL) {
vpr_printf(TIO_MESSAGE_ERROR, "in get_num_bends_and_length: net #%d has no traceback.\n", inet);
exit(1);
}
inode = prevptr->index;
prev_type = rr_node[inode].type;
tptr = prevptr->next;
while (tptr != NULL) {
inode = tptr->index;
curr_type = rr_node[inode].type;
if (curr_type == SINK) { /* Starting a new segment */
tptr = tptr->next; /* Link to existing path - don't add to len. */
if (tptr == NULL)
break;
curr_type = rr_node[tptr->index].type;
}
else if (curr_type == CHANX || curr_type == CHANY) {
segments++;
length += 1 + rr_node[inode].xhigh - rr_node[inode].xlow
+ rr_node[inode].yhigh - rr_node[inode].ylow;
if (curr_type != prev_type
&& (prev_type == CHANX || prev_type == CHANY))
bends++;
}
prev_type = curr_type;
tptr = tptr->next;
}
*bends_ptr = bends;
*len_ptr = length;
*segments_ptr = segments;
}
void print_wirelen_prob_dist(void) {
/* Prints out the probability distribution of the wirelength / number *
* input pins on a net -- i.e. simulates 2-point net length probability *
* distribution. */
float *prob_dist;
float norm_fac, two_point_length;
int inet, bends, length, segments, index;
float av_length;
int prob_dist_size, i, incr;
prob_dist_size = nx + ny + 10;
prob_dist = (float *) my_calloc(prob_dist_size, sizeof(float));
norm_fac = 0.;
for (inet = 0; inet < num_nets; inet++) {
if (clb_net[inet].is_global == FALSE && clb_net[inet].num_sinks != 0) {
get_num_bends_and_length(inet, &bends, &length, &segments);
/* Assign probability to two integer lengths proportionately -- i.e. *
* if two_point_length = 1.9, add 0.9 of the pins to prob_dist[2] and *
* only 0.1 to prob_dist[1]. */
two_point_length = (float) length
/ (float) (clb_net[inet].num_sinks);
index = (int) two_point_length;
if (index >= prob_dist_size) {
vpr_printf(TIO_MESSAGE_WARNING, "index (%d) to prob_dist exceeds its allocated size (%d).\n",
index, prob_dist_size);
vpr_printf(TIO_MESSAGE_INFO, "Realloc'ing to increase 2-pin wirelen prob distribution array.\n");
incr = index - prob_dist_size + 2;
prob_dist_size += incr;
prob_dist = (float *)my_realloc(prob_dist,
prob_dist_size * sizeof(float));
for (i = prob_dist_size - incr; i < prob_dist_size; i++)
prob_dist[i] = 0.0;
}
prob_dist[index] += (clb_net[inet].num_sinks)
* (1 - two_point_length + index);
index++;
if (index >= prob_dist_size) {
vpr_printf(TIO_MESSAGE_WARNING, "Warning: index (%d) to prob_dist exceeds its allocated size (%d).\n",
index, prob_dist_size);
vpr_printf(TIO_MESSAGE_INFO, "Realloc'ing to increase 2-pin wirelen prob distribution array.\n");
incr = index - prob_dist_size + 2;
prob_dist_size += incr;
prob_dist = (float *)my_realloc(prob_dist,
prob_dist_size * sizeof(float));
for (i = prob_dist_size - incr; i < prob_dist_size; i++)
prob_dist[i] = 0.0;
}
prob_dist[index] += (clb_net[inet].num_sinks)
* (1 - index + two_point_length);
norm_fac += clb_net[inet].num_sinks;
}
}
/* Normalize so total probability is 1 and print out. */
vpr_printf(TIO_MESSAGE_INFO, "\n");
vpr_printf(TIO_MESSAGE_INFO, "Probability distribution of 2-pin net lengths:\n");
vpr_printf(TIO_MESSAGE_INFO, "\n");
vpr_printf(TIO_MESSAGE_INFO, "Length p(Lenth)\n");
av_length = 0;
for (index = 0; index < prob_dist_size; index++) {
prob_dist[index] /= norm_fac;
vpr_printf(TIO_MESSAGE_INFO, "%6d %10.6f\n", index, prob_dist[index]);
av_length += prob_dist[index] * index;
}
vpr_printf(TIO_MESSAGE_INFO, "\n");
vpr_printf(TIO_MESSAGE_INFO, "Number of 2-pin nets: ;%g;\n", norm_fac);
vpr_printf(TIO_MESSAGE_INFO, "Expected value of 2-pin net length (R): ;%g;\n", av_length);
vpr_printf(TIO_MESSAGE_INFO, "Total wire length: ;%g;\n", norm_fac * av_length);
free(prob_dist);
}
void print_lambda(void) {
/* Finds the average number of input pins used per clb. Does not *
* count inputs which are hooked to global nets (i.e. the clock *
* when it is marked global). */
int bnum, ipin;
int num_inputs_used = 0;
int iclass, inet;
float lambda;
t_type_ptr type;
for (bnum = 0; bnum < num_blocks; bnum++) {
type = block[bnum].type;
assert(type != NULL);
if (type != IO_TYPE) {
for (ipin = 0; ipin < type->num_pins; ipin++) {
iclass = type->pin_class[ipin];
if (type->class_inf[iclass].type == RECEIVER) {
inet = block[bnum].nets[ipin];
if (inet != OPEN) /* Pin is connected? */
if (clb_net[inet].is_global == FALSE) /* Not a global clock */
num_inputs_used++;
}
}
}
}
lambda = (float) num_inputs_used / (float) num_blocks;
vpr_printf(TIO_MESSAGE_INFO, "Average lambda (input pins used per clb) is: %g\n", lambda);
}
int count_netlist_clocks(void) {
/* Count how many clocks are in the netlist. */
int iblock, i, clock_net;
char * name;
boolean found;
int num_clocks = 0;
char ** clock_names = NULL;
for (iblock = 0; iblock < num_logical_blocks; iblock++) {
if (logical_block[iblock].clock_net != OPEN) {
clock_net = logical_block[iblock].clock_net;
assert(clock_net != OPEN);
name = logical_block[clock_net].name;
/* Now that we've found a clock, let's see if we've counted it already */
found = FALSE;
for (i = 0; !found && i < num_clocks; i++) {
if (strcmp(clock_names[i], name) == 0) {
found = TRUE;
}
}
if (!found) {
/* If we get here, the clock is new and so we dynamically grow the array netlist_clocks by one. */
clock_names = (char **) my_realloc (clock_names, ++num_clocks * sizeof(char *));
clock_names[num_clocks - 1] = name;
}
}
}
free (clock_names);
return num_clocks;
}