Skip to content

UCS/DEBUG: Add table builder API #11477

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Open
guy-ealey-morag wants to merge 9 commits into
base: master
Choose a base branch
from
+1,029 −185
Open

UCS/DEBUG: Add table builder API #11477

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
9 commits
Select commit Hold shift + click to select a range
0120324
UCS/DEBUG: Add table builder API
guy-ealey-morag May 21, 2026
3746aef
UCS/DEBUG: PR Fixes
guy-ealey-morag May 22, 2026
e98c9d4
UCS/DEBUG: Add include config.h
guy-ealey-morag May 22, 2026
fe5e117
UCP/PROTO: Increase buffer size to UCS_ALLOCA_MAX_SIZE
guy-ealey-morag May 26, 2026
90fa3b4
TOOLS/INFO: Center align main diagonal
guy-ealey-morag May 26, 2026
df1761b
UCP/PROTO: Remove trailing newline
guy-ealey-morag May 28, 2026
395fa2b
UCS/DEBUG: Make the row handle unsigned
guy-ealey-morag May 28, 2026
8340767
UCS/DEBUG: allocate table widths arr on the stack
guy-ealey-morag May 28, 2026
66ee664
UCS/DEBUG: Some minor cleanup
guy-ealey-morag May 29, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
194 changes: 104 additions & 90 deletions src/tools/info/sys_info.c
View file
Open in desktop
Original file line number Diff line number Diff line change
@@ -1,5 +1,5 @@
/**
* Copyright (c) NVIDIA CORPORATION & AFFILIATES, 2001-2015. ALL RIGHTS RESERVED.
* Copyright (c) NVIDIA CORPORATION & AFFILIATES, 2001-2026. ALL RIGHTS RESERVED.
* Copyright (C) Shanghai Zhaoxin Semiconductor Co., Ltd. 2020. ALL RIGHTS RESERVED.
* Copyright (C) Tactical Computing Labs, LLC. 2022. ALL RIGHTS RESERVED.
* Copyright (C) Advanced Micro Devices, Inc. 2024. ALL RIGHTS RESERVED.
Expand All @@ -13,7 +13,7 @@

#include "ucx_info.h"

#include <ucs/sys/string.h>
#include <ucs/debug/table.h>
#include <ucs/sys/sys.h>
#include <ucs/sys/math.h>
#include <ucs/time/time.h>
Expand Down Expand Up @@ -61,131 +61,145 @@ static double measure_memcpy_bandwidth(size_t size)
return result;
}

static void print_repeat_char(int ch, int count)
/* Add an empty row used as vertical padding around content rows */
static void print_sys_topo_add_padding(ucs_table_t *table)
Comment thread
guy-ealey-morag marked this conversation as resolved.
{
int i;
ucs_table_row_h row = ucs_table_add_row(table);
unsigned i;

for (i = 0; i < count; ++i) {
putchar(ch);
for (i = 0; i < table->config.n_cols; ++i) {
ucs_table_row_add_cell_empty(table, row, 1);
}
}

static void print_row_separator(int column_width, int first_column_width,
int num_columns, int fill_char,
int separator_char)
/* Add a header row of the shape "<label> | dev0 | dev1 | ...", surrounded
* by empty padding rows and followed by a separator. */
static void print_sys_topo_add_devices_header(ucs_table_t *table,
const char *first_col_label,
unsigned num_devices)
{
int i;
ucs_table_row_h row;
ucs_sys_device_t sys_dev;

printf("# %c", separator_char);
print_repeat_char(fill_char, first_column_width);
for (i = 0; i < num_columns; ++i) {
putchar(separator_char);
print_repeat_char(fill_char, column_width);
}
printf("%c\n", separator_char);
}
print_sys_topo_add_padding(table);

static void print_table_header(const char *title, const char *distance_unit,
int column_width, int first_column_width,
int num_columns)
{
int column;

printf("#\n");
printf("# %s\n", title);
printf("#\n");
print_row_separator(column_width, first_column_width, num_columns, '-',
'+');
print_row_separator(column_width, first_column_width, num_columns, ' ',
'|');
printf("# |%*s ", first_column_width - 1, distance_unit);
for (column = 0; column < num_columns; ++column) {
printf("|%*s ", column_width - 1,
ucs_topo_sys_device_get_name((ucs_sys_device_t)column));
row = ucs_table_add_row(table);
ucs_table_row_add_cell_fmt(table, row, 1, UCS_TABLE_ALIGN_RIGHT, "%s",
first_col_label);
for (sys_dev = 0; sys_dev < num_devices; ++sys_dev) {
ucs_table_row_add_cell_fmt(table, row, 1, UCS_TABLE_ALIGN_RIGHT, "%s",
ucs_topo_sys_device_get_name(sys_dev));
}

printf("|\n");
print_row_separator(column_width, first_column_width, num_columns, ' ',
'|');
print_row_separator(column_width, first_column_width, num_columns, '-',
'+');
print_sys_topo_add_padding(table);
ucs_table_add_separator(table);
}

static void print_sys_topo()
static void print_sys_topo_distances(unsigned num_devices)
{
const unsigned num_devices = ucs_topo_num_devices();
static const int distance_width = 10;
static const char *system_topo_title = "System topology";
static const char *distance_unit = "MB/s";
static const char *memory_latency_title = "NUMA memory latency";
static const char *numa_distance_unit = "nsec";
ucs_table_config_t cfg = {
.n_cols = 1 + num_devices,
.row_prefix = "# ",
.equal_widths = 1,
};
ucs_sys_device_t sys_dev1, sys_dev2;
ucs_sys_dev_distance_t distance;
char distance_str[20];
ucs_status_t status;
int name_width;
ucs_table_row_h row;
ucs_table_t table;

/* Get maximal width of device name */
name_width = 2 + strlen(distance_unit);
for (sys_dev1 = 0; sys_dev1 < num_devices; ++sys_dev1) {
name_width = ucs_max(
name_width, 2 + strlen(ucs_topo_sys_device_get_name(sys_dev1)));
}
printf("#\n# System topology\n#\n");

ucs_table_init(&table, &cfg);

print_table_header(system_topo_title, distance_unit, distance_width,
name_width, num_devices);
print_sys_topo_add_devices_header(&table, "MB/s", num_devices);

/* Print table content */
for (sys_dev1 = 0; sys_dev1 < num_devices; ++sys_dev1) {
print_row_separator(distance_width, name_width, num_devices, ' ', '|');
if (sys_dev1 > 0) {
ucs_table_add_separator(&table);
}

print_sys_topo_add_padding(&table);

row = ucs_table_add_row(&table);
ucs_table_row_add_cell_fmt(&table, row, 1, UCS_TABLE_ALIGN_RIGHT, "%s",
ucs_topo_sys_device_get_name(sys_dev1));

printf("# |%*s ", name_width - 1,
ucs_topo_sys_device_get_name(sys_dev1));
for (sys_dev2 = 0; sys_dev2 < num_devices; ++sys_dev2) {
if (sys_dev1 == sys_dev2) {
/* Do not print distance of device to itself */
strncpy(distance_str, "-", sizeof(distance_str));
ucs_table_row_add_cell_fmt(&table, row, 1,
UCS_TABLE_ALIGN_CENTER, "%s", "-");
continue;
}

status = ucs_topo_get_distance(sys_dev1, sys_dev2, &distance);
if (status != UCS_OK) {
ucs_table_row_add_cell_fmt(&table, row, 1,
UCS_TABLE_ALIGN_RIGHT, "<%s>",
ucs_status_string(status));
} else if (distance.bandwidth > UCS_PBYTE) {
ucs_table_row_add_cell_fmt(&table, row, 1,
UCS_TABLE_ALIGN_RIGHT, "%s", "inf");
} else {
status = ucs_topo_get_distance(sys_dev1, sys_dev2, &distance);
if (status != UCS_OK) {
ucs_snprintf_safe(distance_str, sizeof(distance_str),
"<%s>", ucs_status_string(status));
} else if (distance.bandwidth > UCS_PBYTE) {
ucs_snprintf_safe(distance_str, sizeof(distance_str),
"inf");
} else {
ucs_snprintf_safe(distance_str, sizeof(distance_str),
"%.1f", distance.bandwidth / UCS_MBYTE);
}
ucs_table_row_add_cell_fmt(&table, row, 1,
UCS_TABLE_ALIGN_RIGHT, "%.1f",
distance.bandwidth / UCS_MBYTE);
}
printf("|%*s ", distance_width - 1, distance_str);
}
printf("|\n");

print_row_separator(distance_width, name_width, num_devices, ' ', '|');
print_row_separator(distance_width, name_width, num_devices, '-', '+');
print_sys_topo_add_padding(&table);
}

print_table_header(memory_latency_title, "device", distance_width,
name_width, num_devices);
print_row_separator(distance_width, name_width, num_devices, ' ', '|');
ucs_table_print(&table);
ucs_table_cleanup(&table);
}

printf("# |%*s ", name_width - 1, numa_distance_unit);
printf("|");
for (sys_dev1 = 0; sys_dev1 < num_devices; ++sys_dev1) {
ucs_topo_get_memory_distance(sys_dev1, &distance);
ucs_snprintf_safe(distance_str, sizeof(distance_str), "%.1f",
distance.latency * UCS_NSEC_PER_SEC);
printf("%*s |", distance_width - 1, distance_str);
static void print_sys_topo_memory_latency(unsigned num_devices)
{
ucs_table_config_t cfg = {
.n_cols = 1 + num_devices,
.row_prefix = "# ",
.equal_widths = 1,
};
ucs_sys_dev_distance_t distance;
ucs_sys_device_t sys_dev;
ucs_table_row_h row;
ucs_table_t table;

printf("#\n# NUMA memory latency\n#\n");

ucs_table_init(&table, &cfg);

print_sys_topo_add_devices_header(&table, "device", num_devices);

print_sys_topo_add_padding(&table);

row = ucs_table_add_row(&table);
ucs_table_row_add_cell_fmt(&table, row, 1, UCS_TABLE_ALIGN_RIGHT, "%s",
"nsec");
for (sys_dev = 0; sys_dev < num_devices; ++sys_dev) {
ucs_topo_get_memory_distance(sys_dev, &distance);
ucs_table_row_add_cell_fmt(&table, row, 1, UCS_TABLE_ALIGN_RIGHT,
"%.1f", distance.latency * UCS_NSEC_PER_SEC);
}

printf("\n");
print_row_separator(distance_width, name_width, num_devices, ' ', '|');
print_row_separator(distance_width, name_width, num_devices, '-', '+');
print_sys_topo_add_padding(&table);

ucs_table_print(&table);
ucs_table_cleanup(&table);

printf("# Memory latency is calculated according to the CPU affinity\n");
}
Comment thread
guy-ealey-morag marked this conversation as resolved.

static void print_sys_topo()
{
const unsigned num_devices = ucs_topo_num_devices();

print_sys_topo_distances(num_devices);
print_sys_topo_memory_latency(num_devices);
}

static double measure_timer_accuracy()
{
double elapsed, elapsed_accurate;
Expand Down
2 changes: 1 addition & 1 deletion src/ucp/proto/proto_common.c
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -766,7 +766,7 @@ void ucp_proto_request_select_error(ucp_request_t *req,
size_t msg_length)
{
UCS_STRING_BUFFER_ONSTACK(sel_param_strb, UCP_PROTO_SELECT_PARAM_STR_MAX);
UCS_STRING_BUFFER_ONSTACK(proto_select_strb, UCP_PROTO_CONFIG_STR_MAX);
UCS_STRING_BUFFER_ONSTACK(proto_select_strb, UCS_ALLOCA_MAX_SIZE);
Comment thread
guy-ealey-morag marked this conversation as resolved.
ucp_ep_h ep = req->send.ep;

ucp_proto_select_param_str(sel_param, ucp_operation_names, &sel_param_strb);
Expand Down
Loading
Loading