1OSLAT(8)                    System Manager's Manual                   OSLAT(8)
2
3
4
5# SPDX-License-Identifier: GPL-3.0-only
6

NAME

8       oslat - OS Latency Detector
9

SYNOPSIS

11       oslat [ -shvz ] [ -b bucket-size ] [ -B bias ] [ -c cpu-list ] [ -C
12             cpu-main-thread ] [ -f rt-prio ] [ --json filename ] [ -m
13             workload-mem ] [-t runtime ] [ -T trace-threshold ] [ -w workload
14             ] [ -W bucket-width ]
15

DESCRIPTION

17       oslat is an open source userspace polling mode stress program to detect
18       OS level latency.  The program runs a busy loop with no or various
19       workloads, collecting TSC information and measuring the time frequently
20       during the process.
21

OPTIONS

23       -b, --bucket-size=N
24              Specify the number of the buckets (4-1024).
25
26       -B, --bias
27              Add a bias to all the buckets using the estimated mininum.
28
29       -c, --cpu-list=CPULIST
30              Specify CPUs to run on.  For example, '1,3,5,7-15'.
31
32       -C, --cpu-main-thread=CORE
33              Specify which CPU the main thread runs on.  Default is cpu0.
34
35       -f, --rtprio=PRIORITY
36              Using specific SCHED_FIFO priority (1-99).  Otherwise use the
37              default priority, normally it will be SCHED_OTHER.
38
39       --json=FILENAME
40              Write final results into FILENAME, JSON formatted.
41
42       -m, --workload-mem=SIZE
43              Size of the memory to use for the workload (e.g., 4K, 1M).
44              Total memory usage will be this value multiplies 2*N, because
45              there will be src/dst buffers for each thread, and N is the
46              number of processors for testing.
47
48       -D, --duration=TIME
49              Specify test duration, e.g., 60, 20m, 2H (m/M: minutes, h/H:
50              hours, d/D: days).  By default the unit is s/second.
51
52       -T, --trace-threshold=THRESHOLD
53              Stop the test when threshold triggered (in USEC).  At the
54              meantime, print a marker in ftrace and stop ftrace too.
55
56       -w, --workload=WORKLOAD
57              Specify a kind of workload, default is no workload.  Options:
58              "no", "memmove".
59
60       -s, --single-preheat
61              Use a single thread when measuring latency at preheat stage
62              NOTE: please make sure the CPU frequency on all testing cores
63              are locked before using this parmater.  If you don't know how to
64              lock the freq then please don't use this parameter.
65
66       -W, --bucket-width
67              Interval between buckets in nanoseconds
68
69              NOTE: Widths not a multiple of 1000 cause ns-precision output
70              You are responsible for considering the impact of measurement
71              overhead at the nanosecond scale.
72
73       -h, --help
74              Show the help message.
75
76       -v, --version
77              Show the version of the program.
78
79       -z, --zero-omit
80              Don't display buckets in the output histogram if all zeros.
81

AUTHOR

83       oslat was written by Peter Xu <peterx@redhat.com>.
84
85
86
87                                August 17, 2020                       OSLAT(8)