1INTEL_GPU_TOP(1) General Commands Manual INTEL_GPU_TOP(1)
2
3
4
6 intel_gpu_top - Display a top-like summary of Intel GPU usage
7
9 intel_gpu_top [OPTIONS]
10
12 intel_gpu_top is a tool to display usage information on Intel GPU's.
13
14 The tool presents data collected from performance counters (PMU), ex‐
15 posed by i915 and other platform drivers like RAPL (power) and Uncore
16 IMC (memory bandwidth).
17
19 -h Show help text.
20
21 -c Output CSV formatted data.
22
23 -J Output JSON formatted data.
24
25 -l List plain text data.
26
27 -o <file>, or -o -
28 Output to the specified file instead of standard output. '-' can
29 also be specified to explicitly select standard output.
30
31 -s <ms>
32 Refresh period in milliseconds.
33
34 -L List available GPUs on the system.
35
36 -d Select a specific GPU using one of the supported filters.
37
38 -p Default to showing physical engines instead of aggregated
39 classes.
40
42 Supported keys:
43
44 'q' Exit from the tool.
45 'h' Show interactive help.
46 '1' Toggle between aggregated by engine class and physical engine mode.
47 'n' Toggle display of numeric client busyness overlay.
48 's' Toggle between sort modes (runtime, total runtime, pid, client id).
49 'i' Toggle display of clients which used no GPU time.
50 'H' Toggle between per PID aggregation and individual clients.
51
52
54 On systems where multiple GPUs are present it is possible to select a
55 specific GPU to be monitored. A GPU can be selected by sysfs path, drm
56 device node or using various PCI sub filters.
57
58 ┌───────┬────────────────────────────────────────────────┬─────────────────────┐
59 │Filter │ Syntax │ GPU selection cri‐ │
60 │ │ │ teria │
61 ├───────┼────────────────────────────────────────────────┼─────────────────────┤
62 │sys │ sys:/sys/devices/pci0000:00/0000:00:02.0 │ Select using the │
63 │ │ │ sysfs path. │
64 └───────┴────────────────────────────────────────────────┴─────────────────────┘
65
66
67 │drm │ drm:/dev/dri/<node> │ Select using the │
68 │ │ │ /dev/dri/* device │
69 │ │ │ node. │
70 ├───────┼────────────────────────────────────────────────┼─────────────────────┤
71 │pci │ pci:[vendor=%04x/name][,device=%04x][,card=%d] │ Select using the │
72 │ │ │ PCI address. Vendor │
73 │ │ │ is hexadecinal num‐ │
74 │ │ │ ber or vendor name. │
75 └───────┴────────────────────────────────────────────────┴─────────────────────┘
76
78 JSON output will be correctly terminated when the tool cleanly exits,
79 otherwise one square bracket needs to be added before parsing.
80
82 • Not all metrics are supported on all platforms. Where a metric is un‐
83 supported it's value will be replaced by a dashed line.
84
85 • Non-root access to perf counters is controlled by the
86 perf_event_paranoid sysctl.
87
89 Report bugs on fd.o GitLab:
90 https://gitlab.freedesktop.org/drm/igt-gpu-tools/-/issues
91
93 IGT Developers <igt-dev@lists.freedesktop.org>
94
96 2009,2011,2012,2016,2018,2019,2020,2023 Intel Corporation
97
98
99
100
101igt-gpu-tools 1.27.1 2023-01-27 INTEL_GPU_TOP(1)