jetsontools package

Subpackages

Module contents

Package for bounding boxes and their operations.

Submodules

info

Tools for getting information about the Jetson device.

Classes

Tegrastats

Runs tegrastats in a separate process and stores output in a file.

Functions

get_info

Get information about the Jetson device.

get_data

Parse the output of parse_tegrastats further to get specfic data.

get_powerdraw

Parse the output of parse_tegrastats to get all energy information.

filter_data

Filter the Tegrastats output by selections of timestamps.

set_log_level

Set the log level for the jetsontools package.

parse_tegrastats

Parse a file written by Tegrastats/tegrastats

class jetsontools.Tegrastats(output: Path | str, interval: int = 1000, *, readall: bool | None = None)

Bases: object

Runs tegrastats in a seperate process and stores output in a file.

reset() → None

Reset the Tegrastats process and data file.

start() → None

Start running Tegrastats.

stop() → None

Stop running Tegrastats.

jetsontools.filter_data(data: list[dict[str, str]], timestamps: list[tuple[float, float]]) → tuple[list[dict[str, str]], list[tuple[tuple[float, float], list[dict[str, str]]]]]

Filter data based on sets of timestamps.

The data given must have the Python timestamps included, i.e. have been generated by Tegrastats.

Parameters:

  • data (list[dict[str, str]]) – The data to filter, generated by Tegrastats

  • timestamps (list[tuple[float, float]]) – The timestamps to filter the data by. Each tuple should represent the (start, stop) where start/stop are time.time() calls.

Returns:

The unique entries within all the timestamps and the entries within each timestamp set.

Return type:

tuple[list[dict[str, str]], list[list[dict[str, str]]]]

Raises:

KeyError – If the timestamp data cannot be found

jetsontools.get_data(data: list[dict[str, str]], names: list[str], parsefunc: Callable[[str], float | int]) → dict[str, Metric]

Parse the output from parse_tegrastats to specific output.

Parameters:

  • data (list[dict[str, str]]) – The output of parse_tegrastats

  • names (list[str]) – The entries of data to get

  • parsefunc (Callable[[str], T]) – The parsing function to apply to each entries values

Returns:

The parsed values, each value is the parsed values

Return type:

dict[str, list[T]]

jetsontools.get_info(*, verbose: bool | None = None) → JetsonInfo

Get information about the Jetson device.

Parameters:

verbose (bool, optional) – If True, print additional information, by default None

Returns:

The information about the Jetson device.

Return type:

dict[str, str]

Raises:

RuntimeError – If the subprocess stdout streams cannot be opened.

jetsontools.get_powerdraw(data: list[dict[str, str]]) → dict[str, Metric]

Parse the output from parse_tegrastats to give power draw values.

Computes two new metrics: VDD_TOTAL and VIN_TOTAL. Useful for profiling overall powerdraw of a Jetson system.

Parameters:

data (list[dict[str, str]]) – The output of parse_tegrastats

Returns:

The parsed values, each value is the parsed values

Return type:

dict[str, list[float]]

jetsontools.parse_tegrastats(file: Path | str) → list[dict[str, str]]

Parse a file written by Tegrastats or tegrastats.

This a general purpose parser which does not seperate entries or handle types. All values are still strings in original format.

The differences from raw entry/values are: 1. values such as (cached 0MB) become entires: CACHED: 0MB. However these pieces of data are part of a two-piece entry for the memory subsystems, so representation in fixed format (dict[str, str]) required a change. 2. temperature entries such as cpu@49.01C become: CPU_TEMP: 49.01C This change upholds that all data generated from tegrastats has uppercase keys and that cpu frequencies will not clash with cpu temperature.

Whether or not the file was written by the class Tegrastats or directly by tegrastats will be determined automatically. Output generated by Tegrastats (Python class) will have a timestamp field. The Python timestamp is present as ‘timestamp’.

Example entry (from Orin AGX 64GB):

{ ‘timestamp’: ‘1729630462.0975535’, # optional entry ‘DATE’: ‘10-22-2024’, ‘TIME’: ‘15:54:22’, ‘RAM’: ‘23336/62841MB’, ‘LFB’: ‘199x4MB’, ‘SWAP’: ‘0/31421MB’, ‘CACHED’: ‘0MB’, ‘CPU’: ‘[0%@2201,50%@2201,0%@2201,0%@2201,0%@2201,0%@2201,0%@2201,0%@2201,0%@2201,0%@2201,0%@2201,0%@2201]’, ‘GR3D_FREQ’: ‘0%’, ‘CPU_TEMP’: ‘48.812C’, ‘TBOARD_TEMP’: ‘38.5C’, ‘SOC2_TEMP’: ‘44.843C’, ‘TDIODE_TEMP’: ‘38.875C’, ‘SOC0_TEMP’: ‘46.156C’, ‘GPU_TEMP’: ‘43.687C’, ‘TJ_TEMP’: ‘48.812C’, ‘SOC1_TEMP’: ‘45.656C’, ‘VDD_GPU_SOC’: ‘5191mW/5190mW’, ‘VDD_CPU_CV’: ‘1996mW/2032mW’, ‘VIN_SYS_5V0’: ‘4528mW/4539mW’, ‘VDDQ_VDD2_1V8AO’: ‘401mW/411mW’ }

Parameters:

file (Path, str) – The path to the file.

Returns:

The parsed data, each entry in the list is one line from tegrastats

Return type:

list[dict[str, str]]

Raises:

FileNotFoundError – If the file does not exist

jetsontools.set_log_level(level: str) → None

Set the log level for the jetsontools package.

Parameters:

level (str) – The log level to set. One of “DEBUG”, “INFO”, “WARNING”, “ERROR”, “CRITICAL”.

Raises:

ValueError – If the level is not one of the allowed values.