Home
last modified time | relevance | path

Searched +full:profile +full:- +full:traces (Results 1 – 25 of 228) sorted by relevance

12345678910

/external/bcc/man/man8/
Dprofile.81 .TH profile 8 "2020-03-18" "USER COMMANDS"
3 profile \- Profile CPU usage by sampling stack traces. Uses Linux eBPF/bcc.
5 .B profile [\-adfh] [\-p PID | \-L TID] [\-U | \-K] [\-F FREQUENCY | \-c COUNT]
6 .B [\-\-stack\-storage\-size COUNT] [\-C CPU] [\-\-cgroupmap CGROUPMAP] [\-\-mntnsmap MAPPATH] [dur…
8 This is a CPU profiler. It works by taking samples of stack traces at timed
10 executing, and by how much, including both user-level and kernel code.
14 not 50, is to avoid lock-step sampling.
16 This is also an efficient profiler, as stack traces are frequency counted in
19 at the end of the profile, greatly reducing the kernel<->user transfer.
24 for an older version that may work on Linux 4.6 - 4.8.
[all …]
/external/bcc/tools/
Dprofile.py2 # @lint-avoid-python-3-compatibility-imports
4 # profile Profile CPU usage by sampling stack traces at a timed interval.
7 # This is an efficient profiler, as stack traces are frequency counted in
10 # at the end of the profile, greatly reducing the kernel<->user transfer.
15 # a version of this tool that may work on Linux 4.6 - 4.8.
25 # 15-Jul-2016 Brendan Gregg Created this.
26 # 20-Oct-2016 " " Switched to use the new 4.9 support.
27 # 26-Jan-2019 " " Changed to exclude CPU idle by default.
28 # 11-Apr-2023 Rocky Xing Added option to increase hash storage size.
69 # -EFAULT in get_stackid normally means the stack-trace is not available,
[all …]
Dprofile_example.txt1 Demonstrations of profile, the Linux eBPF/bcc version.
4 This is a CPU profiler. It works by taking samples of stack traces at timed
9 # ./profile
10 Sampling at 49 Hertz of all threads by user + kernel stack... Hit Ctrl-C to end.
18 - cp (9036)
23 - sign-file (8877)
34 - dd (25036)
41 - func_ab (13549)
46 This default output prints stack traces, followed by a line to describe the
53 It's common for user-level software that hasn't been compiled with frame
[all …]
/external/mesa3d/src/intel/ci/
Dgitlab-ci.yml2 - local: 'src/intel/ci/gitlab-ci-inc.yml'
4 anv-jsl:
6 - .lava-acer-cb317-1h-c3z6-dedede:x86_64
7 - .anv-test
9 DEQP_SUITE: anv-jsl
11 # force fixed kernel, since too many flakes and unexpected results seen with 6.13-rc4
12 EXTERNAL_KERNEL_TAG: "v6.11-rc7-mesa-c984"
15 anv-jsl-full:
17 - anv-jsl
18 - .anv-manual-rules
[all …]
Dgitlab-ci-inc.yml1 .intel-common-rules:
4 - changes: &intel_common_file_list
5 - src/intel/*
6 - src/intel/blorp/**/*
7 - src/intel/common/**/*
8 - src/intel/compiler/**/*
9 - src/intel/dev/**/*
10 - src/intel/ds/**/*
11 - src/intel/genxml/**/*
12 - src/intel/isl/**/*
[all …]
/external/bcc/tools/old/
Dprofile.py2 # @lint-avoid-python-3-compatibility-imports
4 # profile Profile CPU usage by sampling stack traces at a timed interval.
7 # This is an efficient profiler, as stack traces are frequency counted in
10 # at the end of the profile, greatly reducing the kernel<->user transfer.
16 # Kernel stacks are post-process in user-land to skip the interrupt framework
18 # of frames to skip with -s, provided you know what that is. If you get -s
21 # Note: if another perf-based sampling session is active, the output may become
37 # 15-Jul-2016 Brendan Gregg Created this.
73 ./profile # profile stack traces at 49 Hertz until Ctrl-C
74 ./profile -F 99 # profile stack traces at 99 Hertz
[all …]
Dprofile_example.txt1 Demonstrations of profile, the Linux eBPF/bcc version.
4 This is a CPU profiler. It works by taking samples of stack traces at timed
9 # ./profile
10 Sampling at 49 Hertz of all threads by user + kernel stack... Hit Ctrl-C to end.
18 - cp (9036)
23 - sign-file (8877)
34 - dd (25036)
41 - func_ab (13549)
55 - swapper/0 (0)
64 - swapper/1 (0)
[all …]
/external/perfetto/docs/instrumentation/
Dheapprofd-api.md1 # heapprofd Custom Allocator API - Early Access
20 Join our [Google Group](https://groups.google.com/forum/#!forum/perfetto-dev)
27 First, [check out Perfetto](https://perfetto.dev/docs/contributing/build-instructions):
38 perfetto/ $ tools/install-build-deps --android
39 perfetto/ $ tools/setup_all_configs.py --android
40 perfetto/ $ ninja -C out/android_release_incl_heapprofd_arm64 \
57 git rev-parse HEAD > perfetto-version.txt
100 ## Profile your App
104 script to get a profile to generate textpb of the config.
109 [Learn how to install protoc](https://grpc.io/docs/protoc-installation).
[all …]
/external/mesa3d/.gitlab-ci/piglit/
Dpiglit-traces.sh6 . "${SCRIPTS_DIR}/setup-test-env.sh"
8 section_start traces_prepare "traces: preparing test setup"
10 set -ex
15 INSTALL=$(realpath -s "$PWD"/install)
16 S3_ARGS="--token-file ${S3_JWT_FILE}"
21 # Needed because yq and ci-fairy are installed there.
24 if [ "$PIGLIT_REPLAY_SUBCOMMAND" = "profile" ]; then
25 yq -iY 'del(.traces[][] | select(.label[]? == "no-perf"))' \
29 export PIGLIT_REPLAY_EXTRA_ARGS="--keep-image ${PIGLIT_REPLAY_EXTRA_ARGS}"
37 if [ -n "${VK_DRIVER}" ]; then
[all …]
/external/perfetto/docs/quickstart/
Dtraceconv.md3 _This quickstart demonstrates how Perfetto traces can be converted into other trace formats using t…
5 ![](/docs/images/traceconv-summary.png)
9 - A host running Linux or MacOS
10 - A Perfetto protobuf trace file
14 - `text` - protobuf text format: a text based representation of protos
15 - `json` - Chrome JSON format: the format used by chrome://tracing
16 - `systrace`: the ftrace text format used by Android systrace
17 - `profile` : pprof-like format. Either for traces with with
18 [native heap profiler](/docs/data-sources/native-heap-profiler.md) dumps or
19 [callstack sampling](/docs/quickstart/callstack-sampling.md) (note however
[all …]
/external/pytorch/docs/source/
Dtorch.compiler_profiling_torch_compile.rst5 -------------------------------
7 …pful for understanding the performance of your program at a kernel-level granularity - for example…
9 …erstand kernel-level performance, other toosl exist. NVIDIA's ncu tool can be used, or :ref:`induc…
13 Basics of using torch.profiler and viewing traces
14 -------------------------------------------------
18 * Include a warm-up run to wait for compilation to complete (this will warm up systems like the CUD…
19 * Use :code:`torch.profiler.profile()` context for profiling the section we are interested in
22 .. code-block:: python
39 with torch.profiler.profile() as prof:
46 **Viewing chrome traces**: In the Chrome browser, open chrome://tracing and load the json file. Use…
[all …]
/external/tensorflow/tensorflow/core/profiler/
Dprofiler_options.proto20 // Device type to profile/trace: (version >= 1)
29 // We don't collect the dataset ops by default for better trace-viewer
34 // - Level 0 is used to disable host traces.
35 // - Level 1 enables tracing of only user instrumented (or default) TraceMe.
36 // - Level 2 enables tracing of all level 1 TraceMe(s) and instrumented high
39 // - Level 3 enables tracing of all level 2 TraceMe(s) and more verbose
40 // (low-level) program execution details (cheap TF ops, etc).
44 // - Level 0 is used to disable device traces.
45 // - Level 1 is used to enable device traces.
46 // - More levels might be defined for specific device for controlling the
[all …]
/external/perfetto/docs/
Dtracing-101.md2 *This page provides a birds-eye view of performance analysis.
17 complicated, having a lot of components and a web of cross-interactions.
21 **Tracing** and **profiling** are two such widely-used techniques for
22 performance analysis. **Perfetto** is an open-source suite of tools, combining
30 Traces contain enough detail to fully reconstruct the timeline of events.
31 They often include low-level kernel events like scheduler context switches,
41 The level of detail in traces makes it impractical to read traces directly
68 Usually metrics map to high-level concepts. Examples of metrics include: CPU
74 compute metrics on resulting traces? In some settings, this may indeed be the
75 right approach. In local and lab situations using **trace-based metrics**,
[all …]
/external/jazzer-api/docs/
Dadvanced.md3 * [Passing JVM arguments](#passing-jvm-arguments)
4 * [Coverage instrumentation](#coverage-instrumentation)
5 * [Trace instrumentation](#trace-instrumentation)
6 * [Value profile](#value-profile)
7 * [Custom hooks](#custom-hooks)
8 * [Suppressing stack traces](#suppressing-stack-traces)
9 * [Export coverage information](#export-coverage-information)
10 * [Native libraries](#native-libraries)
11 * [Fuzzing mutators](#fuzzing-mutators)
13 <!-- Created by https://github.com/ekalinin/github-markdown-toc -->
[all …]
/external/mesa3d/src/gallium/drivers/virgl/ci/
Dgitlab-ci.yml2 - local: 'src/gallium/drivers/virgl/ci/gitlab-ci-inc.yml'
4 virpipe-on-gl:
6 - .deqp-test
7 - .virpipe-test
9 DEQP_SUITE: virpipe-gl
10 GPU_VERSION: virpipe-gl
14 virgl-on-gl:
16 DEQP_SUITE: virgl-gl
17 GPU_VERSION: virgl-gl
21 - kvm
[all …]
/external/pytorch/test/dynamo/
Dtest_profiler.py16 # dynamo_timed functions should appear in profile traces.
26 with torch.profiler.profile(with_stack=False) as prof:
34 # dynamo_timed functions should appear in profile traces.
46 with torch.profiler.profile(with_stack=False) as prof:
68 with torch.profiler.profile(record_shapes=True):
80 with torch.profiler.profile(record_shapes=True):
92 with torch.profiler.profile(record_shapes=True):
118 (torch.autograd.profiler.profile, lambda prof: prof.function_events),
119 (torch.profiler.profiler.profile, lambda prof: prof.events()),
155 prof = torch.profiler.profile(
[all …]
/external/tensorflow/tensorflow/core/profiler/g3doc/
Dcommand_line.md3 * [Command Line Inputs](#command-line-inputs)
4 * [Start `tfprof`](#start-tfprof)
6 * [Profile Python Time](#profile-python-time)
7 * [Profile Graph Time](#profile-graph-time)
8 * [Profile Checkpoint Value](#profile-checkpoint-value)
9 * [Profile Model Parameter](#profile-model-parameter)
10 * [Profile Device Placement](#profile-device-placement)
11 * [Define Customized Operation Type](#define-customized-operation-type)
12 * [Non-interactive Mode](#non-interactive-mode)
19 <b>--profile_path:</b> A ProfileProto binary proto file.
[all …]
/external/pytorch/torch/mps/
Dprofiler.py1 # mypy: allow-untyped-defs
7 __all__ = ["start", "stop", "profile"]
10 def start(mode: str = "interval", wait_until_completed: bool = False) -> None:
19 The interval mode traces the duration of execution of the operations,
40 def profile(mode: str = "interval", wait_until_completed: bool = False): function
46 The interval mode traces the duration of execution of the operations,
/external/perfetto/docs/case-studies/
Dmemory.md6 * [ADB](https://developer.android.com/studio/command-line/adb) installed and
13 /docs/data-sources/native-heap-profiler.md#heapprofd-targets) for more
19 `dumpsys meminfo` which gives a high-level overview of how much of the various
31 ------ ------ ------ ------ ------ ------ ------ ------
54 the [mmap() system call](https://man7.org/linux/man-pages/man2/mmap.2.html).
59 VMAs can be of two types: file-backed and anonymous.
61 **File-backed VMAs** are a view of a file in memory. They are obtained passing a
65 File-backed VMAs are used, for instance, by the dynamic linker (`ld`) when
69 **Anonymous VMAs** are memory-only areas not backed by any file. This is the way
87 From a memory-consumption viewpoint, individual pages within a VMA can have the
[all …]
Dandroid-boot-tracing.md1 # Recording traces on Android boot
4 recording traces on boot. This can be useful to profile the boot process.
50 # 10s trace, but can be stopped prematurely via `adb shell pkill -u perfetto`.
53 * Put the file on the device at `/data/misc/perfetto-configs/boottrace.pbtxt`:
55 adb push <yourfile> /data/misc/perfetto-configs/boottrace.pbtxt
65 `/data/misc/perfetto-traces/boottrace.perfetto-trace`. The file will be
68 adb pull /data/misc/perfetto-traces/boottrace.perfetto-trace
73 * `boottrace.perfetto-trace` can now be opened in
/external/perfetto/docs/reference/
Dheap_profile-cli.md5 heap_profile - record heap profile on Android device
10 See [Recording traces](/docs/data-sources/native-heap-profiler.md) for more
11 details about the data-source.
14 usage: heap_profile [-h] [-i INTERVAL] [-d DURATION] [--no-start] [-p PIDS]
15 [-n NAMES] [-c CONTINUOUS_DUMP] [--disable-selinux]
16 [--no-versions] [--no-running] [--no-startup]
17 [--shmem-size SHMEM_SIZE] [--block-client]
18 [--block-client-timeout BLOCK_CLIENT_TIMEOUT]
19 [--no-block-client] [--idle-allocations] [--dump-at-max]
20 [--disable-fork-teardown] [--simpleperf]
[all …]
/external/perfetto/python/tools/
Dheap_profile.py8 # http://www.apache.org/licenses/LICENSE-2.0
40 UUID = str(uuid.uuid4())[-6:]
94 "https://perfetto.dev/docs/data-sources/native-heap-profiler#troubleshooting.",
99 return ('https://perfetto.dev/docs/data-sources/native-heap-profiler'
100 '#known-issues-android{}'.format(number))
114 'ro.build.version.release_or_codename']).decode('utf-8').strip()
127 'ro.system.build.version.sdk']).decode('utf-8').strip())
132 'ro.build.version.codename']).decode('utf-8').strip()
136 ORDER = ['-n', '-p', '-i', '-o']
144 return result, action.option_strings[0].strip('-')
[all …]
Dcpu_profile.py8 # http://www.apache.org/licenses/LICENSE-2.0
15 """Runs tracing with CPU profiling enabled, and symbolizes traces if requested.
18 https://perfetto.dev/docs/quickstart/callstack-sampling
38 UUID = str(uuid.uuid4())[-6:]
57 "https://perfetto.dev/docs/contributing/getting-started#bugs.".format(
64 return subprocess.check_output(command).decode('utf-8')
77 """Parses, validates, and returns command-line arguments for this script."""
79 traces if requested.
82 https://perfetto.dev/docs/quickstart/callstack-sampling
86 "-f",
[all …]
/external/pytorch/torch/profiler/
D__init__.py1 # mypy: allow-untyped-defs
5 examine their input shapes and stack traces, study device kernel activity and visualize the executi…
21 profile,
30 "profile",
/external/lottie/baselineprofile/src/main/java/com/airbnb/lottie/baselineprofile/
DStartupBenchmarks.kt16 * Run this benchmark to verify how effective a Baseline Profile is.
20 * Run this benchmark to see startup measurements and captured system traces for verifying
24 …* ./gradlew :baselineprofile:connectedAndroidTest -Pandroid.testInstrumentationRunnerArguments.and…
30 … see the [Macrobenchmark documentation](https://d.android.com/macrobenchmark#create-macrobenchmark)
31 …ntation](https://d.android.com/topic/performance/benchmarking/macrobenchmark-instrumentation-args).
33 … impact of this test is only tangentially related to the impact of the baseline profile for Lottie.
34 …* The benefit of the baseline profile for Lottie is less about startup time and more about time sp…

12345678910