### ImportNvtxt Info Command Example

Source: https://docs.nvidia.com/nsight-systems/UserGuide/index.html

An example of running the 'info' command and its output, showing analysis start and end times in nanoseconds.

```bash
ImportNvtxt info Report.nsys-rep
Analysis start (ns) 83501026500000
Analysis end (ns)   83506375000000
```

--------------------------------

### Basic Container Launch for Preflight Check

Source: https://docs.nvidia.com/nsight-systems/UserGuide/index.html

Perform a basic container launch to verify the Pyxis/Enroot setup before adding Nsight Systems specific configurations. This command ensures the container can be started successfully.

```bash
srun --container-image=nvcr.io#nvidia/pytorch:<tag> /bin/true

```

--------------------------------

### Install Nsight Systems CLI on Ubuntu (Minimal Container)

Source: https://docs.nvidia.com/nsight-systems/InstallationGuide/index.html

Installs the Nsight Systems command-line interface on Ubuntu, suitable for minimal container setups. Requires root privileges.

```bash
apt update
apt install -y --no-install-recommends gnupg
echo "deb http://developer.download.nvidia.com/devtools/repos/ubuntu$(source /etc/lsb-release; echo "$DISTRIB_RELEASE" | tr -d .)/$(dpkg --print-architecture) /" | tee /etc/apt/sources.list.d/nvidia-devtools.list
apt-key adv --fetch-keys http://developer.download.nvidia.com/compute/cuda/repos/ubuntu1804/x86_64/7fa2af80.pub
apt update
apt install nsight-systems-cli
```

--------------------------------

### Example: Flash 6.0.9.1 QNX

Source: https://docs.nvidia.com/nsight-systems/UserGuide/index.html

Example command to run the Docker container for flashing version 6.0.9.1 of DRIVE OS on QNX.

```bash
sudo docker run --rm --privileged --net host \
    -v /dev/bus/usb:/dev/bus/usb \
    -v /tmp:/drive_flashing \
    -it nvcr.io/{MY_NGC_ORG}/driveos-pdk/drive-agx-orin-qnx-aarch64-pdk-build-x86:6.0.9.1-latest

```

--------------------------------

### Example: Flash 6.0.8.0 QNX

Source: https://docs.nvidia.com/nsight-systems/UserGuide/index.html

Example command to run the Docker container for flashing version 6.0.8.0 of DRIVE OS on QNX.

```bash
sudo docker run --rm --privileged --net host \
    -v /dev/bus/usb:/dev/bus/usb \
    -v /tmp:/drive_flashing \
    -it nvcr.io/{MY_NGC_ORG}/driveos-pdk/drive-agx-orin-qnx-aarch64-pdk-build-x86:6.0.8.0-0003

```

--------------------------------

### Example: Flash 6.0.8.0 Linux

Source: https://docs.nvidia.com/nsight-systems/UserGuide/index.html

Example command to run the Docker container for flashing version 6.0.8.0 of DRIVE OS on Linux.

```bash
sudo docker run --rm --privileged --net host \
    -v /dev/bus/usb:/dev/bus/usb \
    -v /tmp:/drive_flashing \
    -it nvcr.io/{MY_NGC_ORG}/driveos-pdk/drive-agx-orin-linux-aarch64-pdk-build-x86:6.0.8.0-0003

```

--------------------------------

### Install Nsight Systems on Compute Nodes

Source: https://docs.nvidia.com/nsight-systems/UserGuide/index.html

Example path for installing Nsight Systems on Slurm compute nodes. Ensure the path is identical and readable across all nodes and within the container's mount namespace.

```bash
/opt/nvidia/nsight-systems/2026.2.1
```

--------------------------------

### Install Nsight Systems CLI on CentOS/RHEL (Minimal Container)

Source: https://docs.nvidia.com/nsight-systems/InstallationGuide/index.html

Installs the Nsight Systems command-line interface on CentOS/RHEL, suitable for minimal container setups. Requires root privileges.

```bash
rpm --import https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1804/x86_64/7fa2af80.pub
sed -i 's/mirrorlist/#mirrorlist/g' /etc/yum.repos.d/CentOS-*
sed -i 's|#baseurl=http://mirror.centos.org|baseurl=http://vault.centos.org|g' /etc/yum.repos.d/CentOS-*
dnf install -y 'dnf-command(config-manager)'
dnf config-manager --add-repo "https://developer.download.nvidia.com/devtools/repos/rhel$(source /etc/os-release; echo ${VERSION_ID%%.*})/$(rpm --eval '%{_arch}' | sed s/aarch/arm/)/"
dnf install -y nsight-systems-cli
```

--------------------------------

### Launch Nsight Systems GUI

Source: https://docs.nvidia.com/nsight-systems/InstallationGuide/index.html

Execute this command to start the Nsight Systems graphical user interface directly. The executable is located in the Host sub-directory of your Nsight Systems installation.

```bash
nsys-ui
```

--------------------------------

### Nsight Systems CLI GPU Video Devices Example Output

Source: https://docs.nvidia.com/nsight-systems/UserGuide/index.html

Example output from the '--gpu-video-devices help' command, showing supported and unsupported GPUs for video accelerator tracing.

```text
Possible --gpu-video-devices values are:
    0: NVIDIA GeForce RTX 3070 PCI[0000:65:00.0]
    all: Select all supported GPUs
    none: Disable GPU video accelerator tracing [Default]

Some GPUs don't support video accelerator tracing:
    Quadro P620 PCI[0000:04:00.0] (reason = Arch Pascal < Turing)
```

--------------------------------

### Install Nsight Systems GUI on Ubuntu (Desktop)

Source: https://docs.nvidia.com/nsight-systems/InstallationGuide/index.html

Installs the full Nsight Systems package, including the GUI, on Ubuntu desktop systems. Requires root privileges.

```bash
sudo apt-key adv --fetch-keys https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1804/x86_64/7fa2af80.pub
sudo add-apt-repository "deb https://developer.download.nvidia.com/devtools/repos/ubuntu$(source /etc/lsb-release; echo "$DISTRIB_RELEASE" | tr -d .)/$(dpkg --print-architecture)/ /"
sudo apt install nsight-systems
```

--------------------------------

### One-Step Dependency Installation for Nsight Systems

Source: https://docs.nvidia.com/nsight-systems/InstallationGuide/index.html

Installs all Nsight Systems CLI and GUI dependencies using pip. Ensure your virtual environment is activated before running.

```bash
python3 -m pip install -r nsys_recipe/requirements/dask.txt -r nsys_recipe/requirements/common.txt -r nsys_recipe/requirements/jupyter.txt
```

--------------------------------

### Automated Nsight Systems Installation (Linux)

Source: https://docs.nvidia.com/nsight-systems/InstallationGuide/index.html

Enables automated installation of Nsight Systems on Linux hosts by accepting the EULA. Use `--quiet` for silent installation.

```bash
sudo apt install nsight-systems --accept
```

--------------------------------

### Install Nsight Systems Dependencies with Root Privileges

Source: https://docs.nvidia.com/nsight-systems/UserGuide/index.html

Run this script to install all required libraries in system directories for Nsight Systems GUI on Ubuntu and CentOS when root privileges are available.

```bash
[installation_path]/host-linux-[arch]/Scripts/DependenciesInstaller/install-dependencies.sh

```

--------------------------------

### Example Nsight Systems Plugin Manifest File

Source: https://docs.nvidia.com/nsight-systems/UserGuide/index.html

This is an example of a plugin manifest file in YAML format. It specifies the plugin name, executable path, library path, target environment, and a description.

```yaml
PluginName: unique_name
ExecutablePath: bin/plugin
LibraryPath: libPlugin.so
TargetEnvironment: {KEY: VALUE}
Description: This is an example plugin manifest.

```

--------------------------------

### Install Nsight Systems GUI on CentOS/RHEL (Desktop)

Source: https://docs.nvidia.com/nsight-systems/InstallationGuide/index.html

Installs the full Nsight Systems package, including the GUI, on CentOS/RHEL desktop systems. Requires root privileges.

```bash
sudo rpm --import https://developer.download.nvidia.com/compute/cuda/repos/ubuntu1804/x86_64/7fa2af80.pub
sudo dnf install -y 'dnf-command(config-manager)'
sudo dnf config-manager --add-repo "https://developer.download.nvidia.com/devtools/repos/rhel$(source /etc/os-release; echo ${VERSION_ID%%.*})/$(rpm --eval '%{_arch}' | sed s/aarch/arm/)/"
sudo dnf install nsight-systems
```

--------------------------------

### Install Netcat on Ubuntu

Source: https://docs.nvidia.com/nsight-systems/UserGuide/index.html

Installs the Netcat utility on Ubuntu-based systems. Netcat is a required dependency for Nsight Systems on the target device.

```bash
sudo apt-get install netcat-openbsd
```

--------------------------------

### Install xcb-util-cursor on RHEL/CentOS/Fedora

Source: https://docs.nvidia.com/nsight-systems/UserGuide/index.html

This command installs the necessary xcb-cursor package on RHEL-based systems to resolve issues with the Qt xcb platform plugin.

```bash
sudo dnf install -y xcb-util-cursor

```

--------------------------------

### NVTX Capture with Specific Message and Any Domain

Source: https://docs.nvidia.com/nsight-systems/UserGuide/index.html

Launches an application, profiling starts when the first range with the message 'profiler' is opened in any domain.

```bash
nsys launch -w true -p profiler@* ./app
```

--------------------------------

### Install Nsight Systems Dependencies Without Root Privileges

Source: https://docs.nvidia.com/nsight-systems/UserGuide/index.html

Execute this script to install Nsight Systems dependencies into a specified directory when root privileges are not available. Ensure the target directory is user-writable.

```bash
[installation_path]/host-linux-[arch]/Scripts/DependenciesInstaller/install-dependencies-without-root.sh [dependencies_path]

```

--------------------------------

### NVTX Capture with Specific Message and Domain

Source: https://docs.nvidia.com/nsight-systems/UserGuide/index.html

Launches an application, profiling starts when the first range with the message 'profiler' is opened in the 'service' domain.

```bash
nsys launch -w true -p profiler@service ./app
```

--------------------------------

### Launch Chrome for WebGL Profiling

Source: https://docs.nvidia.com/nsight-systems/UserGuide/index.html

Use this command structure to profile WebGL content in Chrome with Nsight Systems. Ensure Chrome is installed in the default location.

```bash
"C:\Program Files (x86)\Google\Chrome\Application\chrome.exe" --inprocess-gpu --no-sandbox --disable-gpu-watchdog --use-angle=gl https://webglsamples.org/aquarium/aquarium.html
```

--------------------------------

### Format and Mount Power-Safe File System

Source: https://docs.nvidia.com/nsight-systems/UserGuide/index.html

Commands to format a specified partition with a power-safe file system and then mount it. This example targets the `vblk_ufs40` partition.

```bash
mkqnx6fs /dev/vblk_ufs40 -q
mount -o rw /dev/vblk_ufs40 /

```

--------------------------------

### QNX Kernel Symbols Configuration

Source: https://docs.nvidia.com/nsight-systems/UserGuide/index.html

Example JSON configuration for QNX kernel symbols, mapping guest VMs to their respective symbol files.

```json
{
    "guest_cfg": [
        {
            "guest_id": 0,
            "guest_name": "Guest VM 0",
            "symbols": "/root/symbols/procnto-smp-instr-safety.guest_vm.bin.sym"
        },
        {
            "guest_id": 1,
            "guest_name": "Update service",
            "symbols": "/root/symbols/procnto-smp-instr-safety.update_vm.bin.sym"
        },
        {
            "guest_id": 2,
            "guest_name": "Resource Manager Server"
        },
        {
            "guest_id": 3,
            "guest_name": "Storage Server"
        },
        {
            "guest_id": 4,
            "guest_name": "Ethernet Server"
        },
        {
            "guest_id": 5,
            "guest_name": "Debug Server"
        }
    ],
    "symbol_files": {
        "Sidekick": "/root/symbols/sidekick.unstripped"
    }
}
```

--------------------------------

### Flash Drive Orin p3663

Source: https://docs.nvidia.com/nsight-systems/UserGuide/index.html

Example of running the flash script for a Drive Orin p3663 board.

```bash
./flash.py /dev/ttyUSB1 p3663

```

--------------------------------

### ImportNvtxt Help Command

Source: https://docs.nvidia.com/nsight-systems/UserGuide/index.html

Prints the help message for the ImportNvtxt utility. Use this to get an overview of available commands and options.

```bash
-h [ --help ]
```

--------------------------------

### Check QNX Filesystem Writable

Source: https://docs.nvidia.com/nsight-systems/UserGuide/index.html

Use this command to verify if the target QNX filesystem is writable before connecting Nsight Systems. This is necessary for installing target-side binaries.

```bash
echo XX > /xx && ls -l /xx
```

--------------------------------

### Enable NVML Metrics

Source: https://docs.nvidia.com/nsight-systems/UserGuide/index.html

Add this option to nsys profile or start commands to enable NVML power and temperature sampling. Arguments can be specified as needed.

```bash
--enable nvml_metrics[,arg1[=value1],arg2[=value2], ...]
```

--------------------------------

### NVTX Capture with Specific Message in Default Domain

Source: https://docs.nvidia.com/nsight-systems/UserGuide/index.html

Launches an application, profiling starts when the first range with the message 'profiler' is opened in the default domain.

```bash
nsys launch -w true -p profiler ./app
```

--------------------------------

### Print Tool Version Information

Source: https://docs.nvidia.com/nsight-systems/UserGuide/index.html

Use this command to display the installed version of the Nsight Systems CLI tool. This is a simple query to verify the tool's version.

```bash
nsys -v
```

--------------------------------

### Launch Nsight Systems GUI with Dependencies Environment

Source: https://docs.nvidia.com/nsight-systems/UserGuide/index.html

Source this script to set up the environment for Nsight Systems GUI after installing dependencies without root privileges. Then, launch the UI.

```bash
source [installation_path]/host-linux-[arch]/Scripts/DependenciesInstaller/setup-dependencies-environment.sh [dependencies_path] && [installation_path]/host-linux-[arch]/nsys-ui

```

--------------------------------

### Filter Trace Data with grep

Source: https://docs.nvidia.com/nsight-systems/UserGuide/index.html

This example demonstrates how to export trace data to a SQLite file and then process it using a command-line utility like grep to filter specific information. The output name starting with '@' indicates a command to be executed.

```bash
nsys export --stats=cuda_api_sum --output=@grep -E (-|Name|cudaFree) test.sqlite
```

--------------------------------

### Install Pip and Venv on Ubuntu

Source: https://docs.nvidia.com/nsight-systems/InstallationGuide/index.html

Installs pip and venv for Python 3 on Ubuntu systems. Run 'sudo apt-get update' first on a fresh install.

```bash
sudo apt-get update
sudo apt-get install python3-pip
sudo apt-get install python3-venv
```

--------------------------------

### Create XHV Directory (QNX)

Source: https://docs.nvidia.com/nsight-systems/UserGuide/index.html

Creates the necessary directory structure and copies configuration files for XHV sampling on QNX systems.

```bash
cd /drive_flashing
mkdir -p xhv/hypervisor/configs/t234ref-release/pct/qnx xhv/schemas
cp -rv /drive/drive-foundation/virtualization/hypervisor/t23x/configs/t234ref-release/pct/p3710-10-a03/qnx/pct.json ./xhv/hypervisor/configs/t234ref-release/pct/qnx/
cp -rv /drive/drive-foundation/schemas/event ./xhv/schemas/
```

--------------------------------

### Launch Application with Nsight Systems Launcher

Source: https://docs.nvidia.com/nsight-systems/UserGuide/index.html

Use the Nsight Systems launcher binary to start your application. The application will daemonize and wait for a SIGUSR1 signal to resume execution after Nsight Systems attaches.

```bash
$ /opt/nvidia/nsight_systems/launcher ./my-binary --arguments

```

--------------------------------

### Validate GPUDirect Storage Installation

Source: https://docs.nvidia.com/nsight-systems/UserGuide/index.html

Run the `gdscheck.py` tool to verify that GPUDirect Storage is installed correctly and that the intended filesystem type is supported. This script is typically found in the CUDA installation directory.

```bash
/usr/local/cuda/gds/tools/gdscheck.py -p
```

--------------------------------

### Interactive CLI: Launch App, Start Collection Manually

Source: https://docs.nvidia.com/nsight-systems/UserGuide/index.html

Launch an application and then manually start the data collection. This method allows for control over when profiling begins after the application has already started running.

```bash
nsys launch -w true <application> [application-arguments]
```

```bash
nsys start
```

--------------------------------

### Run nvtx_cpu_topdown Recipe

Source: https://docs.nvidia.com/nsight-systems/AnalysisGuide/index.html

This sequence of commands sets up a directory for reports, collects CPU profiling data using a provided script, and then runs the nvtx_cpu_topdown recipe on the collected data. Ensure the path to collect_cpu_topdown.sh is correct for your system.

```bash
mkdir reports && cd reports
<path to target-linux-sbsa-armv8>/CpuProfiling/collect_cpu_topdown.sh ./myApp
nsys recipe nvtx_cpu_topdown --input .
```

--------------------------------

### Create Recipe Directory and Copy Files

Source: https://docs.nvidia.com/nsight-systems/AnalysisGuide/index.html

Use these Linux commands to set up the directory structure and copy necessary files for a new custom recipe. Ensure you replace placeholders with your specific directory and recipe names.

```bash
> cd <install-dir>/target-linux-x64/python/packages/nsys_recipe
> mkdir new_metric_util_map
> cp gpu_metric_util_map/metadata.json new_metric_util_map/metadata.json
> cp gpu_metric_util_map/heatmap.ipynb new_metric_util_map/heatmap.ipynb
> cp gpu_metric_util_map/gpu_metric_util_map.py new_metric_util_map/new_metric_util_map.py
```

--------------------------------

### Install libxcb-cursor0 on Debian/Ubuntu

Source: https://docs.nvidia.com/nsight-systems/UserGuide/index.html

This command installs the necessary libxcb-cursor0 package on Debian/Ubuntu systems to resolve issues with the Qt xcb platform plugin.

```bash
sudo apt-get install -y libxcb-cursor0

```

--------------------------------

### Verify Nsight Systems Installation

Source: https://docs.nvidia.com/nsight-systems/UserGuide/index.html

Command to verify the Nsight Systems installation on a compute node. This checks if the `nsys` executable is accessible and reports its version.

```bash
$ /opt/nvidia/nsight-systems/2026.2.1/bin/nsys --version
```

--------------------------------

### Profile Local and NVMe-oF Volumes

Source: https://docs.nvidia.com/nsight-systems/UserGuide/index.html

Nsight Systems command line for profiling local storage and NVMe-oF device performance. Use '--storage-metrics' and '--storage-devices=all'.

```bash
./nsys profile --storage-metrics --storage-devices=all <target-application>
```

--------------------------------

### Install Nsight Systems Dependencies on Offline Machine

Source: https://docs.nvidia.com/nsight-systems/InstallationGuide/index.html

Installs Nsight Systems dependencies from a tarball on a machine without internet access. Ensure the tarball has been transferred.

```bash
tar -xvfz recipe-deps.tar.gz
python3 -m pip install recipe-deps/* --no-index
```

--------------------------------

### Display Nsight Systems CLI Help

Source: https://docs.nvidia.com/nsight-systems/InstallationGuide/index.html

Run this command to view a list of available options and commands for the Nsight Systems CLI. For comprehensive documentation, refer to the User Guide.

```bash
nsys --help
```

--------------------------------

### List Available Metric Sets

Source: https://docs.nvidia.com/nsight-systems/UserGuide/index.html

To list the available metric sets for your selected GPUs, use the 'help' option for --gpu-metrics-set.

```bash
$ nsys profile --gpu-metrics-devices=all --gpu-metrics-set=help
```

--------------------------------

### Enable GPU Context Switch Tracing

Source: https://docs.nvidia.com/nsight-systems/UserGuide/index.html

Use the `--gpuctxsw` option when running Nsight Systems from the command line to enable GPU context switch tracing.

```bash
nsys profile --gpuctxsw true
```

--------------------------------

### Bind Mount Host Install into Enroot Container

Source: https://docs.nvidia.com/nsight-systems/UserGuide/index.html

This configuration binds the Nsight Systems installation from the host to the container, making it accessible without modifying the container image. It ensures read-only access and prevents execution from the mount point.

```bash
/opt/nvidia/nsight-systems/2026.2.1 /opt/nvidia/nsight-systems/2026.2.1 none bind,ro,nosuid,nodev 0 0
```

--------------------------------

### Python Equivalent for Finding Innermost NVTX Range

Source: https://docs.nvidia.com/nsight-systems/AnalysisGuide/index.html

A Python function to find the innermost NVTX range that encompasses a given kernel's start and end times. It iterates through provided NVTX ranges and returns the text of the latest starting range that fully contains the kernel.

```python
# Python equivalent:
def find_innermost_nvtx_range(kernel_start, kernel_end, nvtx_ranges):
    # Find NVTX ranges that completely contain the kernel
    containing_ranges = []
    for nvtx in nvtx_ranges:
        if nvtx['start'] <= kernel_start and nvtx['end'] >= kernel_end:
            containing_ranges.append(nvtx)

    # Return the innermost (latest starting) range
    if containing_ranges:
        return max(containing_ranges, key=lambda x: x['start'])['text']
    return None

```

--------------------------------

### Delayed Start Profiling Run

Source: https://docs.nvidia.com/nsight-systems/UserGuide/index.html

Starts profiling after a specified delay (20 seconds) and sets an environment variable for the target application. Collection ends upon application exit. This command traces CUDA, OpenGL, NVTX, and OS runtime libraries, and collects CPU sampling and thread scheduling data.

```bash
nsys profile -e TEST_ONLY=0 -y 20
    <application> [application-arguments]
```

--------------------------------

### Run gpu_vram_usage_trace Recipe

Source: https://docs.nvidia.com/nsight-systems/AnalysisGuide/index.html

Execute the gpu_vram_usage_trace recipe using the Nsight Systems CLI. This recipe analyzes VRAM usage patterns from a collected report.

```bash
nsys recipe gpu_vram_usage_trace --input [report file path]
```

--------------------------------

### Flash Firespray p3710

Source: https://docs.nvidia.com/nsight-systems/UserGuide/index.html

Example of running the flash script for a Firespray p3710 board.

```bash
./flash.py /dev/ttyACM1 p3710

```

--------------------------------

### Interactive CLI: Start Collection, Launch App, Stop Manually

Source: https://docs.nvidia.com/nsight-systems/UserGuide/index.html

Initiate an interactive CLI session to start collecting data from the beginning of an application's execution. The collection is set to stop manually. This sequence is suitable for applications that may run for extended periods, but users should be aware of potential storage issues if collections are not stopped.

```bash
nsys start --stop-on-exit=false
```

```bash
nsys launch --trace=cuda,nvtx --sample=none <application> [application-arguments]
```

```bash
nsys stop
```

--------------------------------

### Docker/Podman Run Command for Nsight Systems Profiling

Source: https://docs.nvidia.com/nsight-systems/UserGuide/index.html

This command demonstrates how to run a Docker or Podman container with Nsight Systems bind-mounted and configured for profiling. It includes necessary GPU access and capability flags.

```bash
docker run --rm --gpus all \
  --cap-add=SYS_ADMIN \
  -v /opt/nvidia/nsight-systems/2026.2.1:/opt/nvidia/nsight-systems/2026.2.1:ro \
  my-workload:latest \
  /opt/nvidia/nsight-systems/2026.2.1/bin/nsys profile \
    -t cuda,nvtx,mpi \
    -o /reports/run_%p \
    python train.py
```

--------------------------------

### Create XHV Directory (Linux)

Source: https://docs.nvidia.com/nsight-systems/UserGuide/index.html

Creates the necessary directory structure and copies configuration files for XHV sampling on Linux systems.

```bash
cd /drive_flashing
mkdir -p xhv/hypervisor/configs/t234ref-release/pct/linux xhv/schemas
cp -rv /drive/drive-foundation/virtualization/hypervisor/t23x/configs/t234ref-release/pct/p3710-10-a03/linux/pct.json ./xhv/hypervisor/configs/t234ref-release/pct/linux/
cp -rv /drive/drive-foundation/schemas/event ./xhv/schemas/
```

--------------------------------

### OpenMP Event Class Values

Source: https://docs.nvidia.com/nsight-systems/AnalysisGuide/index.html

Defines event class values for OpenMP, including start and finish events.

```text
78 - TRACE_PROCESS_EVENT_OPENMP
79 - TRACE_PROCESS_EVENT_OPENMP_START
80 - TRACE_PROCESS_EVENT_OPENMP_FINISH
```

--------------------------------

### Run Nsight Systems Recipe CLI

Source: https://docs.nvidia.com/nsight-systems/AnalysisGuide/index.html

Use the 'recipe' CLI command to execute advanced analysis recipes. Specify arguments and the recipe name, followed by any recipe-specific arguments.

```bash
nsys recipe [args] <recipe-name> [recipe args]

```

--------------------------------

### Start CUDA Profiling

Source: https://docs.nvidia.com/nsight-systems/UserGuide/index.html

Use cudaProfilerStart() to begin data collection in a CUDA application. Include cuda_profiler_api.h to use this function.

```c++
#include <cuda_profiler_api.h>

// ...
cudaProfilerStart();
```