Get Started

Contents

Get Started

To run the Diagnostics Utility for Intel® oneAPI Toolkits:
  1. Open a terminal and navigate to the directory where oneAPI products are installed.
    If you installed to the default directory, the utility will be located at
    /opt/intel/oneapi/
    .
  2. Change to the diagnostics directory:
cd diagnostics/latest/
  1. Run the Diagnostics Utility for Intel® oneAPI Toolkits using this syntax:
python3 diagnostics.py --filter <ARGUMENT_NAME>
The
filter
 modifier specifies how many checks to run at one time. With a filter, you can run a group of checks or a single check. In this example, you will run all available checks, using the
all
 argument:
python3 diagnostics.py --filter all
The output will display in two places:
  • the active console
  • a JSON file
Checks listed at the top of the list have higher priority and may have dependent checks further down the list. When diagnosing problems, start by troubleshooting the checks at the top of the list.
The default directory of the output files is $HOME/intel/diagnostics/logs for Linux.
To customize the default directory, see Customizing the Diagnostics Utility.
To increase the amount of detail, add the -v argument:
python3 diagnostics.py --filter all -v
The Diagnostics Utility for Intel® oneAPI Toolkits can be customized to output only the data you need. Use these methods to customize the command for your needs:

List of Checks by Check Name

Note: the checks below are all supported on Linux.
The table below shows the list of
checks
 and which
groups
 can also be used to run that check.
Check Name
: the name of the diagnostic check that can be run in the command line.
Group
: Checks are grouped together so that you can run multiple checks using one command. For example, you can run all checks associated with GPUs by using the GPU argument.
Rights
: the permissions needed to run this check.
Description
: a short description of the check.
Check Name
Groups where check is included
Description
hangcheck_check


gpu
profiling
runtime
sysinfo
target

This check verifies that the GPU hangcheck option is disabled to allow long-running jobs.
user_group_check


default
profiling
gpu
runtime
target

This check verifies that the current user is in the same group as the GPU(s).
driver_compatibility_check


compile
default
host
runtime
sysinfo
target

This check verifies compatibility of oneAPI products versions and GPU drivers versions.
oneapi_gpu_check


gpu
sysinfo

This check runs GPU workloads and verifies readiness to run applications on GPU(s).
gpu_metrics_check


gpu
runtime
target

This check verifies that GPU metrics are good.
gpu_backend_check


compile
default
gpu
host
runtime
sysinfo
target

This check shows information from OpenCL™ and Intel® oneAPI Level Zero drivers.
oneapi_app_check


compile
default
host
runtime
sysinfo
target

This check shows version information of installed oneAPI products.
intel_gpu_detector_check


profiling
default
gpu
runtime
sysinfo
target

This check shows which Intel GPU(s) is on the system based on lspci information and internal table.
oneapi_env_check


compile
default
host
runtime
sysinfo
target

This check shows the version information of the oneAPI products installed in the environment.
gcc_version_check


compile
default
host
sysinfo

This check shows information about GCC compiler version.
base_system_check


compile
host
runtime
sysinfo
target

This check shows information about hostname, CPU, BIOS and operating system.
kernel_options_check


runtime
sysinfo
target

This check shows kernel options.
user_resources_limits_check


compile
host
runtime
sysinfo
target

This check shows limits of each resource.
sys_check


not included in any groups

Some oneAPI components may have checks specific to that component. These checks will be available after setting environment variables using the setvars script. Run the script and then run the Diagnostics Utility with the
sys_check
 name to see checks that are specific to installed components.
To learn more about the output, see Resolving ERROR, FAIL, and WARNING Status.

List of Checks by Group Name

Check Name
: the name of the diagnostic check that can be run in the command line.
Group
: Checks are grouped together so that you can run multiple checks using one command. For example, you can run all checks associated with GPUs by using the
gpu
 argument.
For a description of what each
check
 does and what permissions are needed to run the
check
, see List of Checks by Check Name.
Group of Checks
Checks included
compile


driver_compatibility_check
gpu_backend_check
oneapi_app_check
oneapi_env_check
gcc_version_check
base_system_check
user_resources_limits_check

default


user_group_check
driver_compatibility_check
gpu_backend_check
oneapi_app_check
intel_gpu_detector_check
oneapi_env_check
gcc_version_check

host


driver_compatibility_check
gpu_backend_check
oneapi_app_check
oneapi_env_check
gcc_version_check
base_system_check
user_resources_limits_check

runtime


hangcheck_check
user_group_check
driver_compatibility_check
gpu_metrics_check
gpu_backend_check
oneapi_app_check
intel_gpu_detector_check
oneapi_env_check
base_system_check
kernel_options_check
user_resources_limits_check

sysinfo


hangcheck_check
driver_compatibility_check
oneapi_gpu_check
gpu_backend_check
oneapi_app_check
intel_gpu_detector_check
oneapi_env_check
gcc_version_check
base_system_check
kernel_options_check
user_resources_limits_check

target


hangcheck_check
user_group_check
driver_compatibility_check
gpu_metrics_check
gpu_backend_check
oneapi_app_check
intel_gpu_detector_check
oneapi_env_check
base_system_check
kernel_options_check
user_resources_limits_check

gpu


hangcheck_check
user_group_check
oneapi_gpu_check
gpu_metrics_check
gpu_backend_check
intel_gpu_detector_check

profiling


hangcheck_check
user_group_check
intel_gpu_detector_check

Product and Performance Information

1

Performance varies by use, configuration and other factors. Learn more at www.Intel.com/PerformanceIndex.