How to Run a Reference Pipeline using Local Input Files

Download & Install the Pipeline from NGC

Refer to the Setup section to install this Pipeline.

NGC ID for the clara pull pipeline command: clara_hqs_pipeline
Pipeline file: hqs-pipeline.yaml

Record the pipeline ID during pipeline installation, as it will be used in a later section to trigger a job with input files.

Trigger Pipeline

A pipeline can be triggered via Clara CLI. Follow these steps to trigger a pipeline:

Open a terminal and go to the directory which contains the reference pipeline.

cd ~/.clara/pipelines/clara_hqs_pipeline

Trigger the pipeline using the pipeline ID in the previous section:

clara create jobs -n clarahqs -p <pipeline ID>

When the above step has successfully completed, you should see the following result in the terminal (the Job and Payload UUID may be different).

JOB ID:f3170b457f5c41879951e6da8f37341e
PAYLOAD ID:12bb398cb54844d5958d51e66e10713a

Note down this job ID and payload ID as they would be used in a later section to access output files.

Then, start the job:

clara start job -j <JOB ID>

Verify the Pipeline Execution in Dashboard

Go to the Clara dashboard UI using a web browser: The URL is <machine-IP>:8000.
- You should see a job that includes the name of the job you specified (clarahqs-xxxxx).
- You can inspect the status of each operator inside that job.
- Viewing the UI status for Pipeline Jobs executed using the Clara orchestrator option is currently not available. See the orchestrator documentation for more details.
Click the clara-hqs stage and then click LOGS. You should get something similar to the logs below:

:::::::::::::::::::::::Clara Hardware Qualification Suite:::::::::::::::::::::::
NV_DRIVER_VERSION                   : 440.33.01
GPU:
  ATTACHED_GPU(s)                  : 1
  gpu0:
     product_name                  : Quadro GV100
     gpu_virtualization            : None
     PCIe_gen (current/max)        : 3/3
     link_width (current/max)      : 16x/16x
     ecc_errors                    : N/A
     gpu_temperature               : 43 C
     max_clocks:
        graphics_clock             : 1912 MHz
        sm_clock                   : 1912 MHz
        mem_clock                  : 850 MHz
        video_clock                : 1717 MHz
CPU:
  cores                            : 12
  model_name                       : Intel(R) Core(TM) i7-7800X CPU @ 3.50GHz
MEMORY:
  total                            : 62G
  available                        : 19G
STORAGE                             : No info available
MEMORY_RW_TEST:
  operations                       : 9034267.77 per second
  bandwidth                        : 8822.53 MiB/sec
CPU_SPEED:
  single_threaded                  : 518.66 CPU-Ops/sec
  multi_threaded                   : 3559.10 CPU-Ops/sec
STORAGE_TEST:
  write_speed                      : 331 MB/s
  read_speed                       : 5.2 GB/s
GPU_TEST:
  bandwidthTest:
     host_to_device                : 12.3 GB/s
     device_to_host                : 13.1 GB/s
  concurrentKernels:
     time_taken                    : 0.012s
  convolutionSeparable:
     throughput                    : 26416.1900 MPixels/sec
     time                          : 0.00036 s
  transpose:
     transpose_simple_copy:
        throughput                 : 479.8362 GB/s
        time                       : 0.01628 ms
     transpose_optimized:
        throughput                 : 484.3772 GB/s
        time                       : 0.01613 ms
  sortingNetworks:
     throughput                    : 454.1256 MElements/s
     time                          : 0.00231 s

Download and Verify the Outputs

Once the job has succeeded, you can download the output payload using the clara download command.

The output payload files are located under the /operators/\<operator name defined in the pipeline definition\>/ folder. Since the triggered pipeline had an AI operator name of clara-hqs, you can use the following command to download output files:

clara download <Job ID>:/operators/clara-hqs/*

Then, you can see the output hardware metrics in JSON format using the following command:

cat hardware_metrics.json

If you are working with the server machine, the payload folder is directly accessible in the default payload folder /clara/payloads/<Payload ID>/.