Roofline analysis for Jetson AGX Orin

[virtual.ramblings]

We have been trying to use ncu to get a roofline chart for our deep learning workload. We tried running with --set roofline and --set SpeedOfLight_RooflineChart. When the generated .ncu-rep file is opened, we see this message in the details Tab: “No report sections available. See Profile options”.

Any idea on why this could be?
Also, we noticed that dram__bytes.sum is not listed in --query-metrics. Does this mean it is not available on the Jetson?

[carolyuu]

Hi,
Here are some suggestions for the common issues:

1. Performance

Please run the below command before benchmarking deep learning use case:

$ sudo nvpmodel -m 0
$ sudo jetson_clocks

2. Installation

Installation guide of deep learning frameworks on Jetson:

• TensorFlow: Installing TensorFlow for Jetson Platform - NVIDIA Docs
• PyTorch: Installing PyTorch for Jetson Platform - NVIDIA Docs
  We also have containers that have frameworks preinstalled:
  Data Science, Machine Learning, AI, HPC Containers | NVIDIA NGC

3. Tutorial

Startup deep learning tutorial:

• Jetson-inference: Hello AI World guide to deploying deep-learning inference networks and deep vision primitives with TensorRT and NVIDIA Jetson
• TensorRT sample: Jetson/L4T/TRT Customized Example - eLinux.org

4. Report issue

If these suggestions don’t help and you want to report an issue to us, please attach the model, command/step, and the customized app (if any) with us to reproduce locally.

Thanks!

[AastaLLL]

Hi,

Have you profiled it using root or sudo?
Please note that collecting GPU traces requires root authority.

Also, we noticed that dram__bytes.sum is not listed in --query-metrics. Does this mean it is not available on the Jetson?

Yes, if you don’t see this in the metrics, it might not be supported.
Since Jetson is a shared memory system, some memory profiling functionality will be different.

Thanks.